London Stock Exchange Stock

Good morning, everyone, and welcome to LSEG's H1 Results Presentation. Thank you for joining us. As usual, I'm joined by MAP, our CFO; and Peregrine Riviere, our Head of IR.

I'll give you a few highlights of the first 6 months and then hand over to MAP to talk through the numbers in detail. After that, I'm going to spend some time talking specifically about our progress in D&A and our deep engagement with customers as they adapt to an AI world. And then, of course, we will be happy to take your questions.

It has been a great first half. We achieved organic revenue growth of 8.4% with strength across the board. Subscription growth accelerated to 6.3%. Our EBITDA margins improved very strongly, and we're raising guidance to the top of the 80 to 100 basis point range. The top line and margin improvement delivered 17% earnings per share growth, an exceptional growth of 37% in free cash flow per share. We made record returns to shareholders, around GBP 2.6 billion across dividends and buybacks and are back in the market as of today with our next buyback tranche.

This performance and the presentation we're going to share with you today show just how deeply we are engaged with customers across a wide range of data and multiple products, increasingly involving co-development of agents and delivered via both existing and new infrastructure. This engagement demonstrates our deep institutional partnerships, the trust in our data and our engineering expertise. Demand for financial data and analytics is as strong as ever. In fact, segment spend has more than doubled in the last 17 years, while industry headcount has fallen by 1/4. The value of data has decoupled from the number of people using it.

I'm going to talk upfront about our Markets businesses because my progress update later will focus exclusively on AI and the D&A business. Remember that Markets is 40% of LSEG revenue. All our venues have had an exceptional 6 months. Following a very strong Q1, we've seen solid follow-through in Q2, comping a pretty extraordinary prior period in 2025 as well. These platforms are not just about volatility, almost all of them have strong underlying growth drivers, too. And we have invested in them over the years to expand their reach, access new asset classes and develop new protocols to meet customer needs.

To highlight a couple of really notable performances, SwapClear and Equities, both maintained very strong momentum from Q1 into Q2. Total interest rate Swap notional cleared was up 29% across H1 and equities average daily volume on the LSE was up 34%. Some may think that whether our Markets business does well is just a function of market volumes, but LSEG Markets has been doing great for the last 5 years, showing the strength and consistency of execution and the growth drivers we have aligned the business. Annual growth has averaged almost 10% over this period, and it's also been consistent. There have been strong years and really strong years, but no weak years.

We fully intend to maintain that momentum, and we're investing behind it. In the last 6 months, we've done our first private securities market transactions, including 2 high-profile U.K. unicorns this month. We've launched DISH, our platform for real-time settlement, which bridges on-chain and off-chain. And in post-trade solutions, we launched TradeAgent. We just announced our MOU with HSBC to support the U.K.'s first digital gilt instrument and LSE 24, our 24/5 equity trading platform. We'll do a deep dive on our work on the digitalization of market infrastructure covering all of this in early December. And this all translates into our all-weather model.

On this slide, we've shown our organic revenue growth over the last 6 years compared to the change and volatility in a number of measures, which could be seen as drivers of our business, GDP growth, market volatility, equity or debt issuance, for example. As you can see, particularly on the right-hand axis, these measures can bounce around a lot, but you wouldn't know it to look at our revenue growth.

The message is clear, solid and accelerating subscription growth plus attractive market exposures across multiple asset classes generate strong and consistent top line growth irrespective of the external environment, plus whichever way you look at it, the gradient of growth from left to right is clearly trending up.

And now let me hand over to MAP to take you through our very strong financial performance in more detail.

Thanks, David, and good morning to all of you. As you heard from David, we delivered an exceptionally strong financial performance in the first half. Organic revenue up 8.4%, adjusted EBITDA up 14%, adjusted EPS up 17% and free cash flow per share up 37%.

I will now walk you through the building blocks of that performance, starting with revenue growth. Organic revenue growth accelerated to 8.4% and with a 1.5% headwind from FX, reported revenues grew 6.9%. All 4 divisions made a strong positive contribution to that growth as we see on the next slide. D&A was up 5.1%. FTSE Russell and Risk Intelligence both grew between 9% and 10%. Taken together, the subscription businesses accelerated growth to 6.3%, well on track for our 2026 target of 6.5%. Markets had a very strong half, growing 12%.

I will now talk through each of these divisions in more detail, starting with D&A. Workflows continued its good performance, growing 2.8%. Workspace users are responding very positively to the AI tools introduced in the first half, driving additional engagement with the platform. We continue to expand the power of Workspace, integrating FXall more deeply, working towards more seamless Tradeweb integration and expanding initiatives like Open Directory.

Growth in Data & Feeds is accelerating, up 7.5%, driven by our continued innovation and the demand it is supporting across both our real-time and pricing and returns data. I will come back to this on the next slide. Finally, Analytics grew 6%, with good demand for our yield book and LSEG products and supported by 33% growth in usage of our analytic API. Increasingly, customers engage with our D&A product as a single solution as part of our LSEG Data Access agreements or LDAs. These enterprise-wide agreements now drive 18% of D&A revenues, up from 16% at the end of last year. Our largest customers benefit from access to our solution at scale and in return, they give us many years of visible revenue and growth.

Returning to Data & Feeds, customer appetite for our data continues to grow extremely fast. Roughly half of our revenues here come from our real-time services, where our strengths across the latency spectrum position us as the provider #1 globally by some way. We continue to see rapid growth in the volume of Data on this platform, up 70% year-on-year in June with that big spike driven by global fund flows and some big market transaction and up fourfold in 10 years. Demand for historic pricing data continues to grow strongly, too, with 39% annual growth in Tick History usage over the last 2 years. Appetite for our cloud-based solution here is particularly strong.

Turning to FTSE Russell. We continue to see strong demand for our flagship equity indices and benchmarks and good momentum in new products. Subscription revenues grew 6.2%, and we expect this to accelerate to high single-digit growth in the second half. Asset-based revenue performed well, up 15%, driven by higher asset prices and strong inflows. During the half, we launched 52 new ETFs, up 24% from H1 2025. We also announced the introduction of the Russell 9000 Index series, expanding the Russell framework from U.S. to global equity markets.

Moving to Risk Intelligence that delivered another good performance, up 10%. Demand for World-Check was the primary driver of growth, although Digital Identity & Fraud was also very strong with volumes up more than 20% in H1. Looking now at the KPIs we introduced at the start of the year. As a reminder, these give additional insight into our 3 subscription businesses, D&A, FTSE Russell and Risk Intelligence.

Starting with retention, which rose slightly in the half at almost 93%, that speaks to the value we provide to clients as well as the long-term nondiscretionary nature of most of our services. Gross sales of GBP 482 million continue to be strong, increasing 11% compared to June last year. And lastly, the New Product Vitality Index, which is a very healthy 25%, highlighting the high level of innovation across our businesses and customer receptivity to our new or enhanced products. These are the building blocks of our growth that feeds through to ASV growth of 6.1% as we exited Q2, up from 5.9% we reported at year-end.

Our Markets division performed exceptionally in H1, particularly given the incredibly strong prior year comparator. Tradeweb and our OTC derivative businesses grew double digits. And our FX business also had a strong performance, growing 8%. For simplification, we show equities on this slide with some other market activities, but the Equities business grew 12% in H1, driven by strong secondary market activity. We are also seeing traction building across recent initiatives with an encouraging pipeline for our Private Securities Market.

Looking at the whole P&L now, you can see our combination of top line strength and focus on cost discipline and efficiency is delivering good operating leverage throughout the P&L. As I already mentioned, revenue growth of 8.4% translates into 14% growth in EBITDA, 17% growth in operating profit and 17% growth in EPS, all that on an organic constant currency basis.

Taking a closer look at cost on this slide. The 2.5% fall in cost of sales reflects the change to the SwapClear revenue share agreement at the end of last year. This revenue share was at 30% in H1 2025 and is now at 10%. Excluding this, cost of sales grew 8.6%, in line with revenues.

Operating expenses grew well below our revenue growth at 4.6%. Our cost equation looks at labor cost as a percentage of total income. That continues to improve, falling from 30% to 28.1%. It is supported by our workforce insourcing program through which we are internalizing more of our talent and improving our agility and efficiency. As we continue to execute on that program, 77% of our headcount is now internal.

We double-click on the EBITDA margin expansion on the next slide. After adjusting for FX, the improvement in margin is 260 bps, 140 bps of this relates to the change we made to the SwapClear agreement last year, leaving 120 bps of underlying margin expansion in H1. As you can see, this performance derived mostly from a disciplined management of the group labor cost, helped by the strong market performance in Q1 that flowed to the EBITDA. On the 120 bps, I assess the group operating leverage at circa 80 bps and the [ flow-down ] to the market activity at 40 bps. So all in all, that delivers an H1 underlying margin of 52.4%, a very strong margin progression from the 49.8% in H1 last year.

Let me now walk you through our margin expectation for the rest of the year. As you may have read in the RNS, we are raising our EBITDA margin guidance from 80 to 100 bps improvement to around 100 bps improvement in constant currency. Given the strong margin performance in H1, that implies a year-on-year slight decline of about 50 bps in EBITDA margin in H2. This is due to the mathematical impact of the SwapClear revenue share change, which in 2025 was all booked in Q4. That creates a 70 bps headwind in H2. Aside from that, we expect to make continued strong underlying progress in operating leverage in line with H1. And we budgeted in H2 around GBP 25 million of one-off costs to accelerate the continued transformation of the group.

Turning now to net finance expense. You can see that adjusted net finance expense was GBP 149 million this half, up from GBP 66 million in H1 2025. Last year figure benefited from GBP 35 million of gain from a bond repurchase and the end of hedging instrument. The underlying increase was just under GBP 50 million and is mainly driven by the impact of higher global interest rates. Rates have typically been 300 basis points higher as we have refinanced over the last 12 months. We expect net finance expense to be similar in the second half, so a full year expense of around GBP 300 million.

On the next slide, our tax rate is consistent with the 24% to 25% range we guided to, and that remains the right range for the rest of the year. Through that combination of top line strength, cost discipline and operating leverage, we delivered first half adjusted EPS of 245p per share. You can see the strength of this performance for yourself with first half EPS up 17% year-on-year and representing 15% compound annual growth over the last 3 years. The significant allocation of capital to buybacks has seen EPS growth consistently outstrip profit growth.

Now turning to non-underlying items. This continued to reduce as expected with the amortization of intangible assets relating to the Refinitiv acquisition 5 years ago, the main item.

On to cash flow, which grew very strongly, up 29% in H1 to GBP 1.2 billion. Large cash items like working capital and CapEx were unchanged year-on-year. So the big increase in our cash flow simply reflects our increased EBITDA, converting directly into our equity free cash flow. This is the cash generative nature of our business model in action. And then ongoing buybacks means this 29% growth in free cash flow translates into a record 37% growth in free cash flow per share.

We continue to be very active in our allocation of cash, which you can see on this slide. We returned GBP 2.6 billion to shareholders in H1, GBP 2.1 billion via buybacks and GBP 500 million through dividends. We pushed particularly hard on the buybacks given the dislocation we saw in our share price for much of the first half. We plan to execute a further GBP 1.4 billion in share buybacks by the time of our full year results in February 2027. And just today, we have kicked off the latest tranche of this buyback.

With our results today, we announced a 17% increase in our interim dividend to 55p per share, consistent with our progressive dividend policy. Shortly after the period end, we reached agreement to acquire a further roughly 1% of LCH Group from minority shareholders for EUR 70 million. We expect that to complete in the second half. We ended June with net debt-to-EBITDA of 2.1x in the middle of our stated leverage range.

We are very confident of delivering on all our financial guidance for 2026. At Q1, I said the very strong market performance meant it was likely our full year revenue growth would be in the upper half of our 6.5% to 7.5% guidance range. With strength continuing, we are formally raising guidance for revenues to grow between 7% and 7.5% this year. As explained earlier, I'm also raising our margin guidance and expect a full year improvement of around 100 bps. And we are on track to deliver full year capital intensity of around 9.5% of total income and equity free cash flow of at least GBP 2.7 billion.

So in conclusion, we are executing well on our strategy, and we are very confident of delivering on all our promises for 2026. Aided by the multiyear contractual visibility and growth of our LDA agreements and the deep partnership we have with our customer, we are also confident in our medium-term delivery as laid out on this slide.

Now I will hand back to David to talk more about our strategic progress, particularly in AI.

Thank you, MAP. A really strong financial performance in H1. As I mentioned at the start, I'm going to talk about how we are becoming an increasingly critical partner to our customers in data and analytics and how that is playing out in our customer engagements. First, a quick recap. The basic ingredients for AI are data, compute, i.e., chips and data centers, and the model. What we can all see over recent months is that compute is an arms race, but ultimately driven by supply and demand. The model landscape is also shifting. Cheaper models are often open-weight and are closing the performance gap on frontier models. Businesses will orchestrate and optimize.

And as for data, data is more important than ever, making LSEG the enduring partner of choice in an AI world. 90% of our data revenues come from real time or data that is proprietary. We have always had unmatched global reach as well as breadth and depth of data. We have, for decades, been embedded in customer workflows and are becoming more embedded, providing regulated, integrated and secure solutions.

Our data is structured to optimize AI performance, driving repeatable and deterministic outcomes. And now we have added massive new distribution through our partnerships across the AI ecosystem. We are becoming an increasingly critical partner for the industry, much more than just a data provider. We are partnering with customers to design and implement multifaceted AI strategies with our data at the center of them and engineers from LSEG, Microsoft and AWS helping to deploy them. Our customers are facing complex challenges in adopting AI into their processes and workflows and the landscape is evolving rapidly.

Let me highlight why LSEG is so well placed to help our customers navigate these challenges. First, regulation. The industry is already heavily regulated and the pipeline of new regulation is growing day by day. In the appendix, we've produced a summary of the various regulations that govern the use of data in the financial services industry. It gives you a good sense of the regulatory weight and complexity our customers face. This is a core capability for us given our decades of experience supporting customers to manage regulatory risk and change.

Next, cyber risk. The latest models are highlighting cybersecurity vulnerabilities in seconds. LSEG is already deeply embedded in the processes and systems of the world's biggest financial institutions and brings a critical market infrastructure mindset of provision and protection of data. Resilience and security are nonnegotiable.

On IP protection, customers are concerned about the risk of commingling their data in a multi-cloud or frontier model environment or giving away their thinking through their prompts. We have worked with customers' confidential information for decades. They know we'll provide them our trusted data and work with their confidential information in a secure environment.

Similarly, on AI sovereignty, global businesses need to maintain flexibility to use different models in different markets. Our open approach, model and platform agnostic meets that need, whether customers prefer to use an orchestration platform combining multiple models or individual leading models market by market.

Accuracy, I think, speaks for itself. You all have experienced the limitations of even the best LLMs when based on Internet data, answers that are often incomplete, inconsistent or made up. With our accurate, auditable and semantically linked data, you are getting the same outputs time after time.

And finally, of course, token costs and ROI. A number of companies have spoken about the challenges emerging here. We can make a big difference in helping customers manage token spend. Partly, it's about being model agnostic, so customers aren't using frontier models for simple prompts. And partly, it's the way our data is structured and presented to models, which reduces superfluous information and repeat tool calls. Our Head of AI, Emily Prince's recent blog on this topic is worth a read for more detail on this.

In summary, some have been too quick to project the rapid consumer adoption of AI chatbots or the dramatic impact AI has had on coding onto the enterprise AI space. As we've said before, our sector moves slowly. Given the range and complexity of issues to address, this is a marathon, not a sprint, and LSEG is the best running partner.

Next, I want to give you a sense of how AI solutions are evolving. You may remember, we showed a diagram like this at the Innovation Forum last November. This framework continues to evolve. We've also shown on the right-hand side, the customer considerations at each level of the framework to tie into the previous slide.

A couple of key points. One thing that hasn't changed, LSEG's trusted content from data, indices and analytics is a key foundation. On distribution, we are seeing larger customers, in particular, choose to leverage our existing distribution to bring data into their own AI stacks with MCP as an add-on in specific use cases. And then in the consumption layer, we're seeing a blurring of lines and an increasingly hybrid approach. Workspace is stretching beyond the core user interface. Customers are now looking to access it via the Microsoft Teams app, which will allow deep interoperability with Open Directory and other Microsoft products. And we're also working with some customers on what the software industry refers to as a headless approach, enabling them to access the intelligence and content of Workspace in any environment and UI.

You'll see that clearly from the case studies. Some customers are taking that hybrid approach to AI adoption, combining our UI with their own solutions and third-party platforms. To take stock on our progress with AI-ready data and product, let's start with MCP, where interest continues to be strong. We've engaged with over 200 customers on MCP since launch late last year with a good spread by geography, customer type and channel. Usage is really ramping up as customers, both humans and agents engage with the data. We saw tool calls increase nearly 5x from May to June. And we're adding a lot more data over the next few months, which is a key ask from customers.

And it is not just D&A. We're rolling out MCP access across the group. The FTSE Russell fixed income Sandbox, which we demoed to you last year, is available via MCP, and we're getting some good lead generation out of it. Broader FTSE data is coming soon. In our Markets business, MCP will be a key interface for LSE 24, which we announced last week as we see agents playing a greater role in trading in the future.

Now MCP is an important new distribution channel, but I should emphasize it represents around 1/3 of our current AI-related commercial discussions. Although there's been a lot of focus on MCP as an AI channel, AI usage of our product is accessible by more than MCP. You'll see that shortly in the depth and breadth of our customer engagements.

Turning now to Workspace. We have seen a very strong pace of development, both in AI and more widely. Our AI search tool is now generally available, rolled out to all Workspace customers during July. Although we have not marketed it widely to customers yet, we already have 17,000 active users with these numbers growing every day. For the deep research tool, which many of you have tried, the number of users has quadrupled from Q1. Both search and deep research are built on leading models. We're adding more data and enhancing workflows on both tools.

We also have a third AI product in Workspace, Company Intelligence. This is actually the grandchild of Meeting Prep, the first prototype that came out of the Microsoft partnership, and our customers really like it. We're seeing users pull 3,000 or so detailed company reports per week from multiple underlying sources. You can see examples of feedback on the right here, but we have much more, and we get plenty of feedback asking for additional functionality, which just helps us make the product even better.

But as you know, Workspace is way more than the AI tools we're building. It remains a critical workflow tool for traders and a rich source of community and data and the impact of the enhancements we are making continues to scale. In H1, we've integrated the vast majority of FXall functionality into the platform, driving a 10% uplift in engagement. We've invested in the messaging function, which has 40,000 monthly active users, 1,000 customers are piloting our new private markets data sets. In H2, we'll be rolling out interoperability with Tradeweb. That work went into production this month. And as I mentioned earlier, Workspace is also breaking out of its traditional UI as we make its data, intelligence and tools available in customers' own environments as well as the Microsoft ecosystem.

There's real product momentum with Microsoft. The Workspace app is already available in Teams, offering all the AI functionality of the main desktop and deep interoperability between the 2. It will shortly be available in Copilot too, which is significant given the 1.5 million Copilot users in our top 50 customers.

Open Directory rollout is also continuing with over 20 customers onboarded. We're now using it as the default communications platform for new TORA OEMS customers with 3 signed up and we'll make it interoperable with LSEG Messenger's 40,000 active users in H2. So we've made significant investment and progress on the product side. The pace of innovation across LSEG is at its fastest for many years.

This table lays out how we are monetizing this investment, and this is likely to continue to evolve. We are out in the market with this framework today. In fact, customers are demanding it. While we are primarily focusing on adoption, some customers really want to understand what the cost will be as they are signing up.

For use of LSEG data in AI applications, the basic commercial model is an additional use-case license. This is consistent with how we charge for data on any new or additional use case. Where customers take a bulk feed or stream data, we don't have instant visibility on usage. That's the category on the far-left column, where customers are accessing data via API, either directly or through our MCP, that will attract an additional usage-based charge. As AI and MCP drive cross-sell, we expect customers to take additional data sets over time as well.

For our Workspace AI tools, we're taking a slightly different approach. AI Search is included in the Workspace subscription with the value reflected in the annual price review, but will also be subject to a fair use policy, reflecting a certain number of prompts per month. Above that, there will be additional usage-based charges. We are positioning Deep Research as a premium add-on with usage-linked tiers.

As you would expect, our pricing structure reflects our costs. These new products and use will drive additional cloud costs for LSEG. On the AI-ready data, we incur some data platform fees and on the Workspace AI functionality, we incur token costs. These costs are fully factored into our midterm margin guidance.

Let's look at how we are working with customers to implement their AI strategies. The first case study is a global bank with a long-standing enterprise agreement or LDA. We're working with them on multiple fronts, which will involve our own forward deployed engineers. The customer is building a couple of platforms for different user groups that combine their own data with our data. One of these will help relationship managers prepare for meetings, bringing their own internal regulatory and product data together with LSEG news and market data. Another will help the banking and capital markets teams access deal intelligence and client-related news flow. We're also supporting them with MCP access to news, fundamentals and ownership for their wealth advisory business. As per the previous slide, we will monetize this through the AI license and the MCP capability license, including tiered pricing for consumption.

Page 32 features our work with a sovereign wealth fund client. We already provide them with significant foundational data to support investment management insights. Our new collaboration goes much further. We are combining our entity, symbology and ownership data with the customers' own data and other sources to underpin 3 specific use cases: A risk intelligence agent to identify emerging threats and potential portfolio impacts; a counterparty agent to help risk managers identify credit risk factors and a C-level dashboard, bringing together a number of sources of data and intelligence in one place for portfolio monitoring. We're delivering data both via MCP and directly through our existing API. And again, the commercial model reflects this. Note that there is a separate and additional AI license for Risk Intelligence.

And the third, a long-standing industrial customer, which may surprise some of you. We're helping them build FX hedging workflows, combining multiple data sources and AI and also providing treasury insights from structured and unstructured content. This example highlights the potential that our AI and data have for all companies, not just financial institutions and shows how supercharged distribution and usability can open up new markets for LSEG's data.

We picked 3 case studies. I could have shared a lot more of similar depth and breadth. They all demonstrate the value we're bringing to customers, the longevity of our relationships, the importance of our trusted data in a highly regulated sector, our open and flexible approach and our platform-agnostic stance to distribution. While these examples do leverage MCP, this is not just simple plug-and-play. These are complex, sophisticated and multilayered solutions and reflecting on the whole AI disruption story. The market has been debating these topics in great detail for the last 12 months and having what we could call the terminal value debate. In the appendix, we have addressed 5 common misconceptions about the future of our business in an AI world. You've heard us make many of these points in meetings and Q&A, but we have pulled them together in one place as a reference source.

So to wrap up, financial performance is very strong with 8.4% organic revenue growth, accelerating subscription revenue growth, strongly improving margins and 37% free cash flow per share growth. We're driving an unprecedented pace of innovation across the business. We will come back later in the year with a deeper dive on that innovation in markets. And we have returned GBP 2.6 billion or over 5% of our market cap to shareholders in H1 alone, with more to come in H2, starting today.

But just as importantly, you'll notice today the clear shift we are driving in the AI debate based on what we are seeing day-to-day on the ground with hundreds of customers. AI and financial services can drive enormous value, but it comes with significant challenges for our customers. We are the trusted partner to help them address those challenges. We have the infrastructure, the data, the trust, the regulatory expertise and the institutional history. LSEG is even more valuable in an AI world.

And now we will be happy to take your questions. Peregrine?

Thanks, David. [Operator Instructions] Operator, over to you.

[Operator Instructions] So your first question comes from the line of Andrew Lowe from Citi.

It's been a year since the AI disruption narrative really took hold. Could you please provide a little bit more color and specific examples about how LSEG has been affected by AI during the period? What are the biggest changes versus your expectations 12 months ago, both positively and negatively?

Thanks, Andy. So really, the biggest issue by far has been dealing with the perception of the impact of AI versus the reality of the impact of AI. And really more recently, over the last couple of months, I think it's fair to say the level of understanding about AI's potential, what it's good at, what it's not good at, that has matured a lot. I think people now recognize that a frontier AI company is not a data provider, not directly providing what we do. In fact, it's now well understood that for an AI company to generate value for enterprise customers, it actually needs a high-quality provider of data like us.

Over the past year, there has been speculation that AI would wipe out large parts of our business. And in fact, it's just the opposite. AI has enhanced the value of LSEG. AI has increased the need for and therefore, the value of our data because our data is verifiable, it's auditable and its proprietary. And if you look at our performance, our performance demonstrates exactly that. If you compare where we are today versus a year ago, our new sales are 10% higher. Our retention is better. Our subscription revenue growth has accelerated to 6% -- I'm sorry, from 6% last year to 6.3% now. So we're seeing more consumption of our data than ever before. We've got new distribution channels and new products that we didn't have a year ago, and we're getting great traction with them with thousands of users. And we are more closely engaged with our customers than we were a year ago. We're creating value from that engagement.

So that's why we talk about LSEG being a lot more valuable in an AI world. I think it is fair to say the world is moving faster today than a year ago, and it has been a challenge for our people to keep moving faster to really integrate new tech into our products and processes and meet customer expectations in this really dynamic market. But I think we're really rising to that challenge very well. And I expect us to do that more and more and better and better going forward.

Your next question is from the line of Hubert Lam of Bank of America.

So going back to MCP, so how much can MCP add to growth going forward? Is MCP monetization incremental to that 7% subscription revenue target you have for next year? And if so, do you see upside to that now that MCP monetization is starting?

Hubert, it's MAP. So I think we've said very clearly in Q1, and we are reiterating that our priority for this year and for the second semester is to concentrate on usage. Our clients are still very much trying MCP very different use case. And for us, the most important is to make sure that we have the setup, which is the most powerful and valuable to them. So MCP, for sure, will be monetized. And by the way, we are already sending some invoices because the client actually asked us to have a price framework for the rest of the year, but it's minimal. And we will see that more in 2027, but certainly, it won't move the needle in 2026.

It could move the needle in '27 then?

We'll discuss. I mean, clearly, it's part of the acceleration of our subscription businesses. So clearly, it's going to be one more engine to this acceleration.

And maybe, Hubert, the other point I would just add -- yes, the other point I would just add as we just went through in the presentation is that MCP is important, but it is really about 1/3 of the commercial discussions that we're having with our customers. So there are other aspects to this as well.

And your next question is from the line of Mike Werner of UBS.

Just a question on the subscription businesses. We saw 6.3% revenue growth in the first half of this year. You guys are guiding to, I think, 6.5% for the full year. So we need to see another, let's call it, 30, 40 basis points of acceleration in the second half. So I was just wondering what gives you the confidence that you'll get -- what will get you to that 50 basis points of acceleration? And then just clarifying your answer before, when it comes to the subscription revenue growth and the 50 basis points of acceleration in 2027, my understanding is that MCP and the like would be incremental to that, not included in that. But if you could just confirm that, that would be helpful.

So first on 2026, yes, your math are right. So 6.3% in the first semester, acceleration to 6.7% in the second semester. And as we said, circa 6.5% on the year. So we're very confident to reach this 6.5% for the year, fundamentally for 2 major reasons. One is that we had gross sales, which were at record level, if you remember, in Q4 last year. And these gross sales are executed not only at the beginning of the year but for some of them in the second half of the year. So it's something that we already know. So it's giving us a good visibility on the installation pipeline over the coming quarters.

And the second reason is that we have improved massively, as you've seen in David's presentation, our product lineup, not only for D&A, but for the 3 subscription businesses. So we have a far better product lineup. So the combination of better product and the pipe that we know is going to be executed in H2 is giving us this confidence.

As for 2027, you want to cover it, David?

Sure, happy to. So Mike, with respect to 2027, the way this will play out is that we will see slow, steady adoption of these products and therefore, the revenue associated with that. So we don't expect and you shouldn't expect a big spike at any point. I think we've been really consistent about that in terms of how this business -- this industry works. But you have seen us very consistently turning the dial up over the last several reporting periods. You can hear MAP's confidence in terms of what this year will look like for subscription revenues, and we expect that to continue going forward with that kind of slow, steady adoption curve, if I can put it that way.

Your next question is from the line of Benjamin Goy of Deutsche Bank.

Also a question on the MCP connector, please. I noticed that the share of direct connections to LSEG has moved up again rather than via the LLM. Just wondering whether this is now the sales force is in place and you're pushing the product more directly or what is driving that? And yes, if that is a strategic target for you?

Thanks. It's not something that we are pushing. It's really customer demand, and this is how we see the market evolving. There are some customers who want to access our data through MCP. And then there are other customers who may want to access some of our data via MCP and some of our data through other channels. They may want to take it through a regular API. They may want to access it in, for example, a Snowflake or Databricks environment. And we're just seeing this market continue to evolve and continue to develop. And this is, in many ways, one of the strengths of LSEG Everywhere. We are in a position to serve our customers across the different channels they want to use to access our data. As I mentioned earlier, MCP is the channel in about 1/3 of our commercial discussions right now.

And then to your point on direct versus other providers, that's also what we're seeing in the marketplace. In other words, a number of our customers are choosing to go direct instead of using one of these model channels. So this will continue to evolve. We'll continue to share with you all what we're seeing in the way that our customers want to access our data. But from our perspective, it's all good.

Your next question is from the line of Arnaud Giblat of BNP.

Just another question on MCP usage, 202 clients is a big number. I'm just wondering if you could give us a bit of an indication as to what share of revenues these clients represent of your revenue base. I assume it's the largest clients are adopting. And if I may, a quick follow-up. You highlighted OTC revenue growth being really strong. I'm just wondering if you could pick out which areas are seeing -- within OTC are seeing the strongest contribution to that growth.

Sure. So I'll touch on the MCP question, then MAP can answer your question -- your second question. It's actually all over the map in terms of the customers that we are seeing access our data via MCP and via these other channels that I'm talking about. And we have seen a number of our very large customers doing some interesting things, and we have mentioned this in one of the case studies. We're also seeing a lot of smaller funds, hedge funds, asset managers that are really interested in the product and accessing our data in this way. It's also really interesting to see -- it's not fully transparent to us, but we can tell pretty much which users of the data are humans versus agents. And it's very interesting to see the -- we've been -- I think you all asked us on one of the prior calls what the differences were in terms of consumption of our data by agents versus humans.

Take this as anecdata. This is not scientific. But what we see so far is that agents tend to consume roughly 10x, roughly 10x the amount of data that humans do through the MCP channel. So I think it continues to evolve. And maybe the last point I would just reiterate is that MCP is, at this point, just about 1/3 of the AI access and the AI commercial discussions that we're having. So important, a great new distribution channel, but part of what we're seeing and part of the opportunity set that we are taking advantage of with our customers.

Yes. On OTC derivative, it was indeed a great semester with both volumes and new product. And we see the growth being double digit on both SwapClear and RepoClear. So it was very much distributed between our different platform.

Your next question is from the line of Oliver Carruthers of Goldman Sachs.

Oliver Carruthers from Goldman Sachs. Just one question for me. On Data & Feeds, the organic constant currency growth rate has now risen 100 basis points over the last 2 quarters. It's now running at 7.7%. It looks like it's set to overtake workflows as your biggest revenue single line item by the end of this year. It was only GBP 3 million shy of this in the second quarter. I think Slide 10 looks pretty compelling to me in terms of the client consumption of some of your key offerings in the here and now. And as you say, potentially future AI consumption may be additive to this. Just in the context of the 7.7% growth rate, just how should we think conceptually about where this growth rate could go from here and some of the aspirations for this line item?

Thanks, Oliver. So if you go back to our original Investor Day or Capital Markets Day after we acquired Refinitiv, we talked about the growth rates of these 2 businesses. And we expected at that point, workflows to be low single digit. And I think we talked about Data & Feeds to being higher than that. I think at that point, we talked about it being in mid-single digits. And so that has played out over the last several years. We have seen -- and I think we've got this in our materials in one of the appendices, a graph that shows how we have seen a significant reduction over the last 15, 20 years in the number of headcount, the number of people in this industry, and yet we've seen a doubling of the amount of data consumption and data spend.

So you have had a clear decoupling of the demand for data from the number of people in the industry. And that all predates AI. It's important to be really clear about that. That dynamic was long before any of us were talking about the impact of AI on our business. I think going forward, we continue to see a really attractive opportunity for our Workspace interface, and that includes this notion of a headless construct, if you will, in terms of we already have Workspace available through Teams. Workspace is going to be available through Copilot with 1 million-plus users among our top 50 customers. And we have that flexibility, that modularity to make the Workspace content available for our customers in the way that they want to consume it effectively through their user interface. So we think that kind of flexibility is a great opportunity for Workspace for a human interface.

And then to the specifics of your question, Data & Feeds has been a great business. We have been adding a lot to it in terms of both new data sets and new distribution channels and AI really just turbocharges that. I think it adds new distribution channels, whether it's MCP or other ways of consuming our data via AI models. And we are seeing good strong growth there already, and I expect to see that continue.

Your next question is from the line of Ian White of Autonomous Research.

Just given the tailwind from rising markets on the asset-based fees since we last spoke at 1Q results, why is the outlook for subscription-based revenues not improved from the 6.5% that you indicated at 1Q? To put it really precisely, I mean, the ETF AUM is 17% higher quarter-on-quarter at 2Q. That should be about a 20 to 30 bps increment to overall subscription-based revenue growth in 2026. So why is the ambition not higher now than the 6.5% it was previously, please?

Yes. So I mean there are 2 reasons for this. The first thing is that our asset-based revenue is relatively small, as you have seen. So even if you have in there a growth which is more than expected, it's not moving the dial at subscription business completely. That's the first reason. And the second reason is that the part of the agreement we have in that business is not directly linked to volume and it is flat fee. So the combination between the 2 is why we confirm the 6.5% for the year with an acceleration at 6.7% in H2.

If I can possibly just come back on that. I mean the -- without want to get into too much detail, R-squared between your ETF AUM and the asset-based fees 1 quarter ahead is greater than 0.9. So there is quite a strong link between the ETF AUM and revenues in the subsequent periods. And as I say, just taking where we are at 2Q and kind of running ahead, that's 20 to 30 basis points on the entire subscription base. So that is significant in my mind. Is it just something that you've not factored in? Or is there something going in the opposite direction that gets us back to 6.5% for the year, please?

I think it depends on mix between U.S. and global really in terms of asset base. We don't have the same agreement for one and the other. And we look into H2 with confidence, I mean, I understand your calculation. But, again, we're talking about 10 basis points at subscription businesses level, and we said circa. So it's -- I think we are already relatively precise or at least I'm not going to be more precise than that.

Your next question is from the line of Julian Dobrovolschi of ABN AMRO.

I'm sorry to come back on the MCP, but I really want to get something straight there. So I understand that it's not really a driver for '26. It's a small one for '27. And at the same time, operational momentum you reported already on is pretty strong in my view, and you also anticipate this to be robust in the future. So my question is, when should we really expect then the MCP strategy to generate meaningful revenue? And also, how can we crosscheck that with the critical mass on the client base side? So you have 200 now. What will be kind of a level of client base that would be kind of a good reflection for generating meaningful MCP revenue?

Guys, you're all trying to build mathematical formulas into models as to exactly how this is going to play out in 2027. Let me just tell you, we have great confidence in the client adoption of our channels. We are seeing consistent, steady acceleration of both the consumption. We have put out the monetization framework today. It is the framework that we have already seen some of our customers engaging on. And as MAP mentioned earlier, we are already monetizing that. And it will be a consistent, steady contributor to our growth. And as MAP has already indicated, we have driven acceleration of our subscription revenue over the past several quarters, and we expect to continue driving that.

So we're not going to give anything more explicit or more specific than that. We, of course, understand why people are asking. But that is how we expect this to play out, and we have lots of customer engagement and customer proof points to demonstrate that.

Your next question is from the line of Thomas Mills of Jefferies.

Could you talk a bit about momentum around LDA wins? I guess we've seen a few less of those publicly announced of late. But could you give us a sense of what's happening beneath the surface? I guess we've seen LDA contribution to D&A ASV increase from 16% to 18% half-on-half. Could you also comment how the pipeline looks? And then slightly adjacently, I guess one of your competitors has recently spoken about sales cycle getting blown out due to complexity of negotiations around AI-related data consumption. I think you've kind of alluded to something similar. But do you have any sense of when we might expect that to start to normalize when commercial models become more standardized?

That's really interesting because I'm going to link the 2 parts of your question there. So first of all, on LDA, we've signed up a couple more this year. And as you said, the percentage has gone from 16% up to 18%. No huge ones in the first half of the year, continuing ongoing discussion and dialogue with various customers. And I would say with respect to the sales cycle commentary from one of our competitors, I don't agree with that actually. We're not seeing that. And some of that may be due to the strength of our LDA relationships. And what I mean by that, and again, you can see this in one of our case studies is that when we have an LDA arrangement in place with one of these customers, that significantly accelerates the engagement with that customer. And we are basically the first call, the default provider, and we can immediately start engaging with them as to how to build this capability for them.

And we have -- in a few cases, we have our people and in some cases, partnering with, for example, Microsoft people and the industry calls these forward deployed engineers. We've had it for a number of years as our implementation team, but happy to call them FTEs, working on the premises with our customers, building new agents, building new capabilities, making sure that they have access to our data through these new channels. So we have not seen the sales cycle extending. And we continue to have a really good, really robust dialogue with both existing LDA customers, but also with a number of new customers who are attracted by our offerings.

[Operator Instructions] And your next question is from the line of Michael Sanderson of Barclays.

Just a single question as expected, but a small add-on, if that's all right. So the single question was obviously talking a lot about the momentum and sales development. I'm just interested if you can talk me through the sort of the gross sales numbers that you talked about in your new set of metrics that sort of versus last -- end of last year and versus June now, minimal progress. Is there a seasonal element that we should see acceleration in the second half of the year given all the discussion you're talking about, I suppose, in that metric, just to understand.

And the small add-on, if you allow me, was just you're obviously working very closely with your clients on setting up tools and building out solutions. Does this translate into any sort of one-off fee setup fees, et cetera, that you get to benefit from? Or is it all rolled into a longer-term subscription model that you obviously run for the most part?

Thanks, Michael. I'll take your second question, and then MAP can answer the first question on the gross sales. So with respect to setup fees, as you call them, or implementation fees, we -- it depends is the short answer. And so for example, in a typical LDA arrangement, there are often embedded in that these kinds of consulting services where we will commit to a certain number of hours, if you will, of our consulting team going in there and helping build capabilities. In other cases, it is a separate cost to the customers, and we charge for that. And that can be kind of a one-off or in some cases, more periodic implementation fee. And so we see that in terms of both modes where sometimes it's included and sometimes it's incremental.

Yes. On the -- Michael, on the gross sales, I reckon, it's a new indicator that we are giving you. So you're trying to get your head around it. I think the important thing is that have in mind that it's a 12-month rolling that we are giving. And actually, the way I look at it is we had a step-up in -- as you remember, in December 2025 of about GBP 50 million, okay, compared to June 2025. So going roughly from GBP 430 million to GBP 480 million. And actually, I was extremely pleased to match this GBP 480 million in June, meaning that the step-up is now behind us. So I see that as a positive to be clear.

And this concludes today's Q&A session. I will now hand the presentation back to David Schwimmer, CEO of London Stock Exchange Group.

Well, thanks, everyone, for all the questions. And I'll close just by touching on one of the themes of the earlier questions. Here we are a year after the first wave of perceived AI disruption hit last summer. And there's now a year of evidence on the impact of AI. I can't speak for the whole industry, but I can certainly speak for LSEG. And we, as an organization, are moving faster. We're more efficient, and we roll out new product more quickly. We are seeing more consumption of our data. We're monetizing new distribution channels and new products, and we are doing more with our customers. And you all can see that in our results. We have higher growth, higher sales, higher retention, higher margin. And we feel as if we are just getting started.

So with that, thank you for joining today. MAP and I look forward to seeing many of you over the coming days and weeks to continue the discussion.

Good morning, and welcome to the LSEG First Quarter Results 2026 Investor and Analyst Call. [Operator Instructions]. I would like to remind all participants that this call is being recorded.

I will now hand over to David Schwimmer, Chief Executive Officer, to open the presentation. Please go ahead.

Good morning, and welcome to our first quarter results. I'm joined by our CFO, MAP; and our Head of IR, Peregrine Riviere.

Q1 was a record quarter for the group and a perfect example of the value of our model. Our leading multi-asset class trading venues have been critical sources of liquidity, price discovery and risk management, while customer engagement with our trusted data to inform their decision-making has reached new highs. This is reflected in revenue growth of almost 10%, the highest since the acquisition of Refinitiv 5 years ago.

This strong start puts us in an excellent position to deliver on our financial targets for the year. And as you will have seen from this morning's announcement, we expect revenue growth to be in the upper half of the 6.5% to 7.5% guidance range.

We continue to take an agile approach to capital allocation. In the first quarter, we used the dislocation in our share price to buy back GBP 1.1 billion of shares. Including dividends, we expect to return more than GBP 3 billion over the next 12 months.

Q1 was also a quarter of strong strategic progress. We're continuing to innovate and invest to capitalize on the opportunities that the ongoing technological change across our industry is creating. Our LSEG Everywhere strategy is embedding our AI-ready data across financial services, driving further growth in March and April in the number of customers accessing our data via MCP servers. We're also transforming our own products with very strong feedback on the Workspace AI tools we introduced in Q1 and an exciting pipeline of additional enhancements this quarter.

The group's innovation goes far beyond AI. We executed the first transaction on our private securities market in Q1, expanding private market funding through our public markets infrastructure. We're making excellent progress on post-trade solutions in partnership with 11 global banks. We're building digital markets capabilities, including a Digital Settlement House and a Digital Securities Depository, and forging a new distribution channel for financial models through our Model-as-a-Service offering.

I'll say more in a moment about our strong commercial and strategic progress. But first, I'll hand over to MAP to give color on the record financial performance.

Thanks, David. Overall, as David said, it was a very good quarter and further proof of our all-weather model. It was a strong quarter for our subscription businesses. All of them accelerated in Q1. Data & Analytics was up 5.1% as the strong growth sales at the end of last year flow through to higher revenues. We saw particular strength in Data & Feeds up 7.3%. The contribution from pricing and retention in D&A was unchanged compared to last year. FTSE Russell was up almost 9%. Subscription revenues accelerated as the rate of contract renewals normalized, as we said it would.

Growth in asset-based revenue was also strong, reflecting product inflows and higher market levels. And Risk Intelligence grew double digits, 10.5%, reflecting strong demand for our business critical screening and identity verification services. Together, those businesses grew 6.3%, a strong acceleration from the 5.2% last quarter and on track for our expectation of around 6.5% growth for the full year. The quality of our market infrastructure really stands out in the kind of market environment we saw in Q1. David will give you more detail on this in just a moment, but you can see the financial impact on that on this slide.

Markets revenue were up 15.5%, driven by strong performance across all the businesses. Cost of sales benefited from the action we took last year on the SwapClear revenue surplus. And as a result, gross profit was even stronger than total income, up 11.5% in Q1. Clearly, we have had a very strong start to the year. The outstanding performance from markets, combined with the great visibility we enjoy in our subscription businesses sets us up very well to deliver on all guidance for 2026. And in particular for revenue, we are confident in reaching the upper half of our guidance.

In addition to our ongoing investments in the business, we are also returning surplus capital. We repurchased shares worth GBP 1.1 billion in the first quarter. Just over GBP 400 million of this was from buybacks announced last year and nearly GBP 700 million was from the latest buyback announcement in February. Combining the rest of this year's GBP 3 billion buyback and dividends, we will be returning nearly 10% of our market capitalization to shareholders over a 15-month period. As a reminder, even with our high level of investments and large shareholder distribution, we expect to end the year around the middle of our leverage range.

This is all from me, and I will pass back to David.

Thanks, MAP. Customers increasingly want to use our data in AI applications, opening up a new distribution channel. We are embracing that through our LSEG Everywhere strategy, delivering AI-ready data to our customers in their preferred environment, embedding our data in their AI-powered solutions and agents. We're continuing to see strong uptake on MCP distribution. In the roughly 4 months since launch, we now have 90 customers who have connected to our MCP server directly or via one of our AI partners. And we have a pipeline of over 60 more customers looking to connect. This is great progress given the onboarding process can take a few weeks.

You can see from the pie charts that we are seeing a good global spread as well as broad-based interest across buy-side, sell-side and corporate customers. And we're seeing roughly half connect through Claude with the rest split between direct connections and other third parties. In terms of data sets, we are adding new ones to MCP all the time. Just this week, that included estimates, company fundamentals and corporate actions. And overall, we now have over half of our nonreal-time data available via MCP. So the platform is becoming more attractive every day.

Over the coming weeks, we will add transcripts, Lipper funds, FTSE Russell indices and much more. While we are currently focused on driving adoption, we're refining our commercial policies, and we'll share the framework at our H1 results. So strong progress on our AI-ready data, and we are also making great strides embedding AI into Workspace. Our Workspace AI search product is in pilot with around 1,500 users today, and we expect to launch general availability in the next few months.

Our Workspace AI deep research capability answers complex prompts with leading models from Anthropic, OpenAI and Google using our trusted data. We have around 1,600 customers in pilot and deep research is benchmarking very well against competitor products. We're adding much more data over the coming months and rolling it out more extensively throughout 2026.

Today, over half of the take-up is coming from the investment management sector, where we have traditionally had lower penetration, so a positive sign. We're also seeing really deep engagement with our products. When global uncertainty and market volatility rise as they did in Q1, our customers turn to us, a testament to their trust in our solutions. We saw record use of Workspace in Q1. Our oil tools, which have long been popular with users, saw a 75% sustained uptick in usage. Our shipping data experienced a threefold increase in demand.

In Data & Feeds, our real-time business data traffic grew 33% in Q1, and this has continued into Q2 with a new all-time high in early April. We're also really scaling up in some of the new channels we have added in recent years, making it easier for customers to access our data.

Following the enhancements we made in 2023, we have accelerated growth in our cloud-based real-time offering, Real-Time Optimised, and use of that platform rose fourfold in Q1. I've spoken before about the power of the analytics API we built in partnership with Microsoft. In Q1, we drove 44% growth in data consumption through that channel. And making Tick History more easily available via cloud-based solutions continues to drive strong demand with 39% growth in the use of that data in Q1.

Turning to our Markets businesses. As you know, we have intentionally positioned ourselves in areas of strong structural growth, driving the electronification of fixed income trading with Tradeweb, supporting cross-border flows in FX and helping customers manage risk and optimize their capital in our post-trade businesses. We achieved exceptional volumes in interest rate swaps on both our trading and clearing platforms as customers adjusted to shifting market expectations in Q1.

Market conditions also drove strong volumes across the rest of the fixed income franchise as well as FX. That was on top of the strong double-digit growth we have consistently been delivering in FX clearing.

In Equities, we also achieved strong trading volumes. Technology is accelerating the pace of change in our industry. We are investing and innovating to take advantage of that. Our index business, FTSE Russell, is expanding its presence in the digital asset space, attracting 8 digital asset ETFs to track its benchmarks in Q1. We're also seeing good demand for our private markets indices with StepStone. As markets digitize, we're on track to deliver 2 new digital markets capabilities, Digital Settlement House and Digital Securities Depository in Q2 and H2, respectively.

I'll pick out just one more example from this slide, Model-as-a-Service. We made financial models from Societe Generale available through this channel in Q1, the first time we have expanded our analytics API to third-party models. We're adding models from our post-trade business later this quarter, taking further advantage of the powerful distribution capability of the analytics API we built with Microsoft.

So to wrap up, this has been a record quarter of growth that puts us in a strong position to deliver on all our targets for the year. We're driving adoption of our AI-ready data across the industry through a range of AI partnerships. Our innovation is creating powerful new platforms for long-term growth. And we are returning significant surplus capital to shareholders, GBP 1.1 billion in Q1 and more than GBP 3 billion over the next 12 months. We're very excited about the opportunities ahead of us this year and beyond and are very well positioned for continued growth.

And with that, I'll pass to Peregrine for Q&A.

Thank you, David. [Operator Instructions]. Thanks, operator, over to you.

[Operator Instructions] Your first question is from the line of Tom Mills at Jefferies.

I think you've mentioned that you'll be looking to share more on the commercialization MCP as a distribution channel at 1H. I just wondered if you could give us a sense of your early conversations with larger customers, appreciating we're only about 4 months since launch. Is there a recognition on their part that this ultimately won't be included in existing agreement, will be [indiscernible] charges there? And just I noted that you said that you're seeing larger buy-side adoption in this channel versus the [indiscernible]. Why do you think that is?

Tom, we are definitely seeing an understanding and recognition from our customers that this is incremental. This is a new product, a new service. So it has been specifically laid out in our -- for example, our data access agreements. A big part of those discussions, those negotiations are around the existing perimeter of what we provide. And I think it's very clear to them that MCP and the AI distribution channels are outside of that perimeter.

So actually, a lot of the discussions that we are having with our customers are around their eagerness both to access the product and frankly, to understand what the commercial model will be. And so we are in early discussions with a half dozen or so about the commercial framework. And as we mentioned, we will be sharing that framework with the market in our half year results.

So on the buy-side, I think it's just the utility. I think our customers are finding it very helpful, attractive product, easy to use. And so we're not particularly surprised that we're seeing that kind of traction.

Your next question is from the line of Mike Werner of UBS.

I appreciate the presentation. A question on the MCP server. Apologies, I'm going to be focusing on this a little bit. I guess, can you give us a little bit more color as to the economics of the MCP server? If we think about you setting it up and the investment, how should we think about ultimately the variable costs? Is this something where there's a lot of operating leverage or there is a significant amount of consumption-based costs tied to the usage of the server?

Mike, it's MAP speaking. So in terms of economics, as far as MCP is concerned, a couple of points that I can make. As our clients are using LLM models to access MCP, so being OpenAI, Claude or Gemini. It's our clients who are paying the tokens to the LLMs. So this cost is with our clients. Then the cost we have for MCP is mostly coming from 2 things: First, the cloud cost and the cost of the data platform. Both of these costs are indeed variable. So that's something we want to take into consideration while we are establishing the commercial policy for this new product.

And your next question comes from the line of Hubert Lam of Bank of America.

I've got one question. On D&A growth, it was 5.1% in the quarter and only up marginally from the 4.9% in Q4. Can you talk about the different dynamics within the division where it seems like Data & Feeds had decent growth, but workflows slowed marginally? And also, I guess you touched upon it in terms of the enhancements in the Workspace, I guess would this be helpful in terms of driving up further growth within Workflows in terms of pricing or greater demand in the future?

So I would not overinterpret any modest tick up or tick down in terms of workflows in particular. We continue to see really strong interest in the new functionality of Workspace and interest as well in terms of the new functionality that is Open Directory and how that will continue to be expanding over the course of this year and beyond. So we'll continue to add capabilities, add functionality, add product in there, new private markets data in there as well, which is also getting some good interest. So I wouldn't get -- as I said, I wouldn't overinterpret any kind of modest ticks up or ticks down in terms of where workflows are.

And then Data & Feeds business is doing very well. We touched on this in the presentation, but very high demand for the content that we're providing in Data & Feeds as well as Workspace. And we will continue, as you know, to invest in that platform and look forward to continued growth there.

Maybe just the last point -- sorry, Hubert, last point I should emphasize. I think everyone knows this, but just to be clear, no MCP revenue in here.

You have a question from the line of Arnaud Giblat of BNP Paribas.

Yes. Just continuing a bit on the MCP theme. I'm just wondering, out of the 150 clients that have signed up or signing up, how many are new clients to you? Are there any substantial new logo wins of size? Just wondering how this is driving incremental growth in the business?

Arnaud, I cannot give you that answer off the top of my head. What I can tell you is that it's a broad range. We're seeing some large institutions like the big global banks. We're seeing smaller institutions like hedge funds. One dynamic that I can share with you is that the onboarding process can be much quicker with some of the smaller institutions. They're really eager just to get on. There's not a lot of focus or review on some of the compliance or regulatory aspects, whereas with the larger institutions, the onboarding process can take, I'll say, a few to several weeks. And there can be a couple of meetings where we explain the content, we explain how it works, go through a number of the security issues, then there can be some legal discussions and then there's the actual onboarding. So just in terms of timing, that's probably the area where at this point, I can give you the most insight that the bigger institutions tend to be slower than some of the smaller, more nimbler institutions. I hope that helps.

And your next question is from the line of Enrico Bolzoni of JPMorgan.

I just wanted to follow up on your very latest comment, David, on the -- for example, on the fact that it's faster to onboard a smaller institution. So on one hand, I would think on top of my head that it would be easier to generally onboard clients via MCP relative to what has been historically. But AI is a very powerful technology, and I think that there might be some concerns and risk in terms of the perimeter of the usage of data, what AI actually might end up using. So my question is, do you expect that as this type of connectivity increases as a proportion of your, let's say, total clients and total revenues, the sales cycle will actually expand or will it actually shrink over time?

I'm sorry, Enrico, when you say the same cycle, I just want to make sure sales cycle.

Yes. So basically, it's going to take -- you think over time, over the next, let's say, 3 years, is it going to take longer actually to onboard clients or actually it's going to be faster, so you'd be able to do it quickly. I'm just concerned about all the implication of AI for risk, for securities and making sure that the perimeter is well defined. I know there's a lot of legal implications when contracts are signed that involve AI technology.

Yes, I would expect it -- well, first point I should make, it's already quicker relative to the historical onboarding in terms of what I'll call traditional or conventional products if we were setting someone up for a traditional API. So it's already quicker than that. And I would expect over time that it accelerates as our customers get more accustomed to the technology, as there is more and better understanding, particularly as we put our commercial framework out there later this year. This is all very new.

Just to remind everyone, we turned this on, I think, December 23rd. And so we're just a few months into this, both in terms of having our own data sets available in this manner and in terms of our customers really figuring out how to use it. And so a number of them have been in what I'd describe as exploration mode here. But as the comfort level increases and I'm sure that on our end, we'll look to facilitate and accelerate our own processes as well, I would expect to see the sales cycle actually becoming a little bit shorter.

Your next question is from the line of Julian Dobrovolschi of ABN AMRO-ODDO BHF.

I have one on the subscription growth. Wondering about the sustainability of it. So you ended the quarter at 6.3%, which I think is quite healthy. But I think you also indicated that this is partly attributed to normalization in FTSE Russell mandates renewals. So I was just wondering how much is from onetime boost the performance that we have seen in Q1 versus a structural step-up in underlying run rate, please?

Yes. So just to reframe the conversation. So we posted indeed 6.3% for the subscription business in Q1. We reconfirm our guidance of 6.5% for the entire year, which would mean that in the next 3 quarters, we will be between 6.5% and 6.6%. In order to do so, we have a growth which is broad-based both in DNA, FTSE and Risk Intelligence. I have already indicated that we expect Risk Intelligence to carry on being double digit. As far as FTSE Russell is concerned, you're right on the fact that after 2025, which was a bit difficult, we see FTSE Russell going back on a growth trajectory to the high single digit that we used to have and an acceleration in -- progressive acceleration in D&A. So that's the 3 elements that is converging to 6.5% for the year. And as I was saying, we are very confident in it.

Your next question is from the line of Ben Bathurst with RBC Capital Markets.

My question is also on MCP. Presumably, there are also some customers that have elected not to take it up at this stage. I just wondered what the typical pushbacks you're hearing when this is the case? Is it that customers aren't ready or that customers are using other MCP providers or any other reasons? And are there any actions you're planning to take to address any of these points to push connectivity up through the year?

Thanks, Ben. So we're not seeing a lot of pushback. I think to the extent that we have had any questions, it's really been about the availability of certain data sets. So we've shared it with some customers, and they have been looking for particular specific data sets. And so sometimes if those data sets are not yet on, they're a little bit less interested. But as we mentioned this morning, we're adding more data sets all the time. We're now over 50% of all of our nonreal-time data sets available through MCP, and that continues, that just making it more and more attractive.

Your next question is from the line of Oliver Carruthers of Goldman Sachs.

Oliver Carruthers from Goldman Sachs. I've got another MCP question, which follows on a little bit to your answer to the last one, David. But it seems like some of your data and analytics competitors are also making their data sets available via [indiscernible] clients, via MCP servers, but they're only making their data sets partially available. So can you talk a little bit about your philosophy of how you're going to set the perimeter for what data sets you make available for your clients via MCP and then particularly in the context of your LSEG Everywhere strategy, which to me feels quite differentiated in this context?

As I think you all are aware, we're very comfortable making our data available through MCP. And we are adding more and more of our data sets to it. We think it is a very helpful and valuable distribution tool. We think it works very well in terms of, I'll call it, cross-selling. It's a much stronger cross-selling machine than any human could be. We have about 1,500 data sets. And so if you are submitting a query through your model that goes into our MCP server, the way that works is that it is looking across the data sets that it has access to, to respond to that query.

So it is a very powerful natural cross-selling machine. It's also a great lead generation machine because to the extent that we have data available in our MCP server and a customer does not have the license to that data set, then we can structure it so that, that becomes lead generation for us. And then we can interact with that customer and let them know that there is data available that would be responsive to their queries and expand their licensing.

So I understand some of our competitors have more of a closed box mentality to this kind of opportunity set. That's not our approach. And from what we hear from a lot of our users and customers, they prefer our open model in this new era of very powerful AI distribution channels.

And if I may just add, David, we are adding data sets on a fortnight basis. Actually, we added yesterday, Reuters News and macroeconomic. So now we have Reuters News, we have fundamentals, estimate peers, and of the pricing corporate action, ESG ownership, company officers and directors, macroeconomics that we just put yesterday night. And in front of us in 2026, as mentioned in the slide, the major one that are awaited by our clients is deal and ownership data and transcript and filing. And then we will add commodities and Lipper fund data and finally, FTSE Russell. So you see it's a very busy pipeline of data set onboarding that we have in front of us.

Your next question is from the line of Ian White of Autonomous Research.

I'm also on MCP, if that's okay. Maybe can you just elaborate a little bit more around the strategy with respect to MCP. I see that you kind of led with sort of real-time pricing data Tick History, while others have led with maybe more sort of company fundamentals, transcripts, kind of research content. Is there any strategic reason that you sort of see it differently to peers in terms of prioritization? Or is it a case of adding what is readily available more or less as quickly as possible?

And can you just elaborate for us what's the end state here? And when will we reach that? Do you anticipate having more or less everything available via MCP in the medium term? And when is the medium term effectively?

Thanks, Ian. So just, I guess, I'd say a slight correction. We do not make our real-time data available through MCP because of the latency requirements of real time, that is -- could you do it technically? Yes, you could do it. It's just given the current construct and the customer demand, that's not practical. So it's really just a function of making the data sets available in part relative to what we see in terms of customer demand and in part, making sure that the data sets are structured in a way so that they can be interoperable.

And this is actually an important point that people often don't get. If you put a bunch of different data sets in an MCP server sort of willy-nilly, and they're not structured in a way to be interoperable, that can confuse the model in the same way that if you have a model accessing different data sets from different MCP servers that are not designed to be interoperable, that can confuse the model. So we are making sure that we're providing our data sets into our MCP server in a manner such that they are all designed, architected to be interoperable so that a model that is accessing data or content through our MCP server is going to get a very consistent experience with the interoperability amongst the different data sets, which just leads to better performance, higher accuracy in the model. So that's an important point that sometimes gets lost in terms of understanding how this works.

In terms of end state, I expect that we'll have the vast majority of our DNA data available this half. And then as MAP mentioned, there's more coming in terms of FTSE Russell and other data from broader parts of LSEG over the second half. So we see really significant opportunity there in terms of creating an MCP channel to access the vast amount of data that we have across LSEG.

No, I would just add, it's really about what David said, it's -- the reason why we are able to put new data sets on the fortnight on MCP is because these data sets have been rearchitected by our team through our partnership with Microsoft. So it's all the work that we have been doing at rearchitecting the data with Microsoft, which is now coming to fruition and which is allowing us to be so fast at getting the data set ready for MCP. So as David said, by summer, we will be done for all nonreal-time data sets.

[Operator Instructions] And your next question is from the line of Andrew Lowe of Citi.

I'll take one outside of MCP, if that's all right. There's been growing debate about your FXall business. So could you just talk us through the sort of planned investment within that business, where do you think you need to step up functionality? What's going to change over the next year or 2, and what the synergies are with the rest of your business?

So FXall has had a very strong performance, as you would have seen in Q1. We have been continuing to invest in the capabilities in FXall really over the past couple of years and continuing to improve in its functionality, in its speed, in the interface. And I would say probably the area to touch on for this year is the fuller integration from FXall into Workspace and the opportunities that, that brings with this integrated front-end system.

We've got FXall also plugged in as of a year or 2 ago into Tradeweb. We have straight-through FXall execution capabilities into ForexClear, so the kind of end-to-end processing. So again, strong performance this year, continued investment and continued improvement in its functionality, and we think it's a great business.

You have a follow-up question from Arnaud Giblat of BNP Paribas.

Yes. Just in your prepared remarks, you talked a bit broadly about the momentum you're having in post-trade services. I'm just wondering if there are any specific milestones you want to flag here in terms of activity pipeline?

Arnaud, I just want to make sure I heard you right. The momentum in post-trade services, is that what you were asking about?

Yes, yes. And specifically the partnership with the banks.

Yes. Got it. Yes. It's going well. We're seeing -- we -- in Q1, we put trade agent out there, which is a very efficient, helpful platform in terms of OTC processing. We are seeing significant onboarding of new customers. And the real area of focus, now that we have the banks fully involved as of the announcement in Q3 of their investment, there's now active ongoing discussion across the business of really creating more integrated functionality. So when we talked about this business last year, you would have heard us talking about Quantile and Acadia and the different -- SwapAgent and different parts of it coming together.

Now it's becoming much more of an integrated offering, and there's good engagement and dialogue with the banks as partners in terms of where we're taking this business. So good progress and good onboarding. It, at this point, has good growth. It's not a huge contributor to the business yet, but we expect -- as you have seen us deliver on in other parts of our business, we expect a nice long runway of growth.

You have a follow-up question from Enrico Bolzoni from JPMorgan.

Sorry, just one follow-up to clarify as I think there's been a bit of confusion around it. Can you just please, to clarity, confirm that the derivative hedgings or the FX impact that you experienced in this quarter, that was about, I think, GBP 5 million positive, is not included in the reported constant currency growth rate, just for the detail to be clear?

Yes. Sure. No, I confirm that the embedded derivative impact of GBP 5 million is not recorded in the organic growth.

And this concludes our questions via the conference line. I will now hand the presentation back to David Schwimmer, Chief Executive Officer, for closing remarks.

Great. Well, thank you all. Thanks for your questions. To the extent you have any further questions, you certainly know where to find us. Peregrine and the team would be happy to hear from you and wish you all the best. Thanks a lot.

Good morning, and welcome to the investor and analyst call for LSEG's 2025 Full Year Results. [Operator Instructions] I would like to remind all participants that this call is being recorded.

I will now hand over to David Schwimmer, Chief Executive Officer, to open the presentation. Please go ahead.

Good morning and welcome to our 2025 full year results. I'm joined by our CFO, MAP; and our Head of IR, Peregrine Riviere.

We have delivered another year of strong performance and rapid strategic transformation for the group. Revenues grew 7.6% with all businesses contributing positively and Data & Analytics accelerating. Our focus on driving efficient and scalable growth delivered 210 basis points of margin expansion, a little over half of that organic, taking full year EBITDA margins north of 50% for the first time.

Adjusted EPS grew 16%, reflecting our disciplined execution throughout the P&L. We continue to invest in future growth with the Post Trade Solutions transaction in Q4.

We continue to deliver strong cash conversion with GBP 2.8 billion of dividends and buybacks in the year. And today, we've announced our plan to execute a further GBP 3 billion of buybacks over the next 12 months. We see a great opportunity to invest in our own shares during this market dislocation.

This strong performance is a direct result of the strong execution of our long-term strategy and the rapid transformation we're driving across our business. November's Innovation Forum provided insight into how we are innovating across LSEG as we deliver on our AI strategy. We are already seeing the fruits of that innovation.

Our LSEG Everywhere AI data strategy is embedding our trusted data into the AI tooling of financial services. It's only been a few months since we launched our MCP server, but early demand has been very strong. I'll give you some numbers on that later.

We're also innovating to capitalize on the accelerating pace of change in capital markets, building new platforms for growth in digital assets. We launched our digital markets platform last year and, at the start of this year, successfully piloted tokenized cash settlement via our new Digital Settlement House.

The strategic partnership with 11 leading banks we announced at our Q3 results is accelerating growth and helping us unlock the multiyear opportunity in Post Trade Solutions, a great example of our strong customer relationships and partnership-led approach, creating unique opportunities for growth.

Let's take a step back and look at our 2025 performance in the context of the multiyear delivery of our strategy. We have achieved a lot over the last 5 years.

Financial performance has been very strong with organic growth [Technical Difficulty] improving significantly. [Technical Difficulty] all of our businesses. Across the whole group, we've driven significant revenue and cost synergies, built better products and better infrastructure, integrated the operations and unified the brand and culture.

This is all still work in progress. And every day, we discover new ways to transform our business. But right now, I can see more growth opportunity in front of us [Technical Difficulty] trading, settlement and depository have huge potential for future growth.

But let me take a step back. Right now, the market seems to be taking a view on the impact of AI on our business. We do not agree with it, and all fact-based evidence would indicate the negative market narrative is wrong. We feel as confident today about our products, our partnerships and our prospects as we ever have.

[Technical Difficulty] entering into enterprise agreements, run some of the most rigorous procurement, risk and technology processes in any industry. They understand exactly what our products do, they know how their own workflow needs are evolving with AI, and they know the role of our data, analytics and infrastructure in their operations.

[Technical Difficulty] heavily regulated and risk-averse customers are going to rely on outputs compiled from the public Internet. That is how much you should be worried about. But more importantly, let's look at the 98% of group revenues that are not from public data.

I'll cover the rest of D&A in detail later. But in a nutshell, these revenues are derived from [Technical Difficulty] that are proprietary, used in regulated environments [Technical Difficulty] intelligence and particularly World-Check is an industry leader.

Two things that are not well understood about this business. First, its value goes far beyond the thousands of official sources. Our customers make 200 billion checks a year across 700 million of their own end customers. And once anonymized and aggregated, we can use the decision data to improve our own detection and matching capabilities in a huge and constantly evolving content set.

World-Check is the leading product in this space and we are able to continue to improve the product to extend that lead through what is in effect a massive and constant flow of customer contributions.

Second, history is important. Customers need to justify decisions they made about counterparties going back decades, for example, in high-profile tax or fraud cases. We have all that history with the information that was available at the time. AI cannot create that past record for customers.

And moving to Markets, 40% of our business. AI is a tailwind here, too, as more data consumption drives more insights, leading to more trading volumes and ever-growing demand for risk management. So our positioning is strong, and our strategy is working.

I'll say more in a moment about our strong commercial and strategic progress and the opportunities we are seeing with AI. But first, I'll hand over to MAP to discuss our financial performance in more detail.

Thank you, David, and good morning, everyone. It has been a very strong year of financial execution for LSEG. So first, some headlines, and then I will unpack this all in more detail.

Organic growth was 7.1%, slightly above the midpoint of our guidance and another year of organic growth above 7%. EBITDA margin improved by 110 basis points underlying plus another 100 from the Post Trade Solutions transaction.

We delivered this performance through a significant improvement in our labor cost ratio. And this strong growth, combined with operational leverage, translated into EPS growth of over 15%. So that was the P&L.

Now moving on to cash and capital allocation headlines. Capital intensity continues to trend down, as guided, but do note that we are still investing in our business at least twice the rate of our peers. The dividend is increasing by 15%, in line with EPS, and we doubled the rate of share buybacks in 2025. With the growth in cash flow and the reduction in share count, this translates into 14% growth in free cash flow per share, which is actually 60% over the last 2 years.

In summary, we are growing our business strongly. We are investing in our future growth, we are generating significant free cash flow, and we are being very decisive and agile in our capital allocation.

So let's cover revenue over the next few slides. On this slide, you can see that our growth is very broad-based with Risk Intelligence continuing its double-digit momentum and Markets growing high single digits against a huge year in 2024.

FTSE Russell continues its revenue trajectory and D&A accelerated on 2024. As you know, I like to look at our subscription businesses as a whole, and here, we have achieved 6% of growth for the year, as we guided to.

We will start with D&A in more detail on the next slide. We achieved good growth across all lines in D&A. In Workflows, we completed the migration from Eikon, the largest ever of its kind in financial markets. As a result, clients are using the platform more frequently, and we continue to innovate and improve it.

In Data & Feeds, we maintain our strong momentum. We are adding significant new data sets, particularly in private markets, and we have launched our LSEG Everywhere strategy for AI-ready data. As David will cover, the initial uptake here is very strong.

Our Analytics business is well advanced on its acceleration journey. In partnership with Microsoft, we have driven a strong acceleration through the Analytics API and have just launched the Model-as-a-Service platform with our first partner bank, Societe Generale, onboarding its own models.

Overall, D&A is posting a 5% organic growth, accelerating versus 2024, as communicated during our half year results.

Turning to the other subscription businesses, we continue to see strong momentum and healthy demand. In FTSE Russell, we have seen balanced growth between subscription and asset-based fees, and we expect the growth rate to improve again in 2026.

Risk Intelligence had another very strong year. World-Check, which represent the bulk of its revenue, continues to innovate from its position as market leader, launching World-Check On Demand for real-time updates. This platform is increasingly deeply embedded in customers' regulated workflow.

Our Digital Identity & Fraud business accelerated in 2025 with transaction volumes up 16%, and the launch of our global account verification platform.

I will spend a little longer on this slide to talk about some new KPIs we are introducing for 2026, there will be an addition to ASV. And from 2027, we will report only these new KPIs and we will be retiring ASV. We are doing this to give investors more insight into our commercial progress and with measures that are less volatile than ASV.

But let's come back to ASV. As you remember, I previously guided to 5.8% growth at the end of Q4, and we achieved a bit better than this at 5.9%. This reflects a very strong end to the year, which set up well for 2026.

And now on the new measures. Before beginning, I shall tell you that they cover exclusively our 3 subscription businesses: D&A, FTSE and Risk Intelligence. We have given you the baseline here.

Gross sales represents the annualized total amount of new business over the last 12 months, so not contract value but more annual recurring revenue across our subscription businesses. We performed strongly in H2 2025 with a rolling figure increasing around 11% over H1.

On revenue retention, we already mentioned that this typically sits in the low to mid-90s depending on the product. We are formalizing that today at 92.4% on a consolidated basis, you can see that it's pretty much stable on H1.

And finally, we are introducing a KPI that measures the level of innovation or newness in our product set, the new product vitality index, or NPVI. This measures the proportion of revenue from products that are new or enhanced in the last 5 years, giving you insight into how our investment into product is translating into revenues.

Taken across our subscription businesses, this figure sits at a very healthy 24%, growing strongly against 2024 and H1 2025. A significant proportion of this relates to Workspace, as you would expect, and reflects the substantial enhancements to the customer experience that the new product gives to customer. In other business lines, this index sits more in the mid- to high single-digit range, which we expect to increase over time.

Finally, we plan to give these KPIs, including ASV for 2026, twice a year as we see little benefit in reporting them quarter-to-quarter.

Turning now to our Markets businesses. These are incredible strong franchise, which I believe do not get the attention they deserve, and they continue to deliver exceptional performance year in, year out.

In Fixed Income, as I have already reported, Tradeweb had another very strong year with continued high levels of activity across all main asset classes backed up by great execution. And in Foreign Exchange, we recorded our best performance in recent years with 7.5% growth. In OTC Derivatives, our Post Trade businesses went from strength to strength, and David will detail them in a few minutes.

Finally, and for ease of presentation, we have shown Equities on this slide with some other lines from Post Trade. Our Equities business had a solid year with revenue up 5.1%. We launched our Private Securities Markets with the first transaction taking place right now, and we also went live with our Digital Markets Infrastructure, built in partnership with Microsoft.

So now on to EBITDA and the rest of the income statement. We translated the 7.1% organic top line growth into 11.8% growth in adjusted EBITDA, 14.3% growth in AOP and, finally, 15.7% growth in EPS. And as you can see from the main table, EPS growth was 19.4% on a constant currency basis. This is truly operating leverage at work, plus very good control in financing, tax and our share count.

Let's take each of those levers to improve earnings in turn. Number one is cost control with total OpEx up only 3.5%, half the rate of revenue growth. Within this, you can see that we have really managed third-party services very effectively, down 11.6% year-on-year. This is a core part of our labor strategy.

Total headcount is roughly stable, a small decrease of 700, with ratio of internal employees rising to 75%, driven mostly by engineering. As previously mentioned, this is not just about cost. We have seen significant upskilling and improvements to productivity as we build a true engineering culture.

As usual, we show the margin improvement graphically on this slide. Once you adjust for FX at either end, the improvement year-on-year is 210 bps. 100 of this relates to the SwapClear revenue surplus agreement, and that leaves 110 bps of underlying. Actually, the real underlying improvement was 140 bps, taking into account the minus 30 bps of the disposal of our Euroclear stake and its related dividend income stream that ceased.

On net financial expense, we saw a slight reduction year-on-year. The underlying position was broadly similar. But as reported at H1, the numbers include a GBP 23 million credit from the bond tender offer we completed in March and a one-off gain of GBP 12 million following the discontinuance of a U.S. dollar net investment hedge.

We currently expect net financial expense to be in the GBP 260 million to GBP 270 million range for 2026, reflecting the effect of refinancing existing low coupon debt in 2025 by higher rates in 2026 and, obviously, the new buybacks announced today.

On the next slide. Our tax rate came in at the lower end of our guidance range, and we expect the same range for 2026.

So if you take all of those lines together, this is giving 15.7% growth in AEPS for the year, more than double the rate of organic revenue growth. Over the last 4 years, we smoothed out some FX impacts along the way. That's a steady compound growth rate of 11.5%.

And as I said last year, we expected nonunderlying costs to come down in 2025, and they did. Integration costs fell by 41% as we came to the end of the formal Refinitiv process, and we expect them to come down again in 2026 as the other areas of restructuring continue to reduce.

Now turning to cash flow. This continues to be another highlight of the business model. We posted a record free cash flow of GBP 2.45 billion. As I'm sure you remember, we guided at, at least GBP 2.4 billion at constant rates. And we beat that at current rates, absorbing the current weakness of the dollar.

We are posting this very strong result despite a negative variation of working capital of GBP 400 million. There are three main reasons for that. First, a reduction of around GBP 90 million of the pay accrual for our SwapClear partners following the reduction of the revenue share; second, an GBP 80 million reduction in creditors related to the net treasury income, reflecting lower balances and interest rates; and third, we triggered around GBP 150 million of payments that we've made earlier than usual to suppliers to crystallize better procurement conditions before year-end.

Anyhow, going forward, typically a working capital outflow of GBP 100 million to GBP 150 million is a safe assumption to model. Given the ongoing buybacks, this 12% growth in free cash flow translates into 13.6% growth in free cash flow per share.

Turning now to capital allocation on the next slide. Against the GBP 2.4 billion of free cash flow, we deployed GBP 3.5 billion across shareholder returns and M&A activity. Total dividends were just over GBP 700 million, and we are proposing a final dividend of 103p today, up 15.7%, in line with our EPS growth.

We have deployed a net GBP 700 million on the Post Trade Solutions transaction that I already mentioned. And finally, we have had a record year for share buybacks with GBP 2.1 billion completed in the year. This demonstrates our very active approach to capital allocation and reflects our strong view of the deep value inherent in our own shares.

Even with this very active year, we ended 2025 with leverage at 1.8x net debt-to-EBITDA, still slightly below the midpoint of our target range.

So now let's look forward to 2026 and beyond. We are very well positioned as we enter 2026 with a record fourth quarter for gross sales in our subscription businesses and very healthy volume growth already at Tradeweb and our Post Trade businesses.

We are guiding to organic revenue growth of 6.5% to 7.5%, the same as in 2025, but importantly, with a steady acceleration in our subscription businesses as I have mentioned before. Within this, we also expect our D&A business to accelerate.

On margin, we expect 80 to 100 bps of improvement on a constant currency basis. So if you take the midpoint, 90 bps, you will find 60 bps to complete the 250 bps improvement that we committed to for '24, '25, '26 and an extra 30 bps, which comes from the further decrease in revenue surplus share terms at SwapClear.

For CapEx, the steady downward trajectory in intensity will continue, and we are targeting around 9.5% for 2026. And finally, we see this all translating into at least GBP 2.7 billion of free cash flow. And as I mentioned earlier, the tax guidance remains unchanged at 24% to 25%.

On this slide, I want to take a slightly longer-term view on how our cash generation and capital allocation has developed over the last 4 years and into 2026. The most important message here is how purposeful and consistent we have been in deploying capital to build a better business.

We have maintained high levels of capital intensity to invest organically in the business. We have grown the dividend strongly. We have done regular bolt-on M&A to strengthen our offering to customers. And then, when appropriate, we have returned surplus capital through share buybacks. This approach has supported strong top line growth, more innovation, improving margin and strong shareholder returns.

Our plan for 2026 continue that consistency. CapEx will be pretty consistent as an amount, but reducing to around 9.5% of revenue in terms of intensity. Free cash flow will grow strongly to at least GBP 2.7 billion. Dividends will continue to go up in line with earnings. And we remain active in our search for good M&A targets depending on fit and value.

And then today, we announced a further GBP 3 billion buyback over the next 12 months. So you can assume, given we have already done over GBP 400 million this year, that there will be a total of around GBP 3 billion in 2026, and then we will complete the new commitment in early 2027.

And finally, we are updating our medium-term guidance today, so I mean from 2027 to 2029, after several years of strong growth and margin delivery. On revenue, we are confident of mid- to high single-digit growth, including acceleration in our subscription businesses. So after the 6% reported in 2025, you can think of it at around 6.5% for 2026, heading to 7% for 2027, as I mentioned last year.

On EBITDA margin, we will carry on improving our productivity, and we are now guiding to a cumulative improvement of circa 150 bps over the period 2027 to 2029. We will drive this through continued strong revenue growth, investment in technology and other ongoing operational efficiencies, but while allowing room to reinvest in future sustained growth.

On CapEx, we expect intensity to come down to circa 8% in 2029. So think of that as the absolute CapEx figure staying relatively steady at GBP 900 million, GBP 950 million while revenue continues to grow.

And then finally, on cash flow, we are moving to a free cash flow per share metric, and we are guiding to double-digit compound annual growth in this important figure for the years to come.

And now I will hand over to David to take you through our strong strategic progress.

Thank you, MAP. Let's start with the obvious topic, AI. As we discussed at the Q3 results and the Innovation Forum, we're benefiting from our unique position at the forefront of AI-driven change, and we are excited about what that means for our customers, our people and our future growth.

You've seen these three pillars before: trusted data, transformative products and intelligent enterprise. Over the next few slides, I'll update you on how we're bringing this to life today for our customers and our organization. We have a great starting point, and everything we are doing is only making us stronger.

As a reminder, roughly 90% of our Data & Feeds revenues come from proprietary data and solutions. Our customers are using this data to power business-critical activities in highly regulated environments where accurate, timely, trusted data is nonnegotiable. It is often deeply embedded in transactional workflows.

Our breadth and depth are unmatched. Alongside proprietary data sources and exclusive licenses, we also have a network of more than 40,000 contributors proactively contributing data, continuing to enhance the value of our products through strong network effects.

The result is comprehensive industry standard data that we are constantly updating. That puts us in pole position to take share and drive growth as customers are able to interrogate and analyze more data at speed using AI.

As I said at the beginning of the call, our customers can see that our solutions are more valuable in an AI world. In the fourth quarter of last year, global investment banks and asset managers, all highly sophisticated institutions like Citi, Bank of America and Standard Chartered, signed GBP 1.9 billion of long-term data agreements with LSEG.

These organizations are securing their access to our data for up to 7 years invariably in contracts that step up in value over time. And they span a range of different segments: global investment banks, commercial banks, alternative investment firms.

We meet their needs and they are confident we will continue to do so across Workflows, Data & Feeds, Risk Intelligence and FTSE Russell. Only LSEG has this breadth of offering. It is a perfect demonstration of why these businesses are so valuable together.

The demand for and consumption of data is accelerating, and we are facilitating that growth. The history of data consumption growth is the history of technological advancement, the Internet, fiber networks, mobile, the cloud and now AI.

The chart on the left-hand side shows how the amount of data or messages coming through our real-time data feed continues to grow at pace, exceeding 15 million new data points a second in December. This represents a 4x increase in real-time data over our network in the past 10 years, a trend that we expect to continue.

With our direct connection to nearly 600 exchanges and venues, and our ongoing investment in technology and capacity, we are strengthening our market leadership. I often call this business the market infrastructure for all market infrastructure. It is live data delivered over our own infrastructure. AI does not and cannot replicate or replace this. If anything, it creates more demand for this data.

On the right, you can see the demand for our Tick History data, an evolving data set currently spanning 100 million instruments over 30 years. It's proprietary data that links today's price moves with those of the past. This is a critical point that people often don't get.

This data is valuable because it ties 30 years of market moves to the present day. And with hundreds of billions of new data points added each day, without constant updating, this Tick History becomes less and less useful.

We have the past, and we have the present. That is what creates the value. No one else has the past like us, and we are the leading provider of the present. Customers find this combination highly valuable with over 5 million customer requests a month.

I'll say it again, AI does not and cannot replicate or replace this. It will just drive more demand, customer demand for data that is accurate, up-to-date and comprehensive, verified and auditable is significant. That is where LSEG sets the standard.

And by making it easier for customers to access and consume this data through new cloud distribution channels or AI partnerships, we're likely to sell much more of it. And we are only at the beginning of that journey.

Increasingly, customers want to use our data in AI applications, opening up a new distribution channel. We're embracing that through our LSEG Everywhere strategy, delivering AI-ready data to any environment in which our customers want to work.

Since we last showed you this slide, we've added a new partnership with OpenAI, becoming the first financial data provider to enable customers to access their data through ChatGPT. You should expect us to enter into further partnerships in 2026 and beyond where there is customer demand and strategic logic.

These channels have only recently become available, and we are seeing very strong customer interest and engagement. Over 60 financial institutions have connected to our MCP servers directly or via one of our AI partners, connecting hundreds of users. And we have a strong pipeline of customers awaiting connection.

Many of these users are new users at existing customers, by which I mean the bank or asset manager already had an LSEG data license, but these particular teams or individuals were not users of our data, proof that our AI partnerships are increasing reach within existing customers.

Our AI partnerships are also expanding our distribution footprint, attracting new customers through the accessibility and ease of natural language. Already hundreds of prospective customers have attempted to access our data via our AI partners. Since no one can access our data without an LSEG license, this is creating valuable sales leads.

Once connected, customers are engaging with our data and content on an ongoing basis, driving rapid growth in data consumption through our AI partnerships. This is a great start to what we expect to be an important distribution channel for our data and also a natural mechanism for cross-selling.

As we make more of our data available via MCP, the user, whether that is a human, a model or an agent, will naturally discover the full breadth and depth of our data across Data & Analytics, Markets, FTSE Russell and Risk Intelligence. We're moving quickly down this path, investing in our AI-ready data and making more of it available through MCP connectors and multi-cloud environments.

We have a large pipeline of data coming to MCP, as you can see on the left. We're also supporting customers in their migration to cloud-based alternatives, and that is driving meaningful new sales and displacements.

Platforms like Databricks and Snowflake are helping us close big new contracts and drive increased sales of some of our most popular products like DataScope. And we keep investing in expanding the data we offer, whether that's in low latency feeds, ETF data or private markets.

Turning to the second pillar of our AI strategy, transformative products. The success of the migration to Workspace means our customers are now in a modern, modular, customizable platform where we enhance functionality week in and week out. That gives us a strong foundation from which to launch transformative AI-enabled products that bring speed, accuracy and conviction to customers' workflows.

As a reminder, 70% of Workflows revenue comes from trader licenses and activity. These users, humans today, maybe agents tomorrow, need real-time data, a network community and integration with a range of pre- and post-trade tools. This is regulated workflow with transactional features embedded.

And to address a question that comes up from time to time, what if the number of human traders is significantly reduced by AI, could that hurt our Workflows business? We don't see that happening. But also remember that over many years, our Workflows business has been moving away from a per seat model towards one focused more on data consumption or enterprise agreements.

And also if the scenario is that human traders are replaced by AI agents, then each agent will effectively be a licensed LSEG customer. In an AI world of agent-driven workflows, we will have more users consuming more data.

Workspace is getting better and better with hundreds of updates every year. To name a few recent enhancements, we extended trading capabilities through the expansion of Advanced Dealing. We streamlined banker workflows with the integration of DealWatch. And we enhanced our leadership in news with a dedicated app for Wall Street Journal and Dow Jones News.

This is driving real, measurable improvements in engagement. As you can see on the right-hand side, investment management and trading users are accessing roughly 25% more applications than a year ago.

Let's turn now to the Microsoft partnership. We made a lot of progress in 2025, and that pace of delivery continues to accelerate. On Workflows, to continue from the previous slide, our Teams-based collaboration tool, Open Directory, is live with accounts across 3 customer communities: FX, commodities and execution. And we have more than 50 accounts in our onboarding pipeline.

We're also piloting natural language functionality in Workspace interoperable with Teams and other Microsoft products. And Workspace Deep Research provides extensive AI-driven research and analysis, leveraging the full power of Workspace data. We expect to roll out both AI tools in the first half.

In Analytics, we've seen great traction and revenue growth since launching the API with over 50 customers adopting the platform. And just a few days ago, we launched Model-as-a-Service with Societe Generale as the launch partner distributing its own models through our API.

We're seeing great progress in Data-as-a-Service or DaaS. We are accelerating the migration of data into the new integrated architecture and expect to have almost all data sets onboarded by the end of the year. This is increasing our speed to market for new products and driving significant customer demand to access these data sets, whether via Fabric or other platforms like Snowflake and Databricks.

And last point, we have launched our Digital Markets Infrastructure powered by Microsoft Azure, another growth opportunity as tokenization takes off.

Turning now to the final pillar of our AI strategy, deploying AI across our own business, accelerating innovation and improving customer outcomes. I've mentioned before that we are resolving customer queries much more quickly and efficiently through our adoption of an AI-powered question-and-answer application. In December, we made that tool available directly to customers and has had significant traction already, and it will only get better.

Adoption of AI-powered workflows is also driving improvements in efficiency, quality and timeliness of data ingestion. We spoke about this at November's Innovation Forum, 9x faster content extraction, 52% reduction in data quality issues and 11% increase in productivity of our engineering teams. This all contributes to the ongoing margin expansion that MAP highlighted earlier.

I'm going to turn now to our Markets businesses. You've heard me say this before, but the whole premise of LSEG is this. In financial markets, data has become infrastructure. Access to data is just as essential as access to trading infrastructure. That's why these businesses belong together.

Electronification of markets, growth in data-driven decision-making and more sophisticated risk management are all blurring the lines between markets and data activities, deepening their interdependency. This is driving multiyear structural growth in our transactional businesses, delivering a 5-year CAGR of over 13%.

The Markets business delivered further strong growth in 2025 with double-digit growth in clearing revenues across interest rate swaps, FX, CDS and repos. Tradeweb also extended its leadership in trading of interest rate swaps, increasing its share by 180 basis points. Our FX venues saw their strongest volumes ever.

There's sometimes a misconception that growth across our Markets platforms just happens. Nothing could be further from the truth. The growth we're delivering today is the result of innovation and customer partnership going back years, often decades.

We build solutions that solve customer pain points and meet their critical needs, and we become deeply embedded in their core businesses. In that vein of innovation and customer partnership, we're innovating rapidly in digital markets, building the transaction and settlement infrastructure our customers will need as they increasingly adopt digital assets and tokenize traditional asset classes.

As you can see in the lower right quadrant of the slide, we are doing a lot in this space. But it is a big topic, so we will tell you more about it later in the year.

Another good example of our innovation and partnership in Markets is our success in the clearing of OTC products. The growth in this business over the last 15 years is extraordinary, a threefold increase in member banks, a 200-fold increase in clients and tenfold growth in notional value cleared each year to roughly $2,000 trillion.

We have become the global clearing destination of choice for interest rate swaps, FX and CDS. Now in partnership with 11 global banks, we're going after the opportunity in uncleared derivatives, which is roughly the same size as the cleared space.

Our members and clients want to manage their whole book in one place, bringing efficiency to their capital and margin requirements and materially simplifying and standardizing processes. We are uniquely placed to do that given the assets we have built and brought together under one roof, and we're entering 2026 with really good momentum.

Revenue in Post Trade Solutions is growing double digits, we're adding new customers and the network is expanding. We're driving strong growth and building platforms for the future across our business.

We've also integrated our products and platforms for our customers' benefit. This dynamic exists clearly in our data flywheel. The data we generate from our own markets infrastructure feeds into our D&A business. helping customers make better informed decisions when they trade, therefore, creating more data.

Second, Workspace is becoming the fully integrated workflow through which customers can access many of our services, not only for all D&A data but now also for FTSE Russell tools, FX trading, LCH data and, in the next few months, Tradeweb.

And we've established a powerful end-to-end ecosystem in FX, providing a front end in Workspace linking to the execution venues and straight through to our clearing business with FX hedging capability for Tradeweb and our data and benchmarking content adding incremental value along that trade life cycle. We have similar connectivity in swaps given the customer trust in the Tradeweb and SwapClear franchises.

I spoke earlier about the strong demand we've seen for our multiyear data access arrangements. Those integrate services from across our business, from Data & Feeds, Workflows, FTSE Russell, Risk Intelligence and Analytics. And they demonstrate the competitive advantage provided by our full-service business model.

As we've said before, big, sophisticated institutions want to do more with fewer partners. You can see that in the success of our LDA agreements.

Through our unique model, we've positioned our business to have deep moats and highly recurring revenues in areas of growth. Our diversification across products, customers and geographies gives our business model an attractive combination of growth and stability that performs well in environments like this.

Despite big swings in capital markets and the global economy in 2025, we continue to deliver strong and consistent growth, and we expect more of the same in 2026.

So to wrap up, we have achieved another year of very strong financial performance, driving continued top line growth through significant investment in our products and a consistent focus on partnership with our customers. LSEG Everywhere and other innovations like Open Directory, Post Trade Solutions and our Digital Settlement House are establishing platforms for future growth.

Through the transformation of our systems and the use of AI and other technologies across LSEG, we continue to deliver material operating leverage. And we are allocating capital in a thoughtful way to grow the business, drive innovation and return surplus capital to shareholders.

We're very excited about the opportunities ahead of us. With our leading trusted data, ongoing investment in product and the strength of our customer relationships, we are very well positioned for continued growth.

And with that, I'll pass to Peregrine for Q&A.

Thank you, David. Before we start the Q&A, can I please ask you to restrict yourself to one question. We plan to wrap up at about 11:30. Hopefully, we'll get through them all. But if we don't, please follow up directly with me or Chris later today. Thanks.

[Operator Instructions] And your first question comes from the line of Tom Mills from Jefferies.

Thanks for the helpful new disclosures, and that's my question. At a recent conference, the CEO of S&P said of the AI LLM platform, I'd say that our clients are getting additional value by being able to use our data in more ways, more ways they use it, the more value it creates and the better opportunity for value-based conversation at renewal. And we talk to those customers.

We've also seen really nice uptick in demand for add-ons and that's something that's helped with net new revenue. I think that ties in well with the content you provided on Slides 31, 32.

But I'd be curious to hear, you touched on the point about improving the opportunity for value-based discussions at renewal. And any uptick in demand for add-ons that you're seeing via the partnership so far?

Sure, Tom. Thanks. So for now, as you would expect, we are focused on adoption and just seeing the customers sign up and get access to this and seeing the usage grow. And that, as we mentioned on that Page 31, is growing very quickly, and we're really seeing a pretty significant and intense engagement there and, frankly, kind of day by day.

So I think, over time, the really significant opportunity here is in the context of consumption-based pricing and really charging the customers over time for usage. And for now, we're continuing to focus on our, I'll say, traditional subscription model.

But as we move over the course of the next year plus to more of a hybrid model, which is keeping the subscription, we think the subscription model is very attractive and very important, but incorporating into that the consumption-based pricing as well, that will be a very attractive way of capturing that kind of dynamic.

And I mentioned this earlier in my prepared remarks, but the fact that you have a combination of humans, models and agents consuming this data, it's, I think, pretty intuitive for you all to recognize that when an agent or a model is consuming the data, they tend to consume a lot more of that data than a human might.

And we've said in the past that humans barely scratch the surface of the amount of data that we have. So that's another angle here just in terms of as usage shifts to more AI-driven consumption, as we shift our model to more consumption-based pricing, we see that as a very, very attractive trajectory.

Maybe actually just...

Sorry, hold on a second.

Just one other point I want to add, and I touched on this earlier, but I think it also answers your question, kind of captures this dynamic, which is I've described the AI models combined with the MCP server as a very effective cross-selling machine. And the model is not asking for data from a particular data set.

The model is asking for answers to a question. And if that question can be answered by extracting data from multiple different data sets that we are making available through the MCP server, that is a great angle as well just for additional access, additional sales of additional data sets that the customer might not have originally known that we even had. So that's another aspect of this.

Now on to the next question. Thank you.

And your next question comes from the line of Hubert Lam of Bank of America.

I just got one of them. So how should we think about pricing and ability to keep your customers? Will we expect more competition in the future from MCP? I assume MCP makes it easier for users to switch between different data providers. So would it be harder to raise pricing in the future? And would there be more risk on bundling data contracts now that users have more choice, more flexibility as to who they want to consume with?

Hubert, so we see a very consistent pricing environment this year relative to last year. And I think it's about the quality of the data. If you think about the new AI channels and MCP as just another way of accessing the data, that's great for us. That doesn't mean that it is an environment where we're seeing incremental pressure on the pricing.

The quality of the data remains the same. The, in some cases, proprietary nature of the data means that no one else has access to it. And so we see this as a way of accessing more users within existing customers and accessing new customers as well. And as I mentioned, from a pricing perspective, we're seeing a very consistent dynamic this year as we have seen last year and the year before.

And your next question comes from the line of Arnaud Giblat of BNP Paribas.

So my question is on capital returns. So you've announced a GBP 3 billion buyback. That pushes up your leverage ratio perhaps towards the end of the year towards 2.0x, 2.1x net debt to EBITDA. So how should we read into this? Are you still -- I suppose you are leaving yourself the opportunity to step in and do further bolt-on acquisitions.

My question is just how are you seeing any potential dislocation in valuations in private markets? We've seen some significant shifts in public markets with data and software companies coming up quite a lot. Are we seeing the same thing in private markets? And perhaps does that create opportunities for you to step in, in the near term and add some more content inorganically to your platform?

Thanks, Arnaud. Maybe MAP will touch on the first part of your question. I'm happy to take the second part.

Yes, sure. On the buyback, you're absolutely right. We have coined GBP 3 billion in order to do two things. First, having a true increase into the return to our shareholders on the basis of the inherent value that we see in our share. Remember, 2 years ago, we did GBP 1 billion; 2025, GBP 2.1 billion. And here, we're talking about GBP 3 billion.

And by doing this GBP 3 billion, and you've made the calculation right, taking into account the dividend and the second part of the Post Trade Solutions, okay, altogether, this will bring us to 2x net debt-to-EBITDA by the end of 2026, so which will allow us to keep firepower for M&A that fit in terms of strategic alignment, obviously, and value.

And Arnaud, to your question about sort of the state of the markets. Yes, there's obviously been some dislocation. There's clearly some stress amongst some of the private equity holders out there. And you should expect us to always be evaluating opportunities.

And nothing to talk about near term, but as MAP mentioned, we are always evaluating opportunities that could make sense in terms of our both strategic fit and then attractive financial returns. And I think the buyback balances that appropriately in terms of an appropriate return to our shareholders while landing at that 2x net debt to EBITDA and maintaining the right kind of flexibility going forward.

[Operator Instructions] And your next question comes from the line of Enrico Bolzoni of JPMorgan.

I had one on EBITDA margin, please. So it looks like you're clearly doing more than what you initially thought. I remember from calls 1 year ago or so saying that, at some point, EBITDA margin would reach a ceiling because, clearly, there's a need to reinvest in the business. And here we are with a new set of targets that actually guides us towards further improvement.

So I was keen to hear your thoughts on whether you think this is just driven by the operating leverage and revenue accelerating, or you found more ways to cut cost. And perhaps, does this new target include any meaningful benefit from the deployment of AI within the organization?

Thanks, Enrico. I don't think we've ever said that we were planning to hit a ceiling, but I'll let MAP address that.

No, no, but I understand what Enrico is saying. So just a reminder for everybody, we committed ourselves in November 2023 of an increase of margin of 250 bps. 2026 is the third year of this plan. We are delivering the 250 bps. And on top of that, we have 130 coming from our Post Trade Solutions. So 380 that we will have delivered for the period '24 to '26.

Now what we've said is, going forward, because there was some question about what about after '26, that going forward, due to the operating leverage that the group has, I mean, building once and distributing many, obviously, this operating leverage, we can crystallize it into the margin or having a balance between the margin and reinvesting into future growth.

And what you see, Enrico, is 150 bps by 2029 is exactly that. It's the balance between operational efficiencies that we are harvesting, our natural operating leverage, so plus-plus, okay, and the investment we make into talent and technology for future growth.

And to answer the second part of your question, the answer is yes, you're right. We will crystallize in this 150 bps, you have indeed the financial consequences of what we do with AI within the company, particularly on our backbone and our -- and the ingestion of data.

Your next question comes from the line of Andrew Lowe of Citi.

I have one on Tradeweb, please. Would you be willing to give an indication of how much the Tradeweb-generated data sets account for your Data & Feeds revenues? And then whether any of those data sets are exclusively distributed by LSE?

Andrew, I don't think we have broken out and I don't think we intend to break out the amount of the Data & Feeds revenue that comes from Tradeweb. I can tell you that some of it is exclusive and some of it is nonexclusive. But I would also mention that, that is one of several different areas across the group, where we have very strong linkages between Tradeweb and the rest of LSEG.

We've talked in the past about the benefits both to Tradeweb and to FTSE Russell from the usage in FTSE Russell indices of Tradeweb pricing, and that flows both ways. We've used -- I'm not sure we've talked about this in the past, but Tradeweb has benefited from some of our middle and back office functionality in India and in other places.

We, of course, have the straight-through processing, if you will, from the Tradeweb swap execution facility into SwapClear. We've got the FX execution into Tradeweb. So a number of different areas.

And then maybe the last thing I should just touch on is that over the course of the next few months, we will be plugging Tradeweb access into Workspace, which is yet another significant opportunity that should be particularly attractive for Tradeweb users.

Your next question comes from the line of Ben Bathurst of RBC Capital Markets.

My question is on the new medium-term guidance where you're pointing to subscription business acceleration, which I think is like perimeter change versus the D&A revenue growth acceleration you've previously called out and are, in fact, restating again for FY '26.

I just wondered, could you elaborate a bit on the decision to make that change and perhaps make a comment on expectations for D&A growth contribution to that total subscription business acceleration you're talking about?

Sure. So I mean, the reason why we're looking at the subscription business altogether is mainly for 2 main reasons. The first one is it's the same subscription model, okay, which are governing the 3 divisions.

And the second, as it was presented in the slide, they are more and more intertwined. And we have true synergies in between the 3. LDA that David was mentioning at the beginning of the call is the obvious example.

So now on the medium-term guidance and for the subscription business, I hope you got it from my remarks. What we expect in there is we posted 6% in '25, circa 6.5% in '26, going to 7% in 2027. And on this, obviously, D&A will be accelerating, too. I mean, just to be clear, due to the size of it, it's the main lever for this acceleration, for sure.

And if I may just add one more thing, which is you see this slide, I don't remember it was 31 or 32, with this adoption of MCP. So you see that it's extremely strong and we are concentrating of usage. So for sure, AI can be an accelerator of this trajectory that I just mentioned. You see what I mean.

But I mean, it is still the early days. We just switched on the MCP just before Christmas. So you see it's not a long time ago. So it's a bit early to size it. But for sure, it's in the plus category, if you want.

Your next question comes from the line of Julian Dobtovolschi of ABN AMRO.

You've mentioned that a large portion of your data sets are already available now via the LLMs such as Anthropic, Databricks and OpenAI and a bunch of others. I was just curious to know, what percentage of LSEG's total data universe will ultimately be available through the AI-native channels? And if there is a view to keep some of this fully in-house for various reasons?

So I would expect that we are going to be making, and we've got a slide in here that touches on this, I would expect that we're going to be making as much of our data as possible available through these distribution channels and through MCP.

And just to be really clear, the implication of your question is that we might keep some away to somehow protect it. But again, to be really clear, providing access to a model through MCP does not mean that the model then can get that data and never need it again.

And so we can provide access through to MCP to a model and continue to protect and maintain the value and the integrity and the proprietary nature of that data. This seems to be kind of a common misunderstanding that people have.

So we view this as a great channel to distribute our data, whether that's proprietary data, whether that's a linkage of multiple different data sets. And the fact that we are making it available through MCP, think of it as a very structured, disciplined gateway, and we can actually put our usage meter on top of that as well.

So again, I understand your question, but I just want to make sure that I'm clarifying. There should be no misinterpretation of making data available to a model through MCP as somehow vitiating the value of that data or the proprietary nature of that data.

Your next question comes from the line of Benjamin Goy of Deutsche Bank.

One question, please, on your LSEG data access agreements. You mentioned almost GBP 2 billion signed in Q4. But can you give a bit more qualitative color on these agreements, whether it was Q4 or more recently signed? Do you see any change in customer dynamics? Do you see put options or breakup clauses in those contracts now or basically same contracts as you had a year or 2 years ago?

Yes. No sort of structural changes in these. We've talked in the past about how they can take a couple of years to put in place because of the way that we and our customers set them up. It takes some real top-down focus in organization and coordination and planning.

But no, we view this as an increasing recognition by our big important customers of the value of the integrated offering that we are providing. They do have a line of sight not only into what we are providing today, but what we are building for them in the next couple of months and in the next couple of years. They are multiyear in nature.

And I believe the ones that we have announced most recently tend to be out to 7 years. They all have extensions built into them as well. So I think it's just what you see is what you're getting here in terms of our customers really understanding the quality of our offerings and wanting to commit to that for many, many years to come.

And I think just it's worth reiterating this. I understand some people might have had a little trouble hearing at the very beginning of the call. We have the most sophisticated financial institutions on the planet who have very rigorous risk management processes, very rigorous analysis of what their technology needs are, very clear understanding of their requirements.

And they, after extensive work -- and I said, in some cases, these take up to 2 years. After extensive work, they are making decisions to, I used this phrase earlier, they're voting with their wallets to commit to consuming our data through our channels for the next, in many cases, up to 7 years.

And so we think that is a pretty clear indication that these highly sophisticated institutions recognize the value of the content, the data, the workflow that we provide and recognize that, that is increasing in an AI world as opposed to decreasing.

Your next question is from the line of Oliver Carruthers of Goldman Sachs.

Oliver Carruthers from Goldman Sachs. Thanks for the very detailed presentation and the incremental disclosure. Very helpful. I think Slide 31 is really interesting around the growth in customers you're highlighting. In terms of those customers connecting to your MCP server, so that 67 number, I appreciate it's moving a lot, but can you give us a flavor of the types of institutions, investment banks, hedge funds, asset managers, who is using this? And any steer on the use cases would be really, really helpful.

Sure. So it's lots of different kinds of institutions. Typically, we see smaller institutions moving more quickly. But in this case, we're seeing smaller institutions and large institutions.

I'll give you one example. There's one very large institution that is using this service to evaluate, I'm not going to go into specific names, but to evaluate one AI functionality against another AI functionality. And the constant they are using is our data because they know the quality of our data and they know what to expect from us.

So they are using us as the baseline and they are using that to make a decision as to which of the AI distribution channels they want to actually use. But that's just one example.

And Oliver, as you mentioned, this is changing literally day by day. And we're kind of getting a running commentary from the team on how this is growing and how we're seeing increasing and expanding desire to access through this as well as incremental sales leads.

As a very quick follow-up. You still own these customer relationships, even when it's not your own MCP, but even when it's a third party? This is a query tool they come to you, but you still own these customer relationships. Is that the correct way to think about it?

Yes, it is. Thank you for asking that question. Let me be really, really clear about this. The way this works is that if you have a license with LSEG, you can then turn on access to LSEG via, for example, Claude or via ChatGPT. There's a little connector button when you pull up a certain window in these.

And you have to flick that on to get access to LSEG data. You can only do that if you have a license with LSEG directly. And therefore, we maintain the ownership of the customer relationship we're contracting with the customers.

Now when we talk on that Page 31 about over 300 prospective users, what we mean by that is that there are a number of prospective users who are using these channels to try to get access to our data. They're effectively knocking on our door through MCP. And they don't have an existing license.

But the way this is designed is that we are informed of their interest. And so it's a great origination channel, it's a great sales channel for us. We then take those leads. Our sales team directly receives those leads and we follow up with those customers. Does that help?

Yes. Very helpful.

Your next question is from the line of Marina Massuti of Morgan Stanley.

I have a question on the AI adoption given some of your peers have given numbers around the efficiency opportunity from internal AI implementation. Can you also provide a bit more color or be a bit more specific on how much of the current and future AI deployments contributes towards the 150 basis points margin expansion targeted in the medium-term guidance?

Yes. Thanks, Marina. So we haven't put any specific guidance out there in terms of the efficiencies that we're seeing from AI. I did mention in my remarks, and on Page 36 you can see some of the stats, we are seeing meaningful improvement in productivity, in efficiency. We're seeing this in customer service. And then we are also seeing improved efficiency in terms of our engineers and our software development.

We've seen up to this point, and this number is going up pretty regularly, but we've seen at this point, I think I can comfortably say, 11% efficiency in our engineers. So I would bake that into the numbers that MAP was referring to earlier in terms of continuing margin improvement in the business. But we haven't given anything specific around that.

Your next question is from the line of Michael Werner of UBS.

Thank you for the long-term targets in particular. A question on LDAs. I was just wondering with regards to, a, the step-ups that you mentioned in terms of pricing, are they contingent upon certain deliverables? And ultimately, are these step-ups typically higher than what you see in kind of the base rate?

And then also, how does MCP servers fit into those enterprise agreements? Is that already included? Or is that potential upside from a revenue generation or a client wallet share perspective going forward?

Thanks, Michael. So every LDA is a little bit different. And some of the step-ups are a little bit higher than what the regular price rises would be. Some of the step-ups might be a little bit lower. They are all typically in the same general range unless there are, for example, commitments that we have made to add a specific new product or new capability.

Or sometimes in these LDAs, the customer may be locked into another competitor product for a year or 2, and it takes them a year or 2 to get out of those and migrate on to ours. And there may be a step-up associated with that kind of migration. So everyone is a little bit different, but that gives you a sense of some of the different variables.

To your second question, there is a defined perimeter around the LDA agreements in terms of effectively focusing on existing product, and it's well defined perimeter. And in the context of these MCP capabilities and this distribution and AI model consumption, that is, I think I can say with certainly, not included in the data access agreements that we have struck at this point.

Yes, yes. For none of them.

Yes. So that is all incremental usage that's coming through these channels, that is upside.

And this concludes questions on the conference line. I will now hand the presentation back to David Schwimmer, Chief Executive Officer, for closing remarks.

Well, thank you all for your questions. Thanks for spending time with us this morning. I know it's a little bit longer than usual in terms of the presentation. We did feel there was a lot to get through. And to the extent you have any additional questions, please do not hesitate to get in touch with Peregrine or Chris. Thanks again.

Good afternoon, everyone. Thank you for joining us. Great to have you here all in person, and also thank you to those of you who are joining us online. It is great to see such a big turnout. And we are really excited to show you a selection of the many innovations we have developed for our customers at today's innovation forum. We have made huge progress over the last 5 years. And our goal for today is to show you some of the innovation, the transformation, the disruption that we are driving through all of LSEG.

So let me briefly take you through the plan for the afternoon. MAP and I will recap the group strategy and some of the powerful drivers of our business as well as the execution and transformation that we have delivered to date. Then we will hand over to Irfan, our CIO; and Emily, our Head of AI, who will talk in more detail about our engineering transformation and our AI strategy. Next up will be Ron and Gianluca. They will update you on the DNA strategy, progress with Microsoft, the product road map and most importantly, monetization. They will then tee up the DNA product demos, which will all be here in the theater. We will then break you into 5 groups and rotate through presentations and demos of a number of other great products, and we'll cover the logistics of that later. And then finally, we'll be back in here for Q&A with all of the presenters, and we will finish off with some drinks.

So first, let's recap on what LSEG is and why these businesses are so valuable together from a strategic, commercial and financial perspective. All of our businesses have strong competitive positions, typically top 3 in the markets they serve and often #1. And our services perform nondiscretionary functions for our clients, i.e., not nice to have. Over the years, through investment and M&A, we have aligned the group to multiple structural growth drivers. We work with our customers very differently from how our competitors typically do. We have a partnership model based on an open ecosystem. We build products not just for our customers, but with our customers. We often become their strategic partners. And with their core businesses deeply reliant on our services and products, a high level of trust is critical to our customer relationships. Strong businesses aligned with structural growth tailwinds with deep customer partnerships. This all translates into a really strong economic model with all-weather growth and very strong cash generation.

Our markets continue to offer very attractive growth prospects, from mid-single digits up to double digits for FTSE Russell and Risk Intelligence. And we have an outstanding portfolio of assets within these markets. Businesses like real-time data, SwapClear and Tradeweb are undisputed scaled leaders in their fields with long track records of investment, innovation and growth.

World-Check is the global leader in the high-growth sector of screening and compliance. FTSE Russell, Workspace and our non-real-time data, all have strong top 3 positions in their markets, and we are investing in all 3 to build new services for our customers and to grow share.

What I like about our positioning is that we are a top player in each of our businesses. But our growth is not constrained by a high market share. So our markets offer growth and we have room to take share as well. So we have several world-class businesses across our portfolio. They are each great trophy assets on their own, but they become even more valuable as part of an integrated LSEG. We are increasingly linking these products and services closer together for our customers' benefit. This is most evident in our data flywheel, the data we generate from our own markets infrastructure feeds into our DNA business. That data helps customers make better informed decisions as they trade more, creating yet more data through their trading and risk management activity.

Second, Workspace is increasingly becoming the fully integrated workflow through which customers can access many of our services, not only for all DNA data, but now also for FTSE Russell tools, FX Trading, LCH data, and in the near future, Tradeweb. I've spoken before about how we have integrated our FX platforms throughout the group with Workspace, Tradeweb and our clearing business, all underpinned by industry-leading FX data and analytics. This creates an end-to-end proposition. We have the same comprehensive offering in swaps, drawing on our SwapClear and Tradeweb franchises. And as you know, we're powering a number of FTSE Russell fixed income indices with Tradeweb data. This creates another flywheel effect. The more volume traded on Tradeweb, the better the pricing and the FTSE indices. The more usage of the indices, increases the importance of the Tradeweb pricing as the industry standard.

These are some of the product benefits, but there are commercial benefits, too. We have become an important strategic partner to many of our customers, and our long-term contracts are reflecting that. I'll cover these enterprise deals in more detail in a moment.

We have aligned LSEG with a number of very strong and long-term industry trends. The growing demand for data in decision-making is not new, but AI is driving that to new heights. And not just any data, data that is trusted to be accurate and specialized for our customers' use cases. That data is at a premium, and that is our forte.

Electronification and digitization also continue at pace. And through Tradeweb and our digital markets infrastructure, we are at the forefront of that trend. And whether through FTSE Russell, Risk Intelligence, DNA or our post-trade businesses, both cleared and uncleared, we support customers as they navigate ever-changing regulation.

Our diversification is yet another strength. Unlike many other companies we are compared to, we are not disproportionately exposed to a single asset class, geography, customer type or product. We serve customers across the sell side, advisory, buy side, corporate and academia and across a broad spread of asset classes. We're also open in our distribution and always have been. This is something you will hear much more about today. We are just as comfortable serving customers directly with our own front end or working in partnership to provide our content through other channels.

And the final point on why LSEG is so differentiated from a strategic standpoint. It's the unmatched breadth of our offering across the whole trade life cycle and through the whole data value chain. This gives us a unique position from which to serve our customers as strategic partners, not just as data vendors.

And now I'll hand over to MAP to talk about how this all translates into our economic model.

Thanks, David, and good afternoon, everyone. I think of our economic model as the best of both worlds. Nearly 3/4 of our revenue is from recurring subscription services. And in many cases, these services are relied on by our customers and embedded in their processes. They are critical and high value. The other 25% comes from transactional revenue. However, most of these, particularly Tradeweb and Post Trade have structural growth drivers behind them. It means that they are, of course, cyclical to some degrees, but much less than other exchange-type businesses. And you can see that from the 14% compound growth achieved over the last 4 years. And this has translated overall into a very stable top line performance, whatever the weather. You can see here that whether interest rates are up or down, GDP is stronger or weaker, equity markets are rising or falling. We can deliver mid- to high single-digit organic growth. And as David mentioned just now, we are not exposed heavily to any single asset class or sector, which means that we have natural offset throughout the business.

So let's take a look at how the model has delivered over the long term. And of course, I certainly can't take credit for all of this, but I'm confident that we will continue the trend. So earnings per share has compounded at 15% over the last 20 years and dividends per share at 18%. To give you some context, this is the best compound dividend growth and second best earnings growth of the top 20 FTSE 100 companies. We have guided over GBP 2.4 billion of cash this year, 60% higher than 3 years ago. This cash generation has enabled us to fund further M&A, just like the Post Trade deal we just announced a couple of weeks ago, and returned cash to shareholders. By February next year, we will have returned GBP 5 billion via buybacks in 3.5 years. So it's roughly 10% of our market capitalization.

For the next section, we are going to focus on our delivery. Back to you, David.

Thank you, MAP. So let me take you back to the Refinitiv transaction. This is old ground for many of you, I'm sure, but many others of you are newer to our story.

Five years ago, LSEG was a regional, mainly equities-focused, mainly transactional business. Although the long-term track record that MAP just showed you was outstanding, LSEG was subscale, overly exposed to Europe and had limited data capabilities. Refinitiv was a global business with a high proportion of subscription revenue and very strong competitive positions, but with a number of assets that required significant investment. Growth had been anemic for many years with steady market share losses in a healthily growing industry. We knew it was a fixer upper, and that was reflected in the multiple we paid around 11x EV to EBITDA. But the size of the prize was significant. We saw the strategic value in the combination, the creation of a unique group, which unites the full trade life cycle with the full data value chain, each enhancing each other, as I laid out a few moments ago. If it hadn't been for this and the successful integration which followed, we would not feel so confident about the continued growth in front of us.

For the last 5 years, we have been on an ambitious journey to transform the combined business. The first 3 years or so focused on integration. More recently, we have pivoted to transformation of our people, our platform and our product.

MAP will take you through that integration, and I will pick up on the transformation.

Thanks, David. It's fair to say that expectations were low off the back of the Refinitiv deal, mainly because of the perception in the market of the quality of the business. But LSEG has, in fact, delivered in every regard on this transaction.

First, Growth. We set a growth guidance of 5% to 7% for the first 3 years. And investor were skeptical that we could even reach the bottom of that guidance, given the decades of underinvestment at Refinitiv. But in fact, as you can see, growth has exceeded 6% in each of the last 4 years.

Second, Eikon. Many of our investors were unhappy users of the platform and couldn't see a future in it. But with the investment in Workspace and a more resilient back end and improved account management, we have taken a business from many years of revenue declines to 4 years of growth.

Third, could we really achieve the synergy targets, given the size of the acquisition and the task required? And here, we have performed very strongly. Upon announcement, we had initially targeted revenue synergies of GBP 225 million. By the end of 2024, we were at a run rate of GBP 292 million. Similarly, on cost synergies, we exited 2024, running at GBP 562 million against an original target of GBP 350 million. And in total, as you know, we've spent GBP 1.4 billion on achieving these synergies as we expected and as it was reflected in the purchase price.

Now point four, Margin. The Refinitiv businesses had a lower margin than the industry benchmark due to legacy system and operational complexity. At first, we improved margins through the integration cost synergies, net of growth reinvestment. This brought a net 90 bps of margin between 2020 and 2023. Since 2024, we've shifted from integration to transformation. Doing so, we have greatly improved the group operating leverage through the implementation of a holistic and disciplined cost control and investment allocation. As a result, our reported margin jumped by 220 bps in the last 2 years, out of which 180 are underlying and 40 is FX. So this is positioning us very well to reach our underlying target of 250 bps for the period '24 to '26. And remember, we have a further 100 bps on top of that from the recent Post Trade transaction.

Finally, our leverage. We took on GBP 13 billion of additional debt, and our leverage immediately post the Refinitiv deal was 3.3x net debt to EBITDA. But through strong cash generation, reduced capital intensity, a couple of disposal and a disciplined capital allocation, we have reduced our leverage below 2x net debt to EBITDA within 12 months, well before the 24 to 30 months we committed to. We expect to be at around 1.9x at the end of this year, which is right in the middle of our guided range. So I know it's a lot of detail here, but it's important to remind you of the journey we've been on through these 2 phases of first integration, '21 to '23, and then transformation on '24 onwards.

Talking about that, David, do you want to talk more deeply about transformation?

Thank you, MAP. So some of you have asked me over the past few years about changes to our leadership. And what has been driving that? It's true. The team has changed a lot. But to be clear, this is a feature, not a bug in the system. To drive this kind of transformation in many areas, we needed different leadership. We now have a very strong team that has the right capabilities to execute on the next leg of the journey. We have the benefit of strong continuity in markets under Dan Maguire, who has led LCH with such a clear long-term vision and partnership mindset. I see the same continuity, vision and partnership in our risk and legal and compliance functions under Balbir Bakhshi and Catherine Johnson.

In engineering, operations, finance, people and corporate affairs, our leaders are driving significant transformation towards a more capable, agile and efficient organization with a high pace of change supported by deep collaboration. And then across our subscription businesses, we are seeing the benefits of bringing in industry and product experts like Gianluca coming from S&P to co-lead D&A. This is also true below the ExCo level with the likes of Todd Hartmann, joining from FactSet to run data and feeds; David Wilson and Fiona Bassett coming to us as seasoned industry leaders; and Emily Prince, becoming Head of AI for the group. Many of these leaders have made significant changes across their own teams.

Looking at what we call our group leaders, the top 90 or so direct reports of my direct reports, over 1/3 have joined in the last 3 years. They bring new capabilities and enterprise leadership to balance the continuity of the wider population. Looking specifically at engineering, at least 10 of our most senior executives have joined in the last 18 months. But change has, by no means, been limited to LSEG's leadership. We have transformed the way that we work across the organization. Irfan will shortly talk you through the engineering transformation as we build a high-quality, deeply technical workforce where core capabilities are in-sourced, enabling our product ambitions. Pascal, Irfan and MAP are leading our shift to a product-led operating model. Ron stood here 2 years ago talking through the transformation that we have driven through sales and account management, through training, incentives and specialization. And MAP and Pascal have made significant progress in developing lean and scalable enabling functions across the group.

The second P of our transformation is platform, in particular, building modern and scalable infrastructure. I'm not going to go through these in detail, but I have listed here some of the more significant programs that we have been delivering across LSEG. Any one of these on their own would have been a major undertaking and the work goes on. Enterprise resource planning and billing will take another couple of years as well our migration of data and applications to Azure.

Looking at platform through another lens, we have created a strategic platform, which has transformed how we engage with our customers. We have gone from being too sizable, but not always top-tier providers to a single critical partner across data and markets infrastructure. And with our biggest customers, we work in partnership on their strategic road maps and how our products can help them deliver.

Commercially, this comes to life via LSEG Data Access Agreements or LDAs. You have heard us talk about these a lot over the last couple of years. The breadth and depth of our data and workflow offering allow us to grow our share of wallet while reducing total cost of ownership for our customers. It also gives economic certainty over the long term for both parties. The results have been very good. Customers are showing increasing satisfaction with LSEG and account growth comfortably outperforms original terms as we introduce additional products and services. As you can see from the chart on the right, if we complete negotiations on all new LDAs currently under discussion, they will represent around 17% of D&A ASV as we exit this year.

Turning to our product transformation. Here, I've outlined some of our larger scale investments, the areas that we have dedicated the most capital to as we enhance services to customers, starting with Workspace. That has been a double transformation. Not only have we migrated our customers off Eikon and onto Workspace, we have also built a modern customizable and modular platform on which we are adding enhancements at the rate of 2 per day. This platform enables us to roll out new innovations as they become available. And that ties into the Microsoft partnership. Almost 3 years ago, we entered into a long-term agreement to work together on product and cloud. Ron and Gianluca as well as Matt Kerner from Microsoft will cover this in more detail later. But I am really happy with the progress that we are making for our customers.

Just a couple of weeks ago, we announced a new partnership for Post Trade Solutions with 11 leading banks. This is the culmination of several years of strategic planning and gives us a great platform for further growth in partnership with our major customers. And then there is the ongoing growth and innovation at Tradeweb. With consistent execution and new product development and deep understanding of customer needs, Billy, Sara and the Tradeweb team have continued to drive exceptional growth, enhanced where it makes sense by acquisitions. Most recently, ICD has given us access to a whole new asset class and customer group.

Our investment in new product has by no means been limited to these bigger builds. As you can see from this slide, we have been innovating across the board. Every division has launched significant new product in the last 12 months, and we plan to continue in the same vein. Maybe just a few things to call out. We have fully replatformed our trade routing network, Autex, in Azure with Autex now connecting 1,600 brokers and asset managers via the cloud. As a result, it's faster, has much greater capacity and is even more resilient. We have executed the first transaction on our digital markets infrastructure, which is positioned to become an important new capability for trading and settlement. And in Risk Intelligence, we have launched World-Check on demand, with all of our critical data and insight now updated in real time.

So what's next? Where do we go from here? LSEG has changed beyond all recognition in the last 20, 10 and even 5 years, and it will continue to do so as technologies evolve, regulation changes and customers encounter new problems to solve. What we are building and what we will show you today is a company that is disrupting itself for customers. Through the presentations, demos and case studies this afternoon, we will show you how we are transforming through technology, executing a bold AI strategy, advancing our leading data and analytics franchise and accelerating innovation across all of LSEG.

We are focused on delivery, and you will see that today as we bring our commitments on the left here to life in our products. You'll see our data in numerous different environments where different customers work, LSEG Everywhere. You'll see the richness of functionality in Workspace and how we are enhancing it with AI and collaboration. You'll see solutions for real customer pain points that only LSEG can deliver, solutions that drive capital efficiency, manage risk and reduce cost. And all of these will be through the lens of making our demos as real as possible. These are not glossy marketing productions or vaporware, but real products with real use cases step through at a pace where you can follow the workflow.

And first, you will hear from Irfan and Emily, who will demonstrate the progress that we are making on delivering all of this through our engineering transformation and AI strategy. So over to you.

Thank you, David. Hello, everyone. Emily and I will start with brief introductions, and then we'll talk about engineering transformation at LSEG and our AI strategy. I joined LSEG as CIO in January last year. Prior to that, I was at Goldman Sachs for 28 years, where I worked in most of their businesses, starting from FIC to equities, to asset management, to wealth, to consumer. So from working on exotic derivatives, real-time trading, big data analytics, multi-asset portfolio construction and 24/7 credit card transactions, I had the opportunity to learn and lead various engineering domains across finance.

And I'm Emily Prince, Head of AI, LSEG. I've been working at LSEG for 9 years, working as the Head of Analytics. I joined LSEG from BlackRock. And prior to that, I spent 9 years in various quantitative analytics roles, including structuring, portfolio modeling, research across Barclays, Lehman, UniCredit and RBS. I'm also a member of the Bank of England's AI Consortium.

So as I mentioned, I'll first walk you through our engineering strategy and how it's helping us transform the way we build products, then Emily and I will cover AI.

We took a first principles approach to our engineering strategy, focusing on the foundational problems we need to solve in order to accelerate product development and manage our costs and risks better. We ask ourselves, what are the key ingredients to building a world-class product organization, which allows us to continuously capitalize on latest advancements in technology? How do we build a durable and efficient factory to create new products faster, cheaper and with appropriate controls?

We have 3 pillars of this strategy, and they're in line with what David just talked about: having exceptional talent, common platforms and product discipline. Now these pillars may sound very obvious and basic, but they are not. They are foundational and some of the hardest aspects of building a world-class organization. You may also notice that these pillars are technology-agnostic, regardless of whether it's AI, quantum, digital assets, cloud or whatever the latest and greatest is, these are the key building blocks. And as we master these, we can play both offense and defense with any technology by accelerating our product development.

So the talent or people piece is essential. LSEG ultimately serves its customers and build this product by shipping software. We are a fintech firm, and you cannot build amazing products without the best engineer. That's pillar #1. But talent alone is not sufficient. You will take the best engineers and convert them into mediocre performers if we don't give them the tools and platforms to be efficient. That's the second pillar. But if you only do first 2, all you get is speed, meaning you will be able to move -- ship software quickly. But speed alone is not enough.

We don't want to just move faster, we also want to move in the right direction and build the right products for our customers. In other words, what we're really after is speed and direction, which the tech firms would typically refer to as velocity. This is where our third pillar product discipline comes in that sets the direction and is helping LSEG become a product-led firm.

Rather than getting into the weeds of every single pillar, let me give you some concrete examples and metrics to bring these to life, where we were, where we are and where we're going. Last January, we had 17,000 engineers in the firm and only 40% of them were contract -- only 40% of them were employees. The rest were contractors. In general, firms don't get the best engineering talent when they go the contractor route and you can't build the best product with the outsourced staff. Fast forward to today, we currently have about 14,000 engineers with 58% of them being internal engineers, that's an 18% increase. Our goal is to get to 80% by the end of 2027. We didn't just shift these numbers blindly. We shifted them with a clear goal of raising the bar on excellence. We introduced new engineering principles to guide all of our actions. We significantly improved our hiring standards and implemented an independent bar raising protocol to ensure we're consistently hiring the best people.

Prior to this year, if you were an amazing engineer, you had to become a manager to progress your career. And as you know, not all engineers want to manage people. We didn't want to take our top quartile engineers and convert them into bottom quartile managers. Now we have individual contributor tracks where you can grow to have the most senior title in the firm without managing a single soul. And we announced the first batch of our distinguished engineers late last year to recognize the best of our technical talent. So what does it all mean? What's the upshot? Why am I talking about it? In the end, it is about productivity. Our productivity is up 11%, while our head count is down 18%. In other words, 14,000 engineers are producing 11% more output than what 17,000 did in January last year. Our hypothesis that fewer higher-caliber people will produce more output is proving to be true.

Regarding the second pillar, our engineers used to have a lot of friction when they build products. We had 8 different source code repositories, limited automated code pipelines and no common credentials, artifacts or logging systems. Fast forward to today, 96% of our code is now in a single source code repository with common platforms.

Our engineers are actively using AI to build products, and they're seeing up to 34% increase in productivity. And we're not just using AI to do code completion, we're using it to write new apps from scratch, perform cloud migration, upgrade legacy systems and automate test. We have also deployed common cloud platforms to operate across all 3 of the major cloud providers, allowing us to automate, software development and more importantly, automatically enforce cyber and other control policies. So what's the punchline for this pillar? We're seeing up to 25% increase in release velocity while our incidents or outages are down by 55%. Why is that important? It's important because there's a risk that more software changes can mean more instability. These are important metrics that we track. We want to, of course, move fast to serve our customers and our -- but the same customers and our regulators demand the highest level of resiliency and quality, and that's a key part of our product offering.

On Pillar 3, I know David, MAP and Peregrine have talked to you about our journey to become product-led. This involves significant cultural, people and process changes. We're going product by product, team by team and ensuring that we have the right people and the right processes in place to improve our offerings. This means having a dedicated team of product managers, engineers and ops people to own the totality of customer experience regardless of how many teams are involved in delivering the ultimate product. We're driving our decisions with data and ensuring that we're upgrading and attracting the best talent.

To recap, these 3 are the foundational pillar of our strategy, allowing us to have the velocity and quality needed to build the right products for our customers while managing our costs and risks better. They're laying the foundations for us to leverage AI and other technologies so that we can serve our customers better. Without these pillars, it would have been much slower and more expensive for us to incorporate AI in our products.

And speaking of AI, I will now hand over to Emily to kick us off on our AI strategy.

Thanks, Irfan. Now whatever you think about artificial intelligence, whether you're an evangelist or a skeptic, what is remarkable is the way it's allowing us to consider new approaches to solving old problems. At LSEG, our global reach and diverse vast data sets, together with decades of experience in data and analytics, we see AI as a powerful opportunity. We've synthesized LSEG's AI strategy into 3 pillars: trusted data, transformative products and intelligent enterprise.

Let's start with our first pillar, trusted data. You've known LSEG as that trusted provider of content across financial services for a long time. We have reinforced that commitment and now we made our data AI-ready. Data is the basis of AI. And to achieve trust in AI, you must first have trusted data. While the first pillar focuses on the importance of our core trusted content, our second pillar, transformative product, is focused on applying AI to the products we build for our customers. We are in the age of product enablement. And with a single question in a customer's preferred language, we can not only discover new insights, but orchestrate entire new workflows.

With just this data as the basis of AI, knowledge is the basis of transformative products. LSEG is using its depth of market expertise to reimagine how financial service professionals work with speed, simplicity and conviction, which you'll see in some of the demos later today.

And finally, our third pillar, intelligent enterprise. Achieving success in AI starts with our people. It increasingly shapes the velocity with which we can build products, evaluate risks, respond to customer questions with consistency and transform unstructured disparate data into structured insights. Let's spend more time on trusted data.

The depth, breadth and diversity of LSEG's data is hard for the human brain to comprehend. But for a model, it's a game changer. Why is it models care so much about data and especially the 33-plus petabytes that LSEG has? Well, models, of which there are now thousands, are generally trained on publicly sourced data. For models to differentiate, they need differentiated and deep data. When presented with trusted data through the likes of LSEG's MCP server, models can identify relationships and data that generate new insights for end users. Combining LSEG's extraordinary breadth, history and subject matter expertise in areas such as value to pricing, together with powerful AI models, allows LSEG's customers to benefit from unparalleled insights.

Now on this slide, which you heard David discuss as part of our recent results, we point to the level of differentiation we have in LSEG's data. And while I won't step through every number to, I do want to draw your attention to our 90% and 45%. 90% of the Data & Feeds revenue is based on proprietary data, which the LLMs don't have access to publicly train from; 45% represents a proportion of our Data & Feeds revenues, which are real time. Built on a global private network, this is a private content set, not available to AI models and are highly desired for use by our customers in AI products such as agents.

LSEG's trusted data is differentiated and highly valuable in the context of AI. We have an extraordinary mix of proprietary, nonreplicable, historical data brought together with LSEG-defined standards and followed by customers across the globe. Now every day, our trusted data is underpinning decisions across financial services' ecosystem from traders and the largest banks to quants building signals and risk analysts responding to changing market conditions. To achieve this data standard, LSEG's content undergoes significant care to achieve the quality which we are happy with. Our process starts with sourcing and has done for decades. It includes over 40,000 contributors, and of course, our own proprietary data generation. Then there is our data quality, which involves deep, iterative cleansing until it meets our standards.

Coming now to normalizing. The step that means our customers can use the breadth of our data out of the box. This step together with the application of mastering is a hugely important one, and requires a deep level of expertise. Later today, Adam and Tim will go into this in further detail and also share some demos.

Now this brings us to concordance and tagging. This is an enrichment step, which broadens the usability of LSEG's data and represents a very important part of what ensures LSEG's data is AI ready. And finally, distribution. This is not as simple as depositing data in a client environment. LSEG is ensuring consistent delivery of data, where and how our customers need it. Through Databricks Delta Share, Workspace, APIs, Microsoft or Google BigQuery, LSEG is everywhere, and we're meeting our customers in their preferred infrastructure.

LSEG delivers the highest standard and trusted data from source to insight through unrivaled quality, concordance and intelligent distribution.

Now having spent some time on the importance of trusted data, I'd like to spend a few minutes now on what makes our data AI-ready. Building from the quality we enabled as part of trusted data, we're layering this with control, including data rights management and accessibility. Our focus on accessibility with semantic enrichment with MCP, or Model Context Protocol, ensures that models don't just consume vast amounts of data, but truly understand it. We are leveraging consistent taxonomies, ontologies and entity concordance to unify disparate data sets and preserve context, enabling models to reason over meaningful relationships rather than unstructured noise.

The introduction of MCP has shepherd in the ability for LSEG's data to be safely presented alongside LLMs. We are extending the reach of our unique proprietary data while preserving the underlying licensing and controls. And this positions LSEG as the preferred partner and is enabling us to plug into agentic environments such as Microsoft's Copilot Studio, in turn, enabling the creation of trusted agents. With each such partner connection, we are opening new client use cases and opportunities, and you'll hear more about these opportunities shortly from Ron and Gianluca.

Let's now watch a short video to bring this to life.

[Presentation]

We are meeting our customers where they are, from our flagship Workspace integrated experience powered by AI to the enablement of our customers' proprietary solutions and through our strategic AI partners. This broad-based distribution is underpinned by our multi-cloud distribution, AI-ready content, APIs, agents and feeds. And regardless of how our customers prefer to consume and use LSEG's products, are ensure this is underpinned by trusted content. At LSEG, we provide trusted data to our customers to enable their trusted use of AI.

Over the past few months, we've announced a series of strategic AI partnerships from specialist partners, complementing our Workspace business, such as Rogo, to scaled partnerships with the likes of Databricks, Snowflake and cloud. And building the success of our partnership with Microsoft, we've extended our relationship by making LSEG's trusted data available as part of Microsoft's Copilot Studio. This is enabling customers to build custom agents in the Microsoft ecosystem with LSEG's trusted data.

To hear more about this, let me hand over to Irfan and Matt Kerner from Microsoft.

Okay. So Emily just covered the trusted data part of the 3-pillar strategy that we earlier talked about in AI. And our second pillar is transformative products. Instead of talking about product by product, as David mentioned earlier, you'll be seeing these products live in action. So instead talking about them, what we're going to do is we're going to talk to Matt Kerner, who's going to give us his perspective on our partnership with Microsoft and how we are co-building various different products using AI, collaboration and other technology.

So before that, just a quick intro or bio of Matt. Matt is Corporate Vice President, CTO in Microsoft's Commercial Organization. He oversees technology partnership within the worldwide sales and solution organization, collaborating closely across global commercial and enterprise customers and partners. As a Microsoft veteran with 24 years of experience in the company's product group, including roles spanning Windows, Azure and Microsoft Cloud for Industry, Matt knows our space, our customers and he knows LSEG.

Matt, welcome.

Thanks so much for having me. It's great to be here with this informed audience.

And before we start, I know you just landed last night. How is your jet lag?

I'm doing okay. I'm doing okay. I think I'll last through this conversation, but I'll reserve the right to go to sleep afterwards.

All right. So Matt, let's start with Microsoft and especially AI. I know it's a beefy and big topic to start with. But share with us about how you're thinking about AI? How is Microsoft thinking about AI? And what is it that you and Microsoft are most excited about?

AI is changing the way we operate at Microsoft. It's changing the way we develop our products and serve our customers and go to market. We see employees becoming more efficient. And as they become more efficient, they have time to exercise their creativity, and we see them becoming more productive in measurable ways.

We also see people learning new skills. So for example, a person who has no coding experience at all can now create applications and agents to make their jobs better, and a person who is an expert in their area now can focus on the specialized and most complex part of their job, which drives more fulfillment at work. And with that flexibility of people being able to do new things, we also see some new optionality in how we structure teams and distribute work across end-to-end business processes to drive better customer outcomes.

In financial services, what we see is AI transforming the way people do risk analysis. People no longer have to wait for laborious work by a team of analysts, so they can get immediate insights across many more options for actions that they take instead of a narrow set of options that they analyze. And as Emily said, they can also, with AI, analyze a lot more data than they could have made sense of manually. And so we see not only better decision-making, but more demand for the kind of differentiated data that LSEG provides.

We also now see the emergence of autonomous agents that can take on tests that were historically only accomplished by people. And so sort of the first tranche of that has focused on internal scenarios where employees are interacting with HR or IT. We now see this happening externally, customer-facing, where in sales functions and in customer support, we see AI agents driving results. We see it happening across transformation of business processes and perhaps most exciting for me personally is bending the curve on innovation where we see autonomous AI agents now participating in product development, writing code and taking on more jobs that developers have historically had to do themselves. And so that's really quite exciting.

So I think a theme that you'll hear through this conversation and, of course, through the rest of the afternoon is that through our partnership, we're bringing together enterprise workflows and technology with financial services specific workflows and technology. And by bringing those things together, we squeeze a lot of friction out of the system, we reduce cost for customers, we reduce time to value for customers and we give them better and simpler product experiences. So it's a tremendous opportunity with AI between us.

And in terms of the autonomous agent, I know you and I were talking about it this morning. Are you seeing people writing a lot of read-only agents? Or are they also making decisions and updating and actually changing the systems?

Certainly, there's much lower risk. When people are just reading things, you can have any employee develop agents that can read things. As soon as you can start to ride and kind of change the world and drive transactions, you have to be a little bit cautious. And so many times, our customers and internally, we're bringing professional developers and we're having more oversight on those scenarios. And so certainly, governance to manage the risk that comes with writing is going to be important.

Makes sense. Now turning to our partnership, we have -- obviously, it's a strategic partnership between our firms. Microsoft, obviously, is a shareholder in LSEG, and Scott Guthrie joined our Board in 2023. Give me your perspective on the importance of this partnership to Microsoft.

This partnership is a game changer for us. We have a horizontal platform capability that we bring to our customers. And with LSEG's differentiated data and vertical solutions and know-how, we see a lot of doors opening in the market that previously were not available to us. Our value proposition is more relevant to our customers, and we have simpler and better product propositions.

As Microsoft, it's always been our premise that we need to very carefully and thoughtfully and intentionally serve financial services. There's a very large target market for us. It's also the segment that provides growth and stability to the global economy. So it's important to us. And our observation would be that some of the tools, workflows, data distribution methods in financial services have not really kept pace with some of the changes that have happened in the rest of enterprise technology. And so there's a big opportunity for us to bring value and change to customers that they'll -- that will really help in their business.

And looking at LSEG, LSEG has this differentiated data. This data drives insights and actions for so many different market participants, the buy side, the sell side, asset management, insurance, banking, even corporate finance and so we felt the partnership with LSEG would bring us closer and with more relevance to all of those different audiences, which we think of as being very valuable. And the other thing that you could say about LSEG is I think there's multiple centuries of having established trust, which is tremendous. Microsoft also values deeply trust with our customers. And the open philosophy of LSEG is really helpful because we have customers who want to do all kinds of different things. So having an open ecosystem enables their scenarios to work.

Finally, I would say that the partnership mentality and the role that partnerships that play in LSEG across different businesses that you have is very clear, and we see this partnership between us as being no different from that.

From a customer perspective, what customers see is a more thoughtfully integrated out-of-the-box solution that just works, so they have less cost, they have less time -- shorter time to value and simpler products. And we deeply appreciate all the product feedback that you give us. You have helped make many of our products better. I could give an example with Microsoft Fabric. Microsoft Fabric is our data and analytics platform, which we make available to our customers to store all of their enterprise data and then run analysis in the enterprise. To date, we've had fabric generally available for just about 2 years. And in that time, we've acquired 28,000 paying customers. We grew 60% last year. That makes Fabric the fastest-growing data platform in the industry. And it's used by 80% of the Fortune 500.

Now in Fabric, we have this great horizontal platform that's used by many customers, and they have this broad data estate. The work we've done with LSEG is to bring LSEG data as a first-class capability into Fabric. So customers can discover the data, explore it and then access that data and analyze it, not only on its own, but joined with their proprietary data or other commercial data that they've acquired. In order to do that, we had to do a lot of platform improvement to Fabric to meet all the requirements that LSEG had for global data distribution. And so LSEG has played a key role in making Fabric a better product for every customer.

Not only that, we have customers who have lots of different ways that they do business, even inside of their own organization, department by department. So there's native integration with both Azure Databricks and Snowflake. And so LSEG data that shows up in Fabric can be consumed through Fabric workloads or through those partner solutions that also consume that same data set, which makes it very valuable. And we have similar stories around Copilot Studio, as Emily said, for people to create agents. We have that story in Teams and Microsoft 365. We have it in Azure. And across all of these, what you see is the merging again of financial services specific workflows with enterprise-specific workflows and platforms to be a more relevant low-friction, low-cost solution for customers. All of this is about product truth and solution truth, the statements we would make to customers about what they can achieve.

There's also a go-to-market side. Microsoft has had great relationships with the CIO organization for many years at most of our customers. But we don't often have deep relationships with the financial services line of business leaders, and LSEG has those relationships and has that deep domain knowledge. So when we go to customers together and tell our joint story, we can have a much more relevant, cohesive conversation that unifies their tech and line of business conversation so they can much more quickly get to a plan jointly with us on how they want to proceed.

The most exciting thing is when this partnership started, AI was not on our radar as an important thing for us to focus on. And all of these things that we've talked about were things that we set out to do at the beginning of the partnership before the AI inflection point. So a whole bunch of the foundational investments that we've made in these first 2 years of the partnership now put us in this pole position with AI, and we can very quickly adapt these things to bring differentiated AI value through the work that we've already done. So it's a very exciting time. I think we have an innovative future that we can drive.

Great. I'm sure I'm glad you mentioned the word innovation. But before we get to that question, I remember early when I first joined, we were -- AI was a thing, but not that big of a thing. So you're absolutely right. Now, it is becoming a thing and some of the work we've done, especially on the data pipelining, that really sets up really, really well.

So in terms of the innovation, we work very closely together. You and I talk at least once a week, if not daily, especially given many products we're about to ship out. From your perspective, from Microsoft perspective, what does innovation mean to you? And how does it work from a -- we're not just working together ourselves, but working with our partners and our customers. How do you think about innovation in that context?

We've had a lot of conversations with our customers. who have told us how meaningful this partnership is to them, and they've expressed their interest in helping to influence the partnership and shape the products that we build. And that's great for 2 reasons. First, it helps us build the right product because we get a lot of customer feedback to inform it. And second, it sort of prepares this initial tranche of early adopters who can deploy that product more quickly and get value out of it because they have confidence having shaped the product that will be the right one for them. So that direct customer engagement is really important. And as I said, in Fabric, the case of Fabric that I described before, we had many, many customer conversations that shaped how that product would go.

I think another one to talk about a little bit more is Copilot Studio, and we might unpack that one a bit because there's a lot of buzz and many keywords. And so maybe I can just go through it step by step and explain how that works and why customer input is important there. So I think it was 3 weeks ago, LSEG announced that there'd be a Copilot Studio-based connector for LSEG's MCP server to make LSEG AI-ready data available in our Copilot Studio. That's a mouthful.

What is MCP? MCP is Model Context Protocol. This is a way that you can make a tool available to an LLM that it can call to do something. It might push a transaction to a system or it might query a system for information. In the case of LSEG data, it's querying the LSEG data set to get information back for use in an AI workflow. And in addition to making that tool available, MCP lets you describe what the tool is and how to use it. So here's what the tool is, here are the parameters you can pass to it, here's what the results look like, here are some examples of calling the tool and getting the results. All of that text lets the LLM reason about what that tool can do and how to build it into a workflow.

So when LSEG wraps their data with MCP, it makes it easy for LLMs to interact with that data. And that takes trusted, definitive, up-to-date and accurate financial data and makes it available to any AI workflow, which is really important because it helps you ground that AI, which reduces hallucination and makes those results more trusted and reliable. So now LSEG has this capability with MCP. MCP is a standard, which can fit into many different AI systems. Now Microsoft Copilot Studio is one such AI system.

Copilot Studio is a low code and no code agent development platform. So anyone where they're a developer or a novice can write down what they want an agent to do in plain English or the language of their choice and have that agent created. And they can then use Copilot Studio to publish that agent to various different channels. You can put it into Microsoft Teams, you can put it into Microsoft 365 Copilot, you can put it into your website or your own application. You could even stick it on the end of a phone number, so you could do IVR and have somebody talk to the agent. And that agent can interact with other systems through connectors. It might be querying Microsoft Dynamics for CRM or ERP data, it might be querying Salesforce, it might be querying SAP or it might be querying LSEG data.

And the feedback we had from customers was, "Hey, we'd like to consume LSEG data, but we want it to be like a super simple, zero configuration task for a completely novice user, and we also want it to be enterprise-grade." When we say enterprise-grade, we mean it should work consistently and it should be governable. In addition to allowing people to create agents, Copilot Studio allows an IT department to inventory and govern those agents, so they can see all the agents that exist in their environment and they can set permissions. For example, every employee ought to be able to create an agent for their own use, but if they want to share it, maybe they can only share it with 10 other internal employees. If they want to share it with more than that, they have to go through a security and compliance and engineering review so we can make sure that the right thing is happening with that agent before we publish it out to the world inside of the organization.

So this -- the result of this customer feedback on wanting enterprise grade has resulted in LSEG being the very first partner of ours to release an enterprise-grade, zero configuration MCP connector for Copilot Studio. And because LSEG is first, it means that LSEG is bumping into some product gaps and some sharp edges and other things in Copilot Studio that we haven't ironed out yet. And I can say over the past 6 weeks, we've had a very tight loop between LSEG and Microsoft product folks, ironing out those bugs, getting those bug fixes pushed to production and paving the path for every subsequent customer who's going to come use that connector in Copilot Studio to get their job done inside of their own organization. So that feedback has been super valuable for us.

I think that's kind of an explanation of how customer feedback drives the stuff on Copilot Studio. Can I talk about DMI? Would that be okay?

Sure, of course. I mean, look, DMI, I think David mentioned earlier, it's a product that we just had our first transaction with a couple of -- a month or 2 ago. And I remember talking to you when I first joined and I know you are considered an expert in blockchain and digital assets. So yes, please talk about DMI.

Sure. Okay. So DMI is this sort of modern cloud-based infrastructure for life cycle management of digital assets from cradle to grave. And we started talking about DMI and we said, "Hey, we can build this thing, and we work together to build this thing on Azure." And once we built it, LSEG started to talk to the world about it, and LSEG got this flood of customer interest from customers who either wanted to onboard assets or transact on the platform. And it's great to have that signal.

The challenge is when you have a new product like this, it's very hard to go from 0 to 1. You have to balance many different considerations. And the thing that, that product feedback did from those customers, the expression of interest and their fine grain feedback on what they wanted to do, combined with LSEG's market knowledge and relationships with those customers, LSEG has orchestrated a very intentional path to go from 0 to 1 and then from 1 to scale. And that's a lot of trade-offs. You have to manage time-to-market, you have to manage which jurisdictions you're in and what regulatory requirements they have. You have to decide which asset classes you want to support, what workflows you want to support and which customers you want to onboard so that you maximize liquidity and flow on the platform to make it relevant and to get scale.

And so it's been great to see LSEG chart the course for this product where we had a technical point of view on how that product would work, but LSEG knows how to take it to scale with customers. And so that's another place where I think customer feedback has really driven very intentional product development and product management and a product mindset. And I would say this is where LSEG's data, LSEG's domain knowledge, LSEG's vertical solutions, plus our platforms, these are all examples of where we're breaking new ground for the industry, and we're doing it hand in glove with customers.

Matt, on your point around the Copilot, we don't mind being guinea pig because not -- so I was -- when I first heard that we were the first enterprise-grade MCP connection on the entire Microsoft plant on the Copilot and Copilot Studio, it was good to hear because we're learning at the same time. And for us to learn much ahead of anybody else makes it better for us as well. So we don't mind co-creating.

So one of the innovation that our teams have been working on and showcasing today is Open Directory. This has been an incredible partnership between our teams. That's been going on for a bit. It solves a clear customer problem. It's secure, it's compliant, it's across from communication augmented by LSEG's Workforce -- Workspace and LSEG's workflows. Tell us more about your thoughts on Teams and on Workspace because now we have brought both of these products together, and it wasn't a snap of the finger off you go, and we have this product life and you invested a lot in it. So from a Microsoft perspective, how do you view Open Directory?

Let's start with Teams. Teams is an enterprise collaboration platform. People can do chat, they can do video calls, audio calls, collaboratively edit documents, work in a shared canvas. There are a lot of things the Teams can do. When we last reported on Teams' usage, it was in our fiscal year '24, so the number is a little bit dated. But at that time, we reported 320 million monthly active users on Teams.

Teams, for those users, is a part of their daily routine. They log on, they interact with it, and it's part of the air they breathe and it's a system that they live in. Similarly, it's integrated into the IT environment in the organization with identity, security, networking, data policies. So all of those things are in place with Teams. What many people don't know is the Teams has a feature called Federation, where 2 different organizations can have their users chat with each other. And so we have this between LSEG and Microsoft. I can type Irfan's name in the address bar of my Teams, and his profile shows up, I click it and I can send him a message. Emily and I were doing this morning with a couple of links we were sharing. And so we can chat back and forth. And that works on desktop, web, mobile.

True story, about 5 or 6 weeks ago, I was in a 12-acre corn maze with my wife and 4 children, and we were lost in the corn maze. And at that moment, Irfan pinged me saying, "We need to talk about an issue we found in Open Directory." And I was like, "Well, I'm lost in a corn maze. Can I reach you after we find our way out?" And he said, "Sure, sure, sure. Prioritize your family and then we can talk later." And then indeed, we did talk later.

And so you get alerts and you get that chat, and it's just like it is inside of your own organization. Microsoft is pretty free and easy with Teams Federation because we love chatting with our customers that way. But many financial institutions do not turn on cross-organization Federation because there's risk associated with having your employees talk outside of the organization. So we don't see high penetration of Federation inside of financial services. And so when LSEG came and said, "Hey, we'd like to do Open Directory." We were quite excited. We said, "This sounds great. Let's go." And they said, "Well, wait a minute. We think there are some things that you need to do in order to make Teams better so that it will be ready for these customers."

And so we've spent 2 years working on a shared backlog of things that we needed to do in the Teams' platform to make it ready for this use case. For example, 2 recent features. This fall, we enabled something called Trust Indicators. So next to every person in conversation, we mark, is it internal or is it external? So that way, a person inside of our organization knows whether they're having a conversation with an external person and they can gauge what they say. I see Nej nodding there, that was very important. We got your feedback.

And then I think the second one, which just became generally available a week before last is granular controls. An IT admin can now say, "This set of users is authorized to chat externally. Maybe it's front office people who have a business need to do that. And there are back-office people who are not authorized to chat externally because they have no business need to do it."

And so you can turn on Federation for just those users who should need it. And so that capability was also very important. These are examples of platform things that we had to do. And then LSEG said, "Look, what we'll do is instead of -- if you have a new member who wants to join a network, with the way Teams Federation works out of the box, that organization would have to go talk to every single one of the other organizations in the network to do KYC and vetting and then technical onboarding to get the Federation turned on both sides." And I'll say, "Look, we'll be the centralized clearing house. We have a KYC business that's a leading business. We'll do the KYC on behalf of the network, and we can facilitate and orchestrate that technical onboarding." And so we said, "That's great. Let us build a solution to do that, which we call Automated Domain Management."

With Auto Data Domain Management, there's a way for an enterprise to come and say, "I want to be part of this." LSEG does that vetting. And then that configuration is taken, and there's a little piece that runs inside of each organization that they deploy when they onboard, which picks up a trusted centrally distributed configuration from LSEG, validates it and deploys it locally so that every member organization picks up that new federation and turns it on right away.

And I think I can announce that this past weekend, LSEG became the very first tenant deployed in production with Open Directory. And in the next 5 or 6 weeks, we're going to go take it to other customers together, which is super exciting. Look, the whole point at some level of financial services is -- at least capital markets is to facilitate transactions across counterparties. And when that stuff does not happen in Teams, we're not living up to our mission to empower every person in business around the world to achieve more. And so we really want all of the sorts of activities that happen in a business to be able to happen on Teams. And so for us, bringing this kind of collaboration into Teams through Open Directory is really strategic and exciting. And so we're delighted to see this thing happen. And as you said, this was not an overnight job. This is 2 years of platform work and solution work. And so part of the thesis of our partnership is we can tackle hard problems and see them through. And so it's great to have this data point show that.

Awesome. It's my last question. I know we're probably a little over time over here, but we covered a lot today. We talked about DMI. We talked about MCP. We talked about Open Directory. We talk about Agentic platforms. I'm not sure people keeping track of it. We've talked about a lot of these products that we're working together. What excites you? What's next? What is it that you think -- what is exciting you for the next few years for us working together?

I like to think about like the foot work that we do to get in position and then the execution we can do once we're in position. So the foot work has been a whole bunch of this foundational work. We've got a team that operates as a joint team. That's no small thing to build that team. It took years of work to build that team. I think we have top-to-top alignment that's very clear. I spent a significant portion of last week with many members of LSEG's Executive Committee in Redmond. And now I'm here today. I have untold frequent flyer status, and I know the people at the hotels around St. Paul's and they know me, and so that's very exciting.

And so we've got sort of all those pieces in place. And then you look at the technology foundation. We have a regulatory compliant footprint of LSEG and Azure. We have LSEG data in Fabric. We have the Copilot Studio integration with the LSEG MCP server and AI-ready data. We have DMI. We have Open Directory and Teams Integration with M365 and with Open Directory. All of these things are now in place. And so you can imagine some very exciting scenarios. And it does not take a big leap to now describe a scenario, and I'll just like hypothesize one.

This 2 years ago would have been inconceivable to talk about this scenario. But now the scenario seems like obvious and achievable. The scenario is this. Let's say you have a conversation going with a counterparty in Teams, someone in a different firm, that connection is facilitated by Open Directory and you have an investment thesis. You go into Workspace and you look at some economic indicators and you produce a chart, you then take the link for that chart, you share it in Teams and it comes through as a first-class thing on the other side. They see the chart embedded in Teams. They can click it and jump in their own Workspace deployment, deep linked, straight to the same context that the originators had. And so that person can do more deep analysis and evaluation of the data in the chart. So they chat about the investment, and you decide, "Hey, this investment is valuable. I want to go pursue the next thing."

So in M365 Copilot, you initiate an agent that you built in Copilot Studio that uses LSEG's Copilot Studio connectivity. And the agent goes and it retrieves the transcript of the chat in Teams and extracts the investment intent from that chat autonomously. Then it takes that and it constructs a scenario, and it delivers that scenario to LSEG's modeling as a service running in Azure with a risk model that runs over that possible investment thesis, using LSEG data coming from Fabric to do a bunch of pricing and risk and volatility analysis, whatever the people who know this stuff know how to do. I don't know how to do that. That does that stuff and it comes back with data. And that data, in turn, goes to a deep reasoning model running in Microsoft's AI foundry. And the deep reasoning model looks over that result of that analysis and puts together a proposed trade with an explanation of why that trade makes sense and maybe what hedges you might want to do and whatever other things, again, these people do that I don't know what it is.

And it comes back and is presented in M365 Copilot, along with an accounting of the recent e-mails, chats, files and meetings that happened in your enterprise that you have access to, to make sure you're not missing something about the context on any of that -- those securities that are in the list. And you look at the trade, maybe you edit it, maybe you approve it and you send it off to Workspace for execution.

This all can happen in like minutes, maybe seconds if the analysis doesn't need to take that long. It doesn't require you to depend on a team of analysts. It's entirely compliant and audited in your enterprise. It's consistent and repeatable. So every single person in the firm can get the same result if they want to ask the same question. This is the new AI standard that we're building to, and this is going to be the direction that we go in the partnership. It's very exciting to contemplate. And again, it's merging together financial services specific workflows, enterprise platforms in a way where Microsoft and LSEG bring complementary strengths to the partnership and we are very excited for the value we'll deliver for the customers.

I know we did not prepare for the scenario you just mentioned, but for the audience, this -- as I hear what Matt just said, you will see many demos today, which brings a chunk of that workflow to life. So I think that it's exciting to hear what you just mentioned. And thanks so much for your time. Really appreciate it. Thank you.

Thank you.

Matt, Irfan, thank you so much. I think one of the things that really resonates for me is how we're bringing the full force of LSEG and Microsoft to co-develop for our customers.

So now let's come back to the third pillar of LSEG's AI strategy and focus on intelligent enterprise. Across LSEG, we are deploying AI to innovate faster for our clients, boost productivity and to transform our data and content operations. At the top of this bubble image, as you heard from Irfan earlier, you see some of the examples of the AI transformation happening in our engineering organization. This AI-enabled transformation is also permeating through other parts of our organization. In our sales team, we're actively using AI to identify prospects and support client management. In our customer support teams, there's already an 87% adoption of LSEG's proprietary QAS, or question-and-answer service, designed to support our customer consultants, providing consistent and timely responses. Indeed, we're already seeing up to a 40% reduction in the overall time to resolve customer queries with 50% being resolved in under an hour.

In content operations, we're similarly seeing the benefits of our AI deployment. We're using the latest techniques in AI combined with our deep data expertise to optimize delivery of the highest quality content to our customers. We're already seeing a 9x faster content extraction rates while simultaneously improving accuracy. We're not only bringing the benefits and speed, we're also seeing the gains in efficiency. We have realized a 51% employee reduction in central sourcing and a 66% reduction in cloud costs as part of data scraping activities.

While we are going faster and with greater efficiency, we are not compromising on quality. We are getting even stronger. Data quality issues reported by customers are down 52% on content volumes that have risen by 45% since the beginning of 2022. And content extraction success rate has increased to 98%. And the breadth and depth of our content just keeps expanding. To say the volume of what LSEG is providing has grown substantially is an understatement. As an example, our exchange-traded fund holdings data has increased by 400%. AI is already creating new opportunities for LSEG and our customers with our trusted data, transformative products and intelligent enterprise, combined with our solid infrastructure and strategic partnerships. LSEG is uniquely positioned at the forefront of this change.

We're now going to take a short break. Our next session will start at 5 to 2:00 with Ron and Gianluca. Thank you.

[Break]

Hello, everyone, and welcome back. I'm Gianluca Biagini, Co-Head of Data & Analytics, alongside Ron. I joined LSEG 3 months ago, and I'm really excited by the tremendous opportunity we have at LSEG to transform how the industry operates. Previously, I was Head of Data Valuation and Risk Analytics at S&P Global.

And for those of you I haven't met, I'm Ron Lefferts. I've been with the group for 4 years, and previously led our sales and account management function, a role I will be handing over to Chris Coleman, when he joins LSEG in January. Prior to joining LSEG, I was a Global Technology Leader of Protiviti, and I've also held senior leadership roles with IBM.

Now let me start with an overview of our current positioning, including how we are partnering and innovating before handing over to Gianluca to detail our plans to accelerate growth within the division. We will also give an update on our partnership with Microsoft.

We are a leader in a GBP 35 billion global market for financial markets data and analytics. Think about the systems in your own institution, order execution, risk management, market surveillance, portfolio management, fund valuation, performance monitoring and many others. They all rely on huge quantities of timely and accurate data to power them. And there is a very good chance those systems are running on our data. 3/4 of our largest customers use 20 or more of our products, typically to help them with highly regulated business-critical activities. Our solutions are deeply embedded in the global financial ecosystem with real strategic partnerships grounded in expertise and trust that have been built up over decades.

And the market in which we operate is growing on multiple fronts. Customer expectations are rising all of the time. Of course, they want real-time data, but some want ultra-low latency feeds, and they want that data to cover a much broader range of asset classes and for that data to be consistent across multiple platforms. Increasingly, they want to run AI on that data, too. And as you heard Irfan and Emily talk about, AI models are incredibly data hungry. This is all driving demand for financial data and analytics, turning trusted and accurate data into actionable insights.

Our customers spend the world's largest banks, asset management firms, corporates, wealth advisers and central banks. For banks and other sell-side firms, the largest proportion of the D&A revenues, we provide fully integrated workflow solutions. Our flagship platform, Workspace, provides a modern, customizable user interface for integrated execution and order management workflows. Investment and wealth managers rely on our unparalleled, Tick History data and analytics. Here, we are the leading global provider of real-time data with unmatched scale, depth and breadth.

Our AI-powered analytics platform helps create market validated models and tools to discover insights faster, while corporates rely on our business-critical data and workflow tools from treasury management to company fundamentals and comprehensive news coverage. But it's the combination of those capabilities anchored in the full breadth and depth and quality of our data that allows us to provide innovative, distinct solutions.

The demand for and consumption of data is accelerating, and we are facilitating that growing wave. The chart on the left-hand side shows the amount of data or messages coming through our real-time data feed. During Liberation Day, up to 20 million data points a second were processed. This information on completed trades, price moves or indications of liquidity, critical market data that participants need. And while Liberation Day was an extreme event, there has been a 4x increase in real-time data over our network in the past 10 years, a trend that we expect to continue. And with our direct connection to nearly 600 exchanges and venues and our ongoing investment in technology and capacity, we are strengthening our market leadership.

On the right, you can see the demand for our Tick History data, covering 100 million instruments over almost 30 years. It's a powerful data set that is highly valued by our customers with roughly 4 million customer data inquiries a month in early 2024. But it's also massive, tens of petabytes in size, which some customers found a struggle to manage when they had to receive it as a file. Last year, we made that data available in the cloud for the first time. And you can see the impact that it's had on consumption of that data. Now it's 6.5 million customer requests a month and climbing. And that speaks to the broader truth in the industry.

Although data is abundant, not all data is equal. Markets don't just run on what's publicly available, they run on what's trusted. Customer demand for data that is accurate, comprehensive, verified and auditable is significant. And that's where LSEG sets the standard. And if you make it easier for customers to access and consume this data through new cloud distribution channels or AI partnerships, you are also likely to sell much more of it. We are only near the beginning of that journey.

The same approach of meeting customer demand through relentless innovation is at work on our Workspace platform, too, where we have driven a lot of change over the last 4 years. In June, we successfully retired Eikon, one of the largest financial services workflow migrations in history, moving more than 350,000 users onto Workspace and establishing a common platform for innovation and growth.

We also continue to enhance functionality week in and week out. With something like 500 updates a year, Workspace today is far more powerful than it was even a couple of years ago. And as you will see shortly, we are accelerating this evolution further in the coming months and years. And the increased power of Workspace is evident on how our customers engage with the platform. They are not using it for just 1 task. In fact, all key customer communities, traders, bankers, investment managers are regularly using 10 or so different Workspace applications. And their engagement with each of these functions is increasing, too, with the average trading customer now using desktop applications 60% more than they were a year ago. This shows the success of the Eikon to Workspace migration. The new platform is easy to use, intuitive and drives much greater engagement from customers as a result.

Our disciplined focus on executing our strategy is translating into faster growing and more resilient business. Customers are keeping our products for longer with retention up 200 basis points since 2021, and we are winning more business with a 700 basis point step-up in win rates over that same period. Revenues are growing as a result, and we have a clear plan to further accelerate these growth rates in 2026 and beyond.

Now let me hand over to Gianluca to talk through our strategy for future growth.

Thanks, Ron. Our ambition is simple: to be the leading provider of trusted data and actionable insights for customers. To do this, we have 4 strategic pillars: one, expanding our data leadership; two, transforming customer workflows; three, maximizing channel reach; and four, building an efficient and scalable platform for growth. As Ron has outlined, we are coming from a position of strength in a combination of trusted data, technology and talent.

Let me spend a few minutes breaking down the priorities under each of these 4 pillars. For decades, our data has been the foundation for critical financial decision, operation and capital flows worldwide, proprietary licensed content with unmatched depth and breadth, spanning asset classes, institutions and geographies. But we're not standing still. We continue to enhance and strengthen our content offering to maintain our lead. This has been so exciting for me personally to join such a powerful franchise and help drive the next phase of growth through data leadership.

In news, we have expanded our leading news content through a partnership with Dow Jones, adding to thousands of revenue sources, including exclusive access to Reuters News. We have also built on our partnership with Reuters with the launch of Reuters Super Summaries, an AI-driven earnings intelligence to deliver concise earning insights at speed. And we are focused on making our news machine readable. This means investors can combine our news output with sources like Tick History to drive valuable insight into what really moves share prices from second to second. Todd and Tim will give a great example of this in a few minutes.

In private markets, an important growth area, we have added leading data sets in Preqin and Dun & Bradstreet. And we have announced last week that LSEG will license Nasdaq eVestment private market data sets. Together, these data sets provide an end-to-end curated view of private markets in a way that others cannot. This is actually a fantastic example of the strength of our distribution platform and flexibility in our strategy. As you know, private markets data is quite fragmented with no single information services company owning comprehensive coverage. Through organic investment and partnership, we have built that coverage ourselves. And this extends across to the FTSE Russell partnership with StepStone as well.

And finally, in the third box, we continue to build our unique assets, expanding real-time in Tick History and embedding Tradeweb enhanced fixed income data in our services.

Under Pillar 2, we are enhancing seamless end-to-end workflows with Workspace. As David often say, it's not AI or a desktop, it is AI in the desktop. We have made countless announcements over the last couple of years, and have some really big developments over the next 6 months. Nej will be bringing just some of them to life for you in a few minutes.

As Ron highlighted earlier, our customer use Workspace for a wide range of applications. It is fully embedded in the workflows. How can we make that even stronger? First, through AI integration to search, summarize and analyze, all built on our trusted and accurate data, which customers can rely on for critical processes and decision-making. Second, through collaboration, whether through Open Directory or our trading functionality. And third, through our application that are dedicated to specific user types, whether it's in commodities, investment banking or wealth. And of course, this will all combine financial services workflows with enterprise workflows end-to-end as we fully integrate with Microsoft Teams and Microsoft 365.

We have always had a multichannel approach to data distribution, as Emily outlined earlier, through our own UI Workspace through direct feeds and distributed by third parties. As demand for data grows and AI use becomes widespread, new channels are opening up all the time. And our customers are working with data in a number of new environments. Our LSEG Everywhere approach is focused on delivering AI-ready data to where our customers are working.

We are using MCP to enable discoverability for LSEG content in LLMs through scalable distribution ecosystem, appropriately governed and licensed. And we have developed multi-cloud content distribution through AWS, Google, Azure and Snowflake, offering customers choice. Again, you will see several of these platforms and use cases demonstrated in a moment. And I will also cover what the monetizations and growth opportunities are.

Finally, we are investing to transform our data infrastructure to deliver more agile, resilient and scalable platforms. The migration of data and application to Microsoft Azure is an enabler for more consistent data onboarding and faster product delivery, providing a more unified customer experience across data sets as well as reducing infrastructure costs. We are making good progress here. And on a real-time network, as Ron mentioned, we have just embarked on a 5-year investment plan to deliver a step change in capacity and intelligence.

So let's see how all of this translates into what matters, how we will capture the value from the huge growth in demand for data and accelerate our growth. Let's start with the traditional levers, retention, displacement and value realization. Our products are getting better and better. And as a result, we are confident we will continue to improve retention and steadily displace competitor over time. Remember, even with amazing products, the rate of displacement can feel slow because changes can be disruptive for customers. But we believe that this can be a steady long-term tailwind.

Next, price realization. Ron showed that our real-time traffic is up 4x over the last 10 or so years. So demand for our services is growing at a huge pace. We think that, that can be reflected more in what our customers pay for our services over time. On desktop, we have said that -- we said that before that for our high-end Workspace users, there is around a 30% price gap to a major competitor. As we import functionality and we build networks with products like Open Directory, that give us the opportunity to close that gap over time. Customers will see the value. It is early days, but we see scope for new revenue streams. If we're driving revenue growth for distribution partners through their compute or subscription, then there is an opportunity to share in the upside that we are generating for them.

And finally, increased usage and users. As we modernize our infrastructure, we are introducing more and more telemetry into our stock. This will allow us to move to a more hybrid subscription and usage model, giving customer control and visibility on the spend and capturing the value of usage growth. As for new user, the spread of AI and new application is democratizing data like never before. Every industry vertical, every professional services firm can make commercial use of financial data. So the opportunity to reach adjacent markets is opening up like never before.

Alongside all of these levers, we have our long-term enterprise agreement, or LDAs. These are selective strategic partnerships. As you saw from David earlier, we expect this to represent a run rate of around 70% of ASV as we exit this year. This cement viable long-term partnership with some of the world's leading institution, building product road maps together and giving good visibility to both parties. The breadth and the depth of our data makes it hard for others to fully replicate. As you can see, we are very excited about the breadth of positive commercial outcomes our strategy will give us.

Ron, back to you to update on the progress with the Microsoft partnership.

Thanks, Gianluca. Our partnership with Microsoft is a key aspect of our overall strategy. And it was great to hear from Matt Kerner earlier about how important it is to Microsoft as well. In fact, Gianluca, David and I as well as a few other colleagues were in Seattle last week for a few days to have a detailed partnership catch-up with a number of Microsoft leaders. We've made good progress over the last 3 years with the partnership moving from production, product ideation to product build and increasingly into product delivery.

However, inevitably, with partnerships on this scale, this process has not always been as fast as we would have liked. In some areas, we needed to build the foundations of the platform before we could scale data migration, which is very important to get right, even if it's not glamorous. AI was barely a thing when we started out and is now fundamental. Customers were initially nervous about new product adoption, particularly around protecting their own data. So compliance proved quite a barrier to onboarding for some time. And our early stage product launches have shown the importance of engaging with customers throughout the design process, informing a broad range of design aspects from product onboarding, iteration and co-innovation as well as the importance of applying a community lens to product rollout to drive adoption.

But let me remind you of what has already been delivered. Each of the products on this slide are either live or will be in the coming weeks. For example -- and applying the community lens I just mentioned, we're rolling out Open Directory to FX and commodities users where our workflow tools are already deeply embedded. And by connecting these users and giving them tools to surface, share and collaborate on content, we will further deepen and extend these communities.

Crucially, Open Directory will be using Microsoft's automated domain management, or ADM. You heard Matt and Irfan talk about this earlier. And ADM supports the onboarding of external customers into Open Directory for secure and compliant intercompany workflows. It's a key differentiator for LSEG in enabling secure, federated collaboration across financial institutions.

We're also making our AI-ready financial data available both in Copilot and Copilot Studio, enabling customers with LSEG licenses to build their own agents, working with our data, embedding our solutions across the finance industry and beyond.

Through the launch of the Analytics API and its extension to Visual Studio Code, we have nearly doubled the rate of growth over the last 18 months. We have fully replatformed our trade routing solution for 1,600 investment managers and banks, creating a first-of-its-kind cloud solution that is faster, more scalable and more resilient. This platform is currently handling trading of roughly 4 billion securities a day.

And staying on the theme of trading, we delivered the first transactions on our new digital markets infrastructure in Q3, deploying distributed ledger scalability and efficiency across the full asset life cycle of a trade from issuance, tokenization and distribution to Post Trade asset settlement and servicing.

Looking ahead to next year, we will accelerate our pace of delivery. We will expand our data leadership. In particular, we will grow our private market data feeds substantially, combining our own proprietary sources with leading data sets from Preqin, Nasdaq AND Dun & Bradstreet, and delivering that through an intelligent combined feed covering private credit, equity, infrastructure and real estate. Gianluca called this out earlier. We will transform customer workflows with the full launch of Workspace AI, including the scale-up of Workspace Teams, Open Directory and the integration with the Microsoft 365 suite, bringing huge benefits to our customers.

We will maximize our channel reach, making all our D&A data feeds AI-ready and building AI-enabled analytics intelligence and automation. And we will continue our work, building an efficient and scalable platform, both through the ongoing migration to Azure and opening up our own Analytics API for customers to distribute and monetize their own models via our infrastructure.

Now there's a lot to digest here. And as we invest in content and accelerate innovation, creating new partnerships and deepening existing ones. But as Gianluca said, our ambition is simple: to be the leading provider of trusted data and actionable insights for our customers. We're excited by the numerous opportunities we see ahead and have a clear strategic focus for delivery.

And before I hand over to Todd Hartmann, a quick word on what you will see over the next hour. Here's the model of content and distribution, which underpins LSEG Everywhere, which Emily covered earlier. So if a customer wants to produce a detailed company report or a piece of fixed income analytics, they can do so in any environment through their own UI, through our UI Workspace based on our direct feeds or through other consumption layers provided by our partners. And we will showcase all of these options.

So thank you. And I'll hand the floor to Todd to kick off the Data & Feeds and Data & Analytics demos. Thank you.

Thank you, Ron and Gianluca. And hello, everyone. Welcome to our session on Data & Feeds. My name is Todd Hartmann, and I lead this business. I'm joined by my colleague, Tim Anderson, who heads up our Tick History and Quantitative Analytics business. I've been with LSEG now for a few months. And prior to that, I was with FactSet for about 19 years where I helped to build their Data & Feeds business. We thought it would be helpful for me to start by providing an overview of our data and the value it provides to our customers. I'm going to cover the challenges our customers face, the value of our data, its breadth and depth and how our customers are deploying AI on our data. I'll then hand it over to Tim to show you an example of how we solve a specific customer need.

So our customers face a number of challenges when using AI on financial data. Before AI can reason, it needs order. The problem is our clients are managing a lot of data from many different sources. Each data set speaks a different language with no single identifier to align them. So even the most advanced AI can't see the full picture in a reasonable amount of time. It's essentially like navigating a new city without a map.

Let me now explain our approach. On the left-hand side of this slide, you can see that we give our clients access to one of the world's most comprehensive libraries of financial and market data, including contributions from over 40,000 customers covering decades of history, and we provide this data in a connected and consistent way.

And on the right-hand side of the slide, as Emily covered earlier, all of that data is curated and mastered. At the heart of our approach, our 2 industry-standard IDs, which link and align data sets across asset classes and systems. It is this ability to structure and connect data that truly sets us apart and sits at the center of everything that you'll see here today.

I mentioned a moment ago our industry standards. LSEG has a unique framework in place. On the left, our normalization allows all data to speak the same language as I covered on the previous slide. That means there's a consistent structure across every data set and asset class, which reduces noise and minimizes AI hallucinations. This means faster and more accurate AI responses. Any model knows exactly where to find the data point because we provide a map of our data structure.

Next is the Reuters Instrument Code. The RIC provides unique point-in-time identifiers for over 80 million listed instruments, connecting across standards like CUSIP and ISIN. And finally, our PermID links entities, people, companies, instruments and sectors, forming the connective tissue between our data. For example, in the column on the right, you can see LSEG's PermID is linked to CEO, David Schwimmer and to the banking and investment services sector, listed on the exchange with RIC LSEG.L. Together, these elements create a data environment ready for AI.

So for Data & Feed specifically today, we'll show you an example of how our clients can use our data with AI in the cloud, transforming how investment workflows operate. We see AI as an extraordinary opportunity to accelerate Data & Feed's growth. Our AI strategy makes our data sets AI-ready and available wherever and however clients need them.

I'm now going to pass this over to Tim, who will walk you through an example. Some may find this a bit technical, but we thought it was important to show the complexity of what's involved. So over to you, Tim.

Thank you, Todd. Hello. I'm Tim Anderson, the Head of Tick History and Quantitative Analytics, formerly from Trading Technologies at Deutsche Bank and JPMorgan.

So what sort of things can our customers do with our data, particularly enhanced by multiple agents as many customers are working towards? Well, here's a use case that we'll be showing today. This is not just a chatbot. It's an AI-generated back-tested analyst report that can show data from recently or look for signals in the past, and the AI insights can be sent to any endpoint a customer wants, and it can be tailored to be as complex as the customer requires.

This example shows what's now possible for LSEG Everywhere, a report built entirely from LSEG data with the option for customers to add their own. It correlates news events, sentiment and ESG scoring, trading volume and performance metrics, all connected automatically. This is insight at machine speed, producing a tailored report focused on investable securities across any exchange, complete with a reason buy, hold and sell rating based on a customer's criteria and in-depth summaries across multiple data indicators. And this intelligence begins with our data foundation.

For the example shown, at the core of that foundation are 4 key products. First, as David had mentioned, Tick History, one of LSEG's flagship data products. It provides a complete timestamp record of every trade and quote on major global exchanges over 30 years. It contains petabytes of data at 1 billion captures per second.

Next, quantitative analytics, our data and analytics environment for quants, portfolio managers and data scientists who need large-scale, high-quality data for modeling and research. It integrates over 60 data sets covering pricing, fundamentals, economics, ESG and sentiment, all AI-ready.

Third, machine-readable news, our AI-ready news feed that transforms Reuters journalism into structured time stamp data, turning headlines into real-time signals.

And finally, markets psych, behavioral and sentiment analytics, quantifying how people feel and talk about the markets in real time. This is data that's quantitative, sentiment-driven and contextual, a living record of the world's markets.

Now we bring all this to life with AI agents. Across LSEG's data sets, we now have agents working in parallel automatically. They query, they calculate and they merge data in concert. From left to right on the screen, you'll see these agents at work, each special analyst scanning the data, collecting, calculating and creating insights live. One agent pools trading volumes and calculates VWAP, another analyzes market sentiment from news and social media, another assesses ESG performance. Each is laser-focused, and they all come together to produce a unified result like worker bees in a hive.

If you want to update just one part of the analysis, say, ESG, you simply modify that agent and it updates instantly. All of this runs seamlessly across Google BigQuery or in the customer's own cloud. It's AI-ready at scale. We can even add regulatory agents that automate reporting and validation, so every output is auditable and compliant.

So in short, you've got a network of intelligent agents doing the heavy lifting, freeing analysts to focus on decisions, not data preparation. And now, how do we handle one of the toughest challenges, transferring large amounts of data to customers? This critical component to making work is cloud delivery. Most data solutions stop at access. Ours goes all the way to delivery. You saw a chart earlier that Ron presented showing how Tick History consumption accelerated significantly when we move to the cloud. Well, this platform runs natively in Google BigQuery, and think of it like a data jigsaw, we snap our data directly into the customer cloud. No downloads, no storage management, just secure delivery, which significantly reduces our customers' total cost of ownership, whereby the customer can query LSEG data on demand and their own data at the same time. LSEG securely shares the data into the customer's environment where AI agents execute securely inside the projects.

Everything stays connected, governed and auditable. Data moves without ever leaving it secure at home. By combining AI-ready content, agents and cloud, now let's see how the agent engine example brings it all together.

[Presentation]

Every customer problem we saw earlier has a direct LSEG solution. AI agents instantly collect, summarize and connect the data. Our data model integrates Tick History, quantitative analytics and news, all working seamlessly and customers can add their own data. RIC and PermID, as Todd went over, ensures consistency across systems. And with context memory, intelligence carries forward from one report to the next. And it's this architecture that creates a new commercial opportunity for LSEG.

What does this mean for our business? More revenue opportunity. Historically, LSEG generated revenue from selling data both directly to the end customer and via third parties like Aladdin. That will continue to be the driver of our growth. Today, we are adding new distribution channels as we accelerate LSEG Everywhere. These new AI and cloud-based applications add value for existing customers and allow us to reach new customers. And in the future, we can monetize the example you've just seen, selling data with multiple functional agents, which extract context, insight and relationships or even purely computational agents performing high-value analysis. In summary, we'll be selling intelligence inside the data where the reasoning itself becomes marketable.

Thank you again for your time, and we look forward to continuing the conversation after the session. We'll now pass to Mikhail and Adam from the Analytics team. Thank you again.

Thank you, Tim and Todd. Good afternoon, everyone, and welcome to the Analytics section of our LSEG Data & Analytics presentation. My name is Mikhail Bezroukov from the Analytics Product Management team, and I'm going to take you through some of our latest developments and core themes. I'll then hand over to my colleague, Adam Towne, who will walk you through some product demonstrations.

So first, an introduction. LSEG Analytics provides our clients with the tools, models and information that they need to make good decisions and drive their businesses forward. Our analytics models cover hundreds of asset classes across instrument pricing and valuation, predictive analytics and risk models. They're deeply embedded in clients' critical operations and ecosystems through our analytics API. And today, they're used for investment research and alpha generation for forecasting and risk management.

Our customers range from small start-ups to the largest financial institutions. And for many years and across multiple turbulent market cycles, thousands of customers have relied on our analytics for accurate, trusted insights. So LSEG Analytics is focused on 3 core areas and benefits for our customers: coverage, scale and efficiency.

So first, we deliver substantial analytics coverage. Our clients draw upon a multitude of input data sources and extensive model libraries to help solve the critical challenges that they're facing. For many years, we've offered hundreds of models through distinct channels, including many Workspace apps and these are now brought together through our Analytics API.

So this brings us to the second point of scale. Rather than offering many disparate solutions, we ensure that clients can access our analytics outputs through a single consolidated analytics API. The API makes our models available to clients in a cohesive manner at large scale and allows for a deep side of customer-specific customizations. It connects directly to users to their internal platforms or to other downstream systems.

Third, we focus on improving our customers' productivity and efficiency of work. We do this by making our Analytics API much easier to access and interact with through our AI-ready initiatives and partnerships. In this, we're fully aligned with our overall LSEG Everywhere strategy that we've spoken about elsewhere today.

So we will speak about the integration of our AI-ready analytics with the Databricks cloud platform, with our MCP-powered AI agents, and we'll talk about our proprietary integration with Visual Studio Code. These initiatives save our clients' critical onboarding time and high technology cost while continuing to give access to our trusted, deterministic analytics models.

Our partners choose to work with us because our analytics are already structured for AI consumption because our models have proven accuracy over many years and are already familiar from Workspace apps and because they are supported by trusted LSEG data that my colleagues, Tim and Todd, have just spoken about. The depth, breadth and accuracy of our analytics is unmatched, and we enable our customers to work in new ways whenever and wherever they choose, be it in the Analytics API, in an AI agent or, of course, in a Workspace app. It's LSEG Everywhere in action.

Now let me hand over to Adam. He will bring this to life through a few examples.

Thank you, Mikhail. I'll be running through 3 demos in which I highlight the ways that our strategy of coverage, scale and efficiency is delivering value to our customers and truly bringing LSEG Everywhere. Our credit quant needs to back test a trading strategy as far back as the global financial crisis. Getting that coverage alone is a challenge, and onboarding data is traditionally slow and error-prone. With LSEG's AI-ready 20-plus year history across millions of securities and integrations with data and AI platforms like Databricks, they can get started building and back-testing training strategies in minutes, not days or weeks. Let's see how.

Here I am in the Databricks Genie AI Assistant. I'm looking at LSEG's historical analytics on government and corporate bonds, seamlessly shared to my account with Databricks Delta Sharing in minutes so that I don't need to spend days, building pipelines to ingest the data. I'm a credit analyst, and I'd like to understand the impact of the global financial crisis on the financial sectors in the U.S. and the EU so that I can back test my strategy. I'm going to prompt Genie AI's natural language interface and ask about option-adjusted spread, or OAS, a key indicator of credit risk during that time.

What was the median OAS from 2007 to 2013 for the financial sectors in the U.S. and EU? LSEG's AI-ready content is well structured for AI use cases and tuned for LLMs. And because of that, Genie is going to be able to answer that question in seconds rather than the hours that might have taken the analyst to write the code previously. You can see that the answer is already returning. First, a table and then a chart soon after. If you look at the chart, you can see that the U.S. had a large spike in OAS early in the global financial crisis. The EU had it there a couple of years later and for a longer stretch. From here, I can easily dig a little deeper, again, with natural language, to understand the specific subsectors that drove these spread movements and can even build correlations that will support risk models. We're generating fast, trusted insights powered by LSEG's AI-ready content, well-structured for LLMs.

Time to value for the content that our customers subscribe to from LSEG has never been faster. And customers who host their models on LSEG's infrastructure can automatically build the same rich histories and share them with their customers, just like we have here.

So what do we just see? Instant access to LSEG Analytics and Databricks and other data and AI platforms with no need to wrangle data, natural language queries, no coding skills needed, trusted results using LSEG's historical analytics on government and corporate bonds. This means customers can make faster, more confident decisions powered by the Analytics API, AI-ready in the customer's cloud, coverage, scale, efficiency.

Now that we've seen how we're enabling customers to interrogate our data with AI, let's head to our second scenario, in which we're enabling a customer to marry our data with theirs in their AI stack. A credit analyst needs to combine their internal portfolio data with LSEG's trusted analytics in their proprietary AI application to aid them in bond pricing activities. And they need to do it securely without writing code. We're seeing an increase in the use of AI across financial services. And LSEG's model context protocol or MCP server enables seamless integration of LSEG content and analytics into the customer's firm-wide AI solutions. Let me show you how.

In this demo, I'm using Anthropic's Claude as my MCP client, but these capabilities work everywhere. I'm a credit analyst, and I'd like to understand my portfolio's risk. So I'm using LSEG's MCP server to marry my internal data to the LSEG content for which I have a license. With a single click, I've connected to LSEG's MCP server so that I can analyze the risk of my portfolio. You can see all the different models that this user can connect to in this drop down. I'm going to start by asking a question about the yield of an individual security in my portfolio.

Now this is connecting to LSEG's Analytics API to retrieve precomputed results from last night. It's returning the price and yield in natural language without my writing a single line of code. Now I want to grab the spread as well, that same risk measure we looked at in Databricks, and it returns almost instantly, again, without a single line of code.

Finally, I'd like to run a scenario and see what would happen to the spread if the price changed. Now this is computing live using LSEG's Analytics API to return the comparison of the spread today versus yesterday. Everything you've seen is running against LSEG's Analytics API, optimized to answer LLM-powered queries and connecting to LSEG's accurate models. And we can extend this to customer models as well running on our infrastructure.

With MCP and LSEG's AI-ready content, customers can run deep analyses, joining their data to LSEG's, all without running a single line of code. And that's driving consumption across our content and our APIs. So what do we just see? A credit analyst subscribed to LSEG's content was able to seamlessly combine LSEG's analytics with their own internal data in their AI stack. Using LSEG's MCP server, they can retrieve results and run dynamic calculations, all without a single line of code. And they can do this with any of LSEG's models or models that LSEG hosts on behalf of our customers in whichever MCP client they are choosing to use. And we're doing this with every LSEG model across every asset class, answering millions of queries per day on our APIs and integrated directly into the customer's AI stack so they can move faster. It's the same 3 pillars in action: coverage, scale, efficiency.

Now that we've seen how we are enabling no-code integration with LSEG content and analytics, let's go to our third and final scenario in which we show how we've made it easier than ever to write code to leverage LSEG models.

A quant in the FX markets needs to routinely hedge an FX position. So they need to write reusable code that can be run automatically. They need to move quickly, but mastering the syntax to build and run models at scale is challenging and can lead to critical bugs that can have a huge impact on your company's bottom line. And that is why we built the LSEG Analytics Visual Studio Code Intelligent AI Assistant. Visual Studio Code is a preferred development environment for 74% of financial services firms. And we're making it easy for customers to build on top of our models with the power of AI. Let me show you how that works.

Here I am in Visual Studio Code, one of the most popular development environments in financial services. I'm looking at LSEG's AI coding assistant extension in the marketplace. FX markets move quickly, and I need a fast, scalable way to build an application that I can use to plan my hedges. I'm going to head over to LSEG's prompt template library. We have many templates optimized with the most common activities of financial services professionals. I'm going to grab one of the pre-canned natural language templates for pulling in an FX forward curve.

Well, it looks like 4 simple steps is actually a lot of code. But with the power of LSEG's AI assistant, I don't need to write that code myself. In seconds and using only natural language, I'll have code that can build a graph that I can use for my analysis. This would have taken hours or potentially days with painful debugging, reading of the documentation and calls to LSEG for technical support. I'm going to save this script and then click run. The results return in seconds, powered by real working code that I can deploy and it took me minutes, not days. And that's true for LSEG's models and for the models that customers host on our infrastructure. We're driving consumption of our APIs, and we're doing it by making writing production code easier than ever.

So what did we just see? An FX quant is able to use LSEG's Visual Studio Code AI coding assistant to build their FX hedging strategy in seconds. LSEG's prompt templates enable them to rapidly write new code without worrying about syntax or the right order of operations so they can focus on building value. The quant can do this across any of LSEG's powerful trusted models, all using natural language. And they have real working code that they can deploy to production. We're expediting strategy development across every model in LSEG's arsenal, accelerating production use cases on the back of the Analytics API, and it's happening where our customers write code.

LSEG Analytics is delivering on its 3 pillars. We are delivering coverage across hundreds of high-value cross-asset analytics models and sources. We are achieving enterprise scale through our high-performance, interactive Analytics API and we are improving efficiency by delivering AI-enabled analytics to customers through their most commonly used channels, like Databricks, Model Context Protocol, Visual Studio Code and LSEG Workspace. In short, we're bringing LSEG Everywhere, helping clients analyze faster, make more confident decisions and innovate at scale.

And now I'll turn it over to Nej D'Jelal from Workflows. Thank you.

Good afternoon, and welcome to the LSEG Workspace session. I'm Nej D'Jelal, Group Head of Workspace, LSEG's customer-facing flagship platform that serves over 350,000 users across the trade life cycle. Now building on what Ron and Gianluca shared earlier, our ambition is to be the leading provider of accurate, trusted data that underpins actionable insights for our customers. And Workspace is where that vision becomes reality for financial workflows.

Across the financial sector, professionals lose valuable time switching between systems and chasing data, inefficiencies that cost global institutions millions. And whilst AI brings speed, value comes from confident decisions and secure collaboration in one place. And this is where LSEG differentiates, bringing together actionable insights underpinned by the market's most comprehensive, trusted and accurate data, as mentioned by Todd and Emily earlier, all of which is delivered through AI capabilities in Workspace. Secondly, integrated workflows, enabled by Workspace working seamlessly with our customers' Microsoft tools. And thirdly, secure intercompany collaboration powered by Microsoft Teams and Open Directory.

The result is an unparalleled package deal of trusted and accurate insights embedded where work happens, driving better decisions and collaboration across the industry. Today, you will see 3 demos that bring this vision to life. Firstly, Workspace AI in action, integrated with Excel, PowerPoint, Teams and Open Directory. We'll also show you trading workflows where we've integrated Teams, Workspace and our analytics partner Tradefeedr. And finally, we'll showcase a proof of concept of Microsoft Copilot agents integrated with Workspace. These demos will show how our core enablers drive commercial impact, boosting license value, expanding reach and unlocking new revenue streams.

Let's introduce the demo. We start with a banker preparing a pitch for a private equity firm. Pitch books are notoriously time consuming, hours spent chasing data across systems switching between Excel and PowerPoint and manual formatting. And when you consider the tens of thousands of professionals that spend 15 to 30 hours a week on a single pitch book, the opportunity to accelerate that process, but without compromising trust or data accuracy unlocks millions in efficiency gains alone. Hence, our ability to bring together trusted and accurate insights, workflow integration and collaboration as one single package into our customers' tools means they can build and share pitch books faster and with more confidence than ever before.

And the commercial benefits for LSEG are clear. Even deeper workflow integration drives higher license value, leading to stronger retention and price uplift. Let's dive into the demo.

[Presentation]

So as you have just seen, Workspace isn't just a terminal. It's the entire financial workflow. We saw trusted insights, integrated workflows and collaboration through Microsoft Teams and Open Directory, delivering speed, confidence and collaboration for our customers. Commercially, this means higher retention, more usage and upsell opportunities. While embedding us deeper into the customers' environment, thereby expanding distribution and reach. Importantly, AI in Workspace and Open Directory are in beta pilots and the office add-in and Workspace app for Teams are both generally available for some data sets.

Now let's move on to the second demo. This time, we will show you how Workspace helps traders make faster and smarter execution decisions. In volatile markets, execution costs can make or break a trade. Yet traders often rely on fragmented data and manual processes that slow down essentially their overall experience and increase risk.

Working with a partner that specializes in trade performance analytics, Workspace brings everything together in the form of natural language queries like what's the best way to execute this order and embedded analytics and execution tickets in one single workflow and collaboration with liquidity providers via Open Directory. For traders, that means speed, accuracy and reduce risk. For LSEG, it means even deeper workflow integration, driving license value, increasing retention and price uplift.

Let's dive into the demo.

[Presentation]

So why is this different? Well, it's trusted analytics insights, it's industry-leading data accuracy, workflow integration across Workspace, Teams and third-party specialists and collaboration with liquidity providers via Microsoft Teams and Open Directory. For traders that means speed, accuracy and smarter decisions. For LSEG, it means higher license value, wider adoption through no-code analytics tools and partner monetization. That's how we turn trading complexity into a seamless value driving experience. And in terms of availability, the Teams app integration with Tradefeedr and Workspace is live with additional enhancements to come.

Now let's move on to the final demo. This time, we'll show you a proof-of-concept that we're working on with plans to release next year. Earlier, Tim showed how we can enable customers to create an analyst report from their own environment. Now let's look at another approach this time from the perspective of a Workspace user. In this instance, we are transforming the way analysts research and make decisions through Workspace integration with Microsoft Copilot agents. Today, buy-side analysts spend days pulling filings, news and their own internal notes, manually modeling scenarios in fragmented workflows. It's slow, error-prone and it delays investment decisions.

So by integrating Workspace with Copilot's researcher, analysts can pull filings, news and internal notes in seconds. They can build comprehensive reports using natural language prompts. And effectively, they move from research to ready in minutes, not days. For our customers, that means speed, confidence and end-to-end insights that combine their data with LSEG's accurate trusted data. For LSEG, it means even deeper integration into our customers' environments, driving license value, increasing retention and expanding distribution. Let's see it in action.

[Presentation]

Here again, we've shown why Workspace is different. Through our integration with Copilot, we have transformed a research workflow that delivers trusted insights in minutes, again, made possible by our three core enablers: accurate, trusted data, entitlement-aware, auditable and combined with customer data, integrated workflows inside Microsoft tools where analysts work and collaboration, which is enabled by preparing insight-ready analysis that can be shared using Open Directory.

For analysts, this means speed, confidence and better decisions. For LSEG, it means higher license value, broader adoption and upsell opportunities via premium AI features. As mentioned, the integration with researcher is currently a proof-of-concept, and we are planning to introduce this next year.

So as we wrap, let me first thank you for joining this session. I'll leave you with one thought. Why does Workspace stand apart? Because Workspace is more than a terminal. It's the entire financial workflow, a packaged deal of actionable insights, accurate and trusted data, integrated workflows and secure collaboration, all in one. It's designed to transform how financial professionals work and that transformation is already underway. Many capabilities are live today and others are advancing through beta pilots. And as we look ahead, we're excited to partner with our customers to shape the future of trusted and accurate financial workflows.

With that, I'll hand over to David for closing remarks. Thank you.

Thank you, Nej. I think that if we get a slide up there in just a moment, there we go. This slide sums up really well what you have seen over the last hour or so. Our D&A business is built on great data, extensive, trusted, accurate. And that is the foundation of everything that we do. Most investors have a very narrow direct experience of our products. It is typically just the left-hand side of this chart, where our data is vertically integrated with our own UI with Workspace.

But as you have seen this afternoon, that is just a small part of our reach, and our reach is expanding every week, whether it is combining tick history and machine-readable news in Google BigQuery or leveraging agents in Microsoft Copilot or doing advanced fixed income analytics via our analytics API or in Databricks, LSEG is everywhere.

Now for the rest of the afternoon, we are going to showcase some great innovations from across our other businesses. What you are about to see is just a small selection of our product portfolio, but it will give you a sense of how close we are to our customers, embedded in their workflows, responsive to their needs and building solutions that help them grow revenue, save costs and manage risk.

So your lanyard will give you your personalized journey for the next couple of hours as you rotate through the different rooms all on this floor or ask any of the hosts or the IR team if you get lost.

The breakout sessions will begin in about 20 minutes at 3:30. And then we will see you back here in the theater for Q&A at 5:15. Thank you very much.

[Break]

Okay. Shall we dive into some Q&A? First of all -- hold on a second. Okay. Thank you all for returning for the Q&A. We know we have thrown a lot of info at everyone today. And our intention was to really give you a lot of information without it being overwhelming. But we're looking forward also to this session as well in terms of taking your questions. And so with that, I saw that quick move on the first hand. Go ahead. And I think do we have microphones coming around. Why don't we come up here in the second row, please?

Thanks very much. It's Ian White, from Autonomous. Thanks for those presentations. Three from my side, please...

Can we actually, sorry, keep the questions to one question per person? We will try to get it around...

Okay. All right. I feel like -- take three for me in this Q&A. So I'll try again. Okay. For my one question then, please. In terms of the 2 sort of partnership products you've discussed with Microsoft today, I'm thinking about Open Directory and Copilot Studio. Can you talk us through what is the enduring advantage gained by LSEG relative to its data vendor competitors from its role in those partnerships specifically? I'm obviously thinking of the non-exclusivity of the partnership with Microsoft. So to kind of put simply, I can understand how Microsoft benefits from LSEG identifying areas where its products could be improved and sort of fine-tuning these solutions. But does LSEG's first mover advantage in those partnerships provide an enduring long-term edge? Can you talk us through some thoughts on that, please?

Ron, do you want to touch on the sort of strategic aspects of Open Directory and then maybe, Emily, if you want to touch on the Visual Studio Code.

Sure. You're right, Open Directory can be federated. But there's something really important that we need to leave you guys -- if you guys walk away with one thing around this. The automated domain management, the ADM tool, which is the ability for us to scale our communities and to be able to manage those communities, that is exclusively licensed to LSEG for financial services. And so yes, others could potentially build those over time. But we have exclusive licensing rights to that in financial services. So I think that's a tremendous advantage for us.

We also, through our messaging platform, already have a well-established large community. So that's yet another benefit for us to be able to leverage that set of communities and then populate our Open Directory and manage that. So that really is key from our perspective.

Then, Emily?

Yes. So on Copilot Studio. So when we bring in the MCP into Copilot Studio, there's a couple of pieces there. When we enable that, it's actually opening up additional use cases and because we're working so closely with Microsoft, not just in the context of Copilot Studio, but actually across that broader ecosystem, which we've been spending on time today, it is actually allowing us to build out entire customer workflows that actually are centered around our data. So it is very meaningful the way that, that was done.

The second thing is when we think about quality in terms of those customer agents and what's happening in terms of Copilot Studio, you heard [ MAP ] talk earlier in terms of the work that we're doing hand-in-hand, Irfan daily conversations. And we are really thinking very deeply, and we're working with customers directly and actually expressing how they think about the opportunities with agents, which is opening up those broader opportunities for us.

I would add one more part, too, which is still emerging. We bet early on Fabric, as you heard before, and Fabric is becoming quite prolific across a number of large accounts. And there's a lot of advantages in terms of co-mingling our data with customer data and advantages about faster integration there. So we expect that to be something to be an advantage for us going forward as well.

And I will -- it's a little hard to see people in the back, but I will make an effort to see if you wave from there as well. But go ahead, Ron.

On the LDAs that you mentioned, could you talk about how these partnerships with your customers are better structured to your -- what your partnerships look like before your agreements look like before, particularly on the pricing side as well, how that is structured differently. And...

Sorry, differently relative to kind of a regular relationship?

Yes. What those relationships look like before you set up the LDAs. And you mentioned 17% of the ASV are now LDAs. What's the aspiration there in 2, 3 years' time? What percentage would that look like?

So I'll answer your aspiration question and then maybe, Ron, if you want to touch on the differences. Look, we view this as sort of an organic development with the LDAs. We don't have a targeted level of we're aiming for X percent. We see them fitting very well with a number of the customer relationships, and Ron will talk about that in a moment. And we see -- as I mentioned in my remarks earlier, we see significant outperformance in those relationships beyond the perimeter of the LDAs because of the structure because we effectively become the default provider for those customers. And even if there is something that is outside the perimeter, we are often the natural first call.

Do you want to touch on some of the structural benefits?

Sure. And we get approached much more for that type of arrangement than we provide. So we have very specific criteria around how we engage with customers in that enterprise type of arrangement. And typically speaking, what we do is without getting into too much detail around it, is we understand where their consumption patterns are from our different services that are in scope of that agreement. And then we lay that out in terms of a joint customer value plan where we expect that growth to continue over that period of time for the agreement.

And we have now a very deep canon of specific cases where value can be unlocked, meaning they can find a competitive displacement, they can use additional services to drive top line or to drive bottom line efficiencies. And we go through that with them and we outline that case very specifically by each customer. And from that, we determine what the commercial relationship is going to be, and then we enter into that arrangement. So that's about as comfortable as I am about talking about the details around it.

But -- what happens is, as David said, we become the default provider and to achieve that synergy case because built in is, of course, our growth. So they are highly motivated to execute on those projects as are we to support them. And through that, we identify tangential opportunities. And as we develop new products, which are outside of that framework, we have a higher propensity to close those deals. And so that's generally led to an outperformance in those accounts relative to peer firms that aren't in that agreement and absolutely relative to what that arrangement was before. I hope that helps.

Anyone in the back there -- in the back row, excellent.

Thank you very much, David. Melwin from Sterling Investments. I think you reminded us today, what a lovely business you have created in the last 7 years that you've been here, David, so fantastic to you and congrats to the team. My question was actually not about LSE this evening. It's about creating LSE a platform for other great companies to list grow, come to the markets in terms of the LSE rather than [ LSEG ]. Any thoughts there in terms of giving momentum, encouraging dual listings, new IPOs, smaller A markets, et cetera?

Sure. So -- you would have heard from Charlie Walker earlier in terms of what's going on with the private securities market. That is just one of many different things that we are doing across our equities franchise. And the LSE itself has been the beneficiary of a huge amount of change over the last few years, where we have driven a bunch of that. We've worked with the government. We've worked with the FCA. AIM itself is going through a consultation right now in terms of what can be done to continue to improve on that. We've seen a significant uptick and benefits from a lot of those changes with the IPO market reopening over the last couple of months. And the pipeline looks very good.

So I think that the notion of breaking down that kind of bright line between public markets and private markets, everything that we're doing on digital market infrastructure, the continuing support for the companies that are already listed on the exchange. We're probably the best market in the world in terms of dual listings, in terms of good partnerships with exchanges in other parts of the world, whether that's Africa, Middle East, Asia, et cetera.

So I actually feel very good about all the progress that's being made in that area. There are some dynamics in the broader global market, whether it is the private equity that we've seen over the last 10, 15 years, a lot of the uncertainty in the U.K. market since Brexit was not particularly helpful, et cetera. But a lot of that is behind us. And I think the pipeline, as I said, looks very good, and a lot of the changes have been very productive. So thank you.

I'm going to go into the third row. So here we go.

It's Andy Lowe, from Citi. You've lent in hard throughout the presentations about your -- the benefit you have in terms of your data being AI-ready, lots of talk about consistency, scrubbing the data. Could you maybe explain a little bit more on that point? And with the ability for large language models to do more with unstructured data, why is that advantage that you have currently not eroded as peers are may be able to do that more easily?

Yes. Emily, you want to...

Yes, happy to. So when we stepped through the slide earlier, we stepped through certain steps. And it starts with sourcing and across a lot of different contributors and then also populated with our own proprietary data generation. Then we get into some of those steps that you were just referencing. Now these are really intricate steps, and they take a lot of deep expertise. And actually, a lot of that knowledge is actually encoded in something we call internally data models, which really describe relationships between data. So if I say to you that you have a bond with a par amount of 1,000 and a bond with a par amount of 100, you as a consumer of that data, that is a very undesirable effect unless you've then gone on to correct it, and it can cause a lot of problems. So it's a very simple example, but we can get further and further and the nuances in financial services are vast and deep. If I say country of risk versus country of issuer, that has completely different meaning and consequences.

Now LLMs can do some of that in terms of understanding the context, but not in the level of detail that we need to achieve the level of quality that our customers really require out of this content. So when we talk about AI-ready content, we're going further in terms of providing all of those semantics and detail like that, that allows for very confident use in the context of LLMs. That is multistep. And one of the other points that I made earlier is how much iterative cleaning goes into this as well. So you can't just take data out of the box. We really have to cleanse it to achieve the level of quality we want.

And then on top of that, make sure it's delivered with all of that history going back decades with the consistency. Why does that matter? Well, if I was -- I'll take an example of a quant and I wanted to build a signal, I want as much history as I possibly can and as much orthogonal information in that data to build the richest type of signal. That's why the breadth and depth of this data matters so much, but also that level of quality that goes into AI-ready content.

Let's see a question over here. Go ahead. We got a microphone coming to you.

Thanks. I mean you gave a lot of insight into the data, and I think you're very much an early leader there. Just in terms of -- in context of the multiyear agreements you have with partners and things like that, just really wanted to understand at what stage you'll be able to monetize that more aggressively? And do you think about that as you monetize that more aggressively in the consumption basis, how does that work with potentially usage, not just in Workspace, but just generally across practitioners going down? And can you ensure that that's a strong net positive for the business? And how are you positioning for that?

Yes. So let me just talk a little bit about usage and consumption-based pricing in general. Today, we already have that in 2 parts of our business. They're relatively small today. As we roll out more and more of this technology, we will be able to do it across much broader parts of the business. And so for example, if data is consumed through an MCP server, that is something that is very conducive to tracking usage and implementing consumption-based pricing if we want to. A lot of what we provide today does not have that kind of metering available. It's on a contractual basis.

And so as we go down this path, you'll see us moving both technologically but also operationally, financially down a path of being able to do more and more of that. So we will be doing this in a thoughtful way, in a careful way, in a way that incentivizes as much data consumption as possible. In other words, we don't want to disincentivize the data consumption with pricing that is too aggressive. This is not a new problem. In other words, other industries have gone through this before. So we'll certainly be learning from that and making sure that we are maximizing the customer usage of the data while at the same time, optimizing our pricing to maximize our revenue intake. But this is going to be a journey that we're going to be on over the next couple of years.

I don't know if there's anyone who wants to add to that. Okay.

You guys spoke about integrating Tradeweb further into Workspace. Can you talk about why now? What enabled that? And what opportunity you see from doing that?

Sure. I mean, fundamentally, it's not that complicated. As you all know, we have been getting closer and closer to Tradeweb, doing more and more in sort of a closely integrated manner. The big issue for us in terms of having Tradeweb come through our front end was the Eikon migration to Workspace. And it didn't make sense to spend time on that until Workspace was sort of fully migrated, established, embedded. It's now something that's, I would say, very close to the top of the list. And as I mentioned in my remarks earlier, I expect to see that access -- access to Tradeweb through Workspace in '26. I don't want to give a specific date at this point, but pretty comfortable with sort of the first half.

So does that help address that?

And the opportunity that you see...

I think the opportunity is consistent with the broader opportunity of Workspace as the front end to so many different parts of our portfolio and Tradeweb, in particular -- from a Tradeweb perspective, there's a lot that goes on in the market that is considered voice trading, and it's actually people chatting on Bloomberg and then executing. If they can move people off of that, and that includes taking advantage of Workspace, taking advantage of Workspace Open Directory, that's a big shift in terms of the ecosystem. So from their perspective, I think that's very interesting, very attractive. I'll let them speak about that specifically.

From our perspective, it is about making us that much more competitive in a critical asset class. And we're seeing the strength that we have had historically in FX play through across the life cycle and kind of the end-to-end offerings that we have to maintain or to build that kind of strength in fixed income is a very attractive proposition from our perspective.

Right, there, yes. I'm trying to mix it up in terms of the questions. Hopefully, we'll get to everybody.

So you mentioned how you have...

Can you just -- a quick intro of yourself?

I'm [ Shibani ], from BNP. My question is regarding Workspace. So you have these different avenues via which you're distributing the data. What incentivizes a customer to opt for Workspace over the other avenues that are there?

To offer the customer...

To opt for Workspace over the other avenues...

So what's the competitive attractiveness of Workspace?

Yes.

Ron, would do you like to take that?

We think it's great. I mean it's tailored to different communities. It has integrated workflows. It's now interoperable with Microsoft. We've got the Open Directory coming. We've got all our unparalleled data that is accessed through it. So we feel pretty strongly, and we're a very compelling case in terms of how we compete in the market.

Was your question specifically around competitiveness of Workspace? Or was it around an alternative to, say, like...

So like -- Sorry if I wasn't very clear. So like within -- like you also have MCPs via which you can provide your data and your customers can access it. So amongst all those avenues of distribution of data, what upside would Workspace have that customers would be inclined to install that instead of just having their own modules?

Okay. Okay. As opposed to just accessing through some other UI. So we want to meet customers wherever they want to be. Some customers want to have their own bespoke solutions. So many of our customers have their own user interfaces, and they just want to leverage our data as part of that -- as part of their own curated workflow, and we've always been open and fully support that as well, too. If customers just want -- or partners also take our feeds. And then in some cases, we co-sell with them, so we'll own the customer relationship.

One of our biggest relationships in that, for example, is Aladdin with BlackRock, where their platform is completely powered by our data front to back. And then when they bring on a new Aladdin client, we contract with that client directly and maintain that relationship. So for example, so we have those type of arrangements. And with AI or any other type of interface, it follows that pattern from our perspective. So if customers want to use that, that's their choice. We feel strongly that integrated workflows across everything that you've heard over today and especially in more complicated and regulated workflows in the trading environment, require all these other capabilities that are nontrivial to build. And so especially in those cases, we feel it's a very compelling option for customers. But if they feel like they have a different case and would like to use another route, we're happy to support that as well, too.

Two rows back from there. Two rows back, there you go. That's -- you just had your hand up. I can't see who it is from here, but person right -- 2 people in front of you. No, that row. I'm calling on people now. You have to have a question. If you don't have a question, that's okay. Who else? There you go right there.

Ben Krause at Wellington. As you think about training your own tools, and it's relevant for World-Check, but I think it's relevant across the business, using customer data to make your tools better and stickier. Like how has that discussion with customers evolved? And do you feel like that is a competitive advantage across LSEG's business?

Do you want to take that?

Sorry, your question was how do we train our model using customer data?

Using customer data and like are there any parts of the business where it sort of creates like a network effect. Whether it's World-Check or other...

Yes, I'll start, and others can chime in as well. So first of all, right now, we're not building any models ourselves, right? So we're not -- we don't think of ourselves in the model game. We're not building any frontier model. We're not a lab. The way we think about this is that our data is what our secret sauce is. So we do work with the LLMs. We do work with our scale partners. But what we want to be able to do is to be able to use those LLM, merge it with our data and allow our customers also to use their data and our data merge it and be able to build solutions. So that's the path we are on in terms of building our products.

And just -- and then maybe part of what the question you're asking. We do have the ability, for example, with our risk intelligence data because we have such a strong position in the marketplace, we can learn from the usage without having anything that is customer identifying. We can learn from the usage. So for example, if there is a particular person that regularly comes up with adverse media, but it's always wrong for some particular case. We can learn that. And we can learn it in this customer case, and we can apply it in this customer case without having any sharing of information across those. That kind of thing we absolutely can, but...

We actually -- in those cases, we are using that to make sure that our answers get better over time, but it's not a training to model. I was specifically trying to answer your training model question. We make our process better every day based on the answers we give to our customers.

Ben Bathurst, from RBC. As a company, I think you spoke 2 years ago about a $50 billion unvended opportunity. I wondered how much closer are you to monetizing that opportunity today? And how have your thoughts changed about how to address that, if at all, with respect to developments around Agentic AI?

Yes. I'm happy to take that one, and anyone should feel free to jump in. But you're referring to what we had talked about a couple of years ago is the opportunity in managed data services basically or managed data as a service. We still think that opportunity is out there. We still think it's a very attractive opportunity, but we have basically prioritized what we're doing in terms of all of the work to embed AI in our functionality, all the Agentic workflow that we're doing right now. It doesn't mean it's gone away, but this has just been a function of prioritization.

Right, second row here. Right in the middle.

Russell Quelch, from Rothschild. We sat here a few years ago now sort of asking the same questions. I'm going to ask you the same question as I asked 2 years ago, actually. The data analytics business was growing at 5%, then it's growing at 5% now. We've had a lot of ambition on the product side. I think it's well recognized that data is good. The technology is getting better. On the slide today, you've said the segment is growing at 5 or mid-single digits. You've changed that to a language-based target. But what is the...

Just to be clear on that. There was no intention to have any kind of signaling or change in terms of that page. And we have not made any change in our guidance or anything along those lines today. So I just want to be super clear on that.

That's my question, done. So I guess the question is...

Come up with one quickly...

Get on with it. What's the ambition? I mean is the ambition to grow at the segment rate? Is the ambition to grow above the segment rate? I'd love to hear you match your ambition on the product side with some ambitions on the financial targets.

Sure. So again, we're not giving any guidance today. We're not -- and there's been no change in terms of any of the guidance. But there is no shortage of ambition. And you have seen us, and MAP went through this in our opening remarks in terms of the consistent uptick in terms of the different parts of the business, including in the subscription-related businesses. We see opportunity here not just to grow at the segment rates. We see the opportunity here to, yes, grow at segment rates and take share and find new customers and address new TAMs.

So without putting any numbers out there, we look at all of our competitors in each of the highest performing segments of each of our competitors. And if they're growing faster than us, we try to figure out why and try to address what do we need to do to match or beat that level. So that's how we think about it. I can't give you a time frame. I can't give you particular guidance. But you see how we are investing in our product. You see the kind of change that we're driving. And hopefully, you can see from today the progress that we're making and the ambition that we do have.

Can we go 2 rows behind, Russell? Perfect.

Yes. Perfect. Yes. Shashwat from [ Landstar ]. So just speaking of ambition, on the Workspace side, do we think of cloud and sort of other LLM solutions as competitors? Obviously, I noticed the partnership last week, but just narrowly from Workspace, are they now competitors? And how is the product/technology gap in terms of meeting the same LLM features that they offer? How is that going, I guess?

Ron, do you want to take that?

Yes. So we do not view cloud to be competitive at this point in terms of the core workflows that we discussed, especially the highly regulated ones and orchestrated ones that are linked within training workflow. It's not even in the same category from our perspective. from a perspective of a UI that can present financial information that may apply to some customers that makes sense. I would say once the accuracy is addressed, if it's within someone's tolerance, then it's an option for a limited set of activity is how we would view it.

And from our perspective, having those kind of technologies out there and how we can potentially leverage that to help our own search within our own workspace, we feel that's an -- that's something we're also going to take advantage of.

So we know that our search currently is obviously highly accurate and works. And as you heard Nej present, we're building out our Workspace AI, and we'll roll that out when we are highly confident that it is highly accurate. And we believe that, that will be a great alternative for someone, especially when they want to have more integrated workflows. But if a customer wants to go a different route, we will support that as well through our data.

Can we get third row microphone here, please?

Can I just pick up on that point a little bit? We saw a number of examples today with, I think, 7 or 8 examples with prompt-based search and a very interesting analysis. Is it good enough yet? And could you maybe kind of qualify that answer with the kind of discussions that you're having with your key customers, mainly the banks and mainly around anything that's proximate to a trading engine? Is it good enough yet to be monetized effectively?

So is that an accuracy question?

Yes. It's okay.

Yes, yes.

Emily, do you want to take that?

Yes, happy to. So let's structure it in 2 parts. So when we think about accuracy, there's a couple of pieces to this. One, it's the accuracy and the underlying content. And one of the things we said earlier is that across those 3 distinct channels, so with Workspace, the proprietary experience with customer proprietary implementations and partners, we uphold that quality in respect to that underlying data. So that holds.

Now I want to divide this into 2 parts. One, when we're exposing data through the likes of MCP or otherwise into Databricks or otherwise, we're still providing that level of quality in the underlying content. When we look at Workspace as an experience, LSEG-led experience, what we're doing is we're taking an extra step. So not only are we making available that 33 petabytes data, which is quite a lot that can come through. We're also doing what Ron was mentioning in terms of that orchestration, AI-powered where makes sense. There's deterministic workflows like with FX where it doesn't always make sense. But where it does make sense, we're creating that interoperability.

And then the third piece is we go an extra mile with Workspace in terms of ensuring the level of quality in respect of someone's prompt, which I think is the basis of your question. So when we get a prompt in Workspace, we go now an extra mile in terms of making sure that the context of that is understood in the context of the data we're serving up. So we really are going an extra step in the presentation of accuracy in respect to Workspace.

Third row, right, in the middle.

It's Enrico Bolzoni, from JPMorgan. We're seeing some reports with eye-watering figures in terms of the CapEx that is expected to go in AI projects across the world. At your recent results, you stood by your guidance for CapEx for the next few years. So I just wanted to understand what gives you confidence that unexpectedly, you might need to actually increase your CapEx just to keep pace with the industry. And if this is not the case, is there a risk that the players that are indeed increasing their CapEx dramatically will try to pass on some of these additional costs to other market participants that are benefiting from it, but they're not doing itthemselves.

So MAP, do you want to take that?

So first of all, in order to square up the numbers, we are not in the business of buying chips for billions of dollars. So I mean, all this -- as Irfan was saying, we are not building an LLM. I think there are other people in there who are doing this very well, and we're using them. But on the CapEx side, I think -- so I think 2 things. The first thing is we're spending the double of our competitors. If you look at the FMI or if you look at the data provider, with 10% of CapEx, we are at the double. So that's number one.

Number two is -- in this, we went from 15 to 12 to 10, and we're going to high single digit next year. Actually, next year, it will not be a decrease of CapEx in terms of millions of pounds. It will be pretty much stable. And the important thing is that we've been investing to cover the technical debt that we have inherited from Refinitiv. And year after year, this investment in the 900-ish million of CapEx is becoming smaller and smaller. So I can't give you a figure because it's not a public figure. But what I can tell you, I can tell you is that it decreased '25 on '24, and it will decrease even more '26 on '25. So it's going to open up for us space to invest into growth and into AI or some other things by keeping our CapEx flat.

And Enrico, was there a second part of your question in terms of what others might be spending? I just want to make sure...

The question was related -- I appreciate you're not in the business of doing chips. But the -- I guess, was a bit more of a philosophical big picture type of question, which is clearly there are some players that are in these sort of businesses, and they are spending a huge amount of money for that. So I was wondering whether there is a risk that to some extent, this cost will cascade through the rest of the industry also across those players that are...

I think -- yes, I got it. Sorry. I think it's actually just the opposite. And what I mean by that is you have massive competition if there are sort of 3 legs of the AI ecosystem stool, it's compute, which is data centers and chips. It's the models, and it's the data. And you're seeing huge capital going into the models. You're seeing huge capital going into the data centers and the chips. That will end up -- there's enormous capital going in there. That's also enormous competition, and there will end up being more commoditization in those 2 legs of the stool.

In the data, you can't just throw money at it to create more data. You can create synthetic data, which is the quality of synthetic data is based on the quality of the underlying data. So fundamentally, there's sort of a defined universe of data. We have the best content set, the best data estate. And so we see that as being protected while those 2 other legs of the stool will, over time, be more commoditized.

There's a mic coming from behind you.

David, this is Nadim Rizk, from PineStone. The question is sort of more longer term. When I look at all these amazing AI tools and products that you showed us today, I can't help but wonder about the long-term employment in the financial industry or if you want to call it, number of seats and think that number of seats will eventually shrink because you'll be able to do so much more with so much less. And if that's the case, how would you think this would get reflected into your business?

So I'm happy to take a shot at that, and then if anyone should feel free, feel free to jump in. We have had this dynamic for a number of years. And in fact, in this theater, we gave our low single-digit guidance for the Workflows business because of -- before AI was such a prominent topic, there was a question as to electronification. Are we going to see more and more data growth, but relatively limited kind of human participation? That may be exacerbated by AI.

I think from our perspective, we are very well positioned to serve our customers, where they want to be served. And if that means a smaller number of humans doing a lot more, great. We're really well positioned for that. If that means more consumption of our data through our data and feeds, great. We're well positioned for that.

One other thing I just want to put out there, okay, this market is changing dramatically. There is an enormous amount of disruption going on in this market. It is not beyond the payout to think about the fact that some of our new customers may be agents, okay? So as there is a profusion of agents, think of each agent needing access to data. That could be a very different model from humans, agents, data and feeds.

And you've already heard a number of companies thinking about their HR function overseeing their people and their agents. You've heard about people putting agents on LinkedIn to be hired. And so again, lots changing. I don't want to tell you, hey, this is what we're charging per agent. But I think it might be overly simplistic to think, hey, people are going away. It's going to be all AI, and that's going to reduce the economics.

It's Bertie Thomson, from Brown Advisory. I enjoyed the conversation with Matt from Microsoft. But if we think back to 2023, David, you were sort of quite confident that we'll see a material contribution to revenue from the partnership in 2025, which sounds like it might have been pushed out a bit. Do we still expect a material contribution from the partnership? And if so, when should we expect to see it?

So absolutely is the answer, but it's also a, I'll say, a steady upward trajectory as opposed to a spike up. And we've gotten this -- or we've had this conversation with a number of you, and I think it was touched on during the course of today's conversation. Part of our business is that there tends to be gradual adoption. And I think I've said on a few occasions, we could introduce the most amazing product in the world tomorrow, and it would still take a few quarters for us to see a fairly gradual uptick in that. I think that's just -- it's the nature of this industry, risk averse big customers, periods of adoption that can take a while.

So we feel very confident about the upside in terms of the revenue generation. And one thing that we have talked about, you've seen our analytics business growth double over this past year. That's the one of the 3 businesses within data and analytics. That's the one area where you do see sort of meaningful in-year sales or in-year adoption as opposed to the other areas, which tend to be longer-term contractual or subscription revenue growth that picks up over time, and we're often displacing other competitors.

It's Hubert Lam, from Bank of America. So a question on your data. I know you pride yourself on the breadth and depth of your data, the proprietary nature of it as well as the petabytes of data that you have. Can you talk about how concentrated is the use of the data? Like how broad-based are the users using across all your data sets? Is there like some sort of 80-20 rule, where 80% of the people only use 20% of the data you have? Or is it more broad-based than that?

I don't know -- anyone want to put a hand up on that one?

It's contextual by community, right, I think, is the answer. So like in investment management, like we don't see a big demand for our real-time data, right, for example, or in the quant business, they clearly want a long history. So it really kind of just depends on which community and which use case that they're focused on. And then we do see a diversity within that.

I don't know, Gianluca, if you'd like to add anything to that?

I also think that it depends on the firm type. So there are some firms which historically are more in the DNA to take massive amount of raw data and analyze the data from quant perspective and so on because they think they can create a competitive advantage on that. There is also a trend from clients who wanted to see more actionable insights and information. So perhaps us doing some of the worker source, but it really depends on the client type.

But it's not...

It's rightly distributed. Hand it, to you, right, there.

Mike Werner, here from UBS. Just a question. You were talking about maybe agents, right, as a potential customer base. Do you see -- or is there a demand from the client base today to potentially unbundle some of your pricing, for example, within workflows? I mean you're adding all these incremental enhancements and opportunities, but do you see a world where you just price that data, not through a data feed, but through, say, an MCP server or something like that. Is that something that you're seeing from clients today?

Not meaningfully at this point. And it gets back a little bit to the question we had over here about consumption-based pricing. We really like the subscription model. We think that's a great model. And so even when we are fully capable of having usage-based pricing, we're not going down the path of saying, "Hey, if you want this little piece of data, over to you, you can pay a small amount for that." We want to maintain the subscription model and then within that, have the usage-based bands and the consumption pricing within that. But at this point, we're seeing -- and I'll combine those last 2 questions. We're seeing this shift to sort of model consumption. enabling our customers to access a much more significant amount of our data as opposed to wanting a little bit here or there.

I don't know, anyone want to add anything?

I'll just emphasize that last point. Traditionally, when people implement the data, they would typically focus on a set of data. That's not the way that models work. Actually, what models do is they encourage a left and light, and you don't need to look up a catalog to say, I want that additional and then you don't need to speak to your technology team to then integrate it. It is there, it's ready, and it's accessible, and the models know where to look.

So that, together with the fact that we've got a lot of relationships between these data sets means that when you ask your first question, it's actually going to really bring back a relevant set of data. And where you've got providers like LSEG providing that breadth of content, it becomes extremely powerful in providing very competent responses back.

I have just checked with the boss here, Peregrine, and we are at time, but we are going to adjourn from here to have some drinks. And of course, we are happy to -- we'll all be there. We're happy to continue having the conversation from there and continue to answer your questions.

So again, thank you very much for your time today. Thank you for joining us. Thanks to those who joined us online. We really appreciate the interest, and we hope you have found it to be an interesting and worthwhile day. Thanks a lot.

Good morning, and welcome to the investor and analyst call for LSEG's Third Quarter 2025 Trading Update. [Operator Instructions] I would like to remind all participants that this call is being recorded.

I will now hand over to David Schwimmer, CEO of LSEG, to open the presentation. Please go ahead.

Good morning, everyone. Thanks for joining the call. I'm here with MAP and Peregrine as usual, and we are also joined by Daniel Maguire, our Head of Markets, to talk about the Post Trade transaction that we announced this morning.

For this quarter, we're going to take a slightly different approach from a normal Q3 given the intense debate in recent months around our business and AI. I'll cover some key aspects of our AI strategy and the excitement we have about the current opportunities, before MAP goes through the Q3 numbers, and Dan covers the Post Trade transaction. Then, of course, we'll be happy to take your questions.

It has been a really busy quarter with great progress on several fronts. Group organic growth continues to be very healthy at 6.4%, with D&A growing at 4.9%, similar to the first half. ASV growth came in at 5.6%, a little better than expected, and we anticipate it being better again in Q4. We're raising our margin guidance to the top of the original range at around 100 basis points of improvement, reflecting strong operating leverage and cost control.

As you may have seen, we've launched a number of AI-related partnerships involving our data, which is valued and relied on by partners old and new as industry standard. We've announced an important transaction today that creates a strong partnership and aligned incentives for the adoption of Post Trade Solutions while also increasing our revenue share from SwapClear and extending the profit-sharing arrangement with our partner banks by 10 years. More on this in a few minutes.

And on the share buyback that we announced at our half year results, the original intention was to complete that by mid-December, but we've taken advantage of a lower share price and accelerated the GBP 1 billion buyback to finish by the end of this month. And we're today announcing a further GBP 1 billion buyback to be completed by our full year results in February of next year. Our strong cash generation gives us the firepower and the flexibility to invest organically to make important strategic moves and to be active in returning cash to our shareholders.

On the next slide, we have summarized our LSEG Everywhere AI strategy under 3 key pillars: trusted data, transformative products and intelligent enterprise. We'll talk more about those second 2 at the Innovation Forum in November. But let me take a minute or 2 to dive into our data and the critical and valuable role it plays now and will play in an AI-rich world.

The easiest way to think about our data is that the content itself and access to it is effectively financial markets infrastructure, something we know a lot about. It is industry standard, deeply trusted, embedded in highly regulated customer workflows and supported by processes and infrastructure that are extremely hard to replicate. And we are and always have been open. We deliver data to wherever our customers want it, their screens, their servers, their cloud and, of course, through third-party providers. Let's unpack this over the next few slides.

Data & Feeds accounts for a little over 1/5 of group revenues. On this slide, we've broken down these by data type. But before we get into that, I want to remind you of the scale of our data. It is the largest pool in the industry, both in terms of breadth and depth. We have over 33 petabytes of data. That is over 3x the so-called common crawl, the data set formed from the public Internet, which is used to train many LLMs.

Let's begin with the 45% of our Data & Feeds revenue derived from real time. This is a business built on physics, not probability. We've built connections to 575 exchanges and execution venues globally with our own infrastructure. In the blink of an eye, we standardize and translate the exchange outputs into a single common language and deliver them directly into the world's financial institutions. Millions of hard facts per second, not probabilistic algorithms. In a nutshell, AI cannot replicate or replace our real-time data.

Then we have 25% of our Data & Feeds revenue, which is specialized and enhanced by our own enrichment. By specialized, we mean proprietary. Think Tradeweb fixed income pricing or exclusive like the Reuters News agreement or contributed like our deals database. So an LLM could not access these data sets through public sources. And then on top of that, we are enriching this data with value-added enhancements and augmentation by our data experts. That is our additional value add. And then that all comes with the LSEG curation standards, accuracy, normalization and tagging. So think of this data as protected by 3 moats. It is either proprietary or exclusive. It is enriched by our own intellectual property, and it is curated, applying the LSEG standards, which have often become the industry standard.

Let me give you an example to bring this to life. Our deals league tables are highly valuable to banks, advisers and law firms. These league tables are widely considered the industry standard with LSEG data obtained daily from thousands of sources co-mingled with data sourced from nearly 2,000 financial and legal advisers actively contributing their deal flow. We get up to 25,000 of these contributions per month. This input, which is from humans, is crucial to the quality, accuracy and completeness of this data. These contributions clarify and correct deal details that appear in the press. They also add additional information to public deals and supply information on other deals that are not reported anywhere. So a data set built solely on public disclosures would be both inaccurate and incomplete.

We further enrich this data with our proprietary calculation of rank value, which sets the standard for deal comps, market share and pitchbooks around the world. We refine this methodology each year through roundtables with advisory firms. So in case anyone is missing the point, no LLM can gather this data from public sources, 3 moats, LSEG proprietary or exclusive data, enriched by LSEG IP and curated by LSEG, applying the LSEG standards.

Let's move on to the next bucket, representing 10% of Data & Feeds revenues. It is almost exactly identical to the previous bucket. It is specialized data, proprietary, exclusive or contributed, with LSEG standards applied. So not accessible by an LLM through public sources, our aftermarket research, for example. And to carry on the analogy with the moats, this is data protected by 2 powerful moats.

Next is another 10% of revenue from data that is indeed public, but to which we apply our enrichment and analysis, similar to what I was talking about with customer contributions on the league tables. And we also applied the LSEG curation standards. Examples here would be earnings estimates and sentiment analytics applied to earnings calls and other sources. So can an LLM access it? Yes, but the data will be incomplete. Here, it is 2 moats applied on public data. So 90% of our revenue is from data that is nonreplicable by an LLM.

That leaves us with the last 10% of Data & Feeds revenue, which represents the data derived from public sources for which we apply LSEG curation standards, data like company filings or economic metrics. This data is rarely sold on a stand-alone basis. Here, there is still one moat, a powerful and important one, and that is our standards, which I will cover on the next slide.

Now that we've established that 90% of Data & Feeds revenue is from data that is simply out of reach or inaccessible to an AI model trawling for public data. Let me take a minute to explain very concretely what I mean by that third moat, the LSEG data curation standards. There are 5 major processes in the curation of LSEG's high-quality trusted data, which are simply nonnegotiable for our customers in regulated activities. These 5 processes are the foundations of what we call the LSEG standards. Let's look at them in a little bit more detail.

We do not build our data sets on probabilistic models. We have constructed them from decades of hard data, much of which is no longer retrievable. We source them from our customer community with over 40,000 customers contributing regularly. And in many cases, our own analysts and experts generate them internally. So that is sourcing. We then extensively cleanse and validate this data to ensure quality, for example, verifying its accuracy and completeness. Publicly sourced data is not reliable without this step.

The third step, normalizing and mastering means creating a single source of the truth, consistent from year-to-year and from security to security, factoring in corporate actions, for example, or restatements or perimeter changes. And then concordance and tagging, which is a critical and differentiated step. This is where the universal symbology of the RIC or Reuters Instrument Codes and our use of perm IDs to tag each piece of data are so powerful. They allow full interoperability across the data estate and create logical semantic relationships between related data, for example, between a company and its directors or a bond it has issued. And the fifth step, distribution. Irrespective of technology platform, data format or channel, the data we distribute to customers is consistent and authoritative. I'll talk more about our distribution strategy in a couple of minutes.

So to summarize, for those who think AI models can scoop up so-called public data from the Internet and displace us, that just does not reflect how this industry works and fundamentally ignores the nonreplicable nature of the vast majority of our data. There's also been a lot of focus on our Workflows business. We have driven a lot of change here over the last 4 years and now have our customers on a modern, modular, customizable platform where we enhance functionality week in and week out, and we're doing more and more.

As we said at H1, it is not AI or a desktop. It is AI in the desktop, fully embedded in financial markets workflow. Workspace is now integrated with Microsoft Teams. We'll be launching Open Directory in the coming weeks and the full Workspace AI platform in the first half of '26, with Agentic tools coming as well. You'll see all of this at the Innovation Forum in a couple of weeks.

So let's look at our Workflows revenue, the same way we did for Data & Feeds. 50% of workflows revenue comes from traders who are deeply engaged with the platform to execute their roles. They need real-time data, a network community and integration with a range of pre- and post-trade tools. Further 20% of Workflows revenue comes from ancillary trading services, such as trade routing and order execution and management. Another 15% comes from investment banking, where we have specialized content across deals, corporate actions and research, as well as integrated productivity tools. That leaves 5% of Workflows revenue from wealth and 10% from investment management.

These customers benefit from our unrivaled data, exclusive Reuters News and portfolio analytics. But in these groups, there are lighter users who are mainly doing desktop research and basic charting, perhaps like many people on this call. Whether someone is a power user deep in trading workflow or a lighter user, all Workspace users will benefit from the significant AI and collaboration enhancements coming over the next few months. They will have the full functionality of some of the newer applications out there, but embedded in their existing workflow and based on data they can trust.

Now over the last couple of months, you can see the pace of execution on LSEG Everywhere, delivering our data to where our customers are working as the partner of choice for financial markets data. This is no change in strategy. We have long provided data to and distributed data through our customers -- I'm sorry, our competitors and partners. For example, we are the #1 data provider to Aladdin. The industry now has new entrants, building new applications and functionality, which we believe can expand our reach and drive additional consumption of our trusted high-quality data. The economics of these deals support our growth aspirations through data licensing, new channels and the potential for usage-based revenue over time.

Rogo is a specialist provider of applications to investment banking and private equity. Customers with Workspace licenses can access certain LSEG data sets through Rogo. The construct with Databricks is similar. These are attractive new distribution channels for our data. Just last week, we took a major step forward in our partnership with Microsoft, introducing certain data sets into Copilot for any Copilot subscriber, and more valuable data sets, both into Copilot and Copilot Studio for LSEG licensees. This will allow customers to build their own agents working with our data. You should expect the list of partners to continue to grow as we look to distribute our data through other major channels. That's the fundamental premise of LSEG Everywhere. A key part of many of these partnerships has been our ongoing build-out of MCP servers as we make more and more data sets available over time.

Before I hand over to MAP, it has also been a very busy quarter in other parts of our business. Just to highlight a couple of significant developments. With Microsoft, we have fully replatformed our trade routing network, Autex, in Azure with Autex now connecting 1,600 brokers and asset managers via the cloud. As a result, it's faster, has much greater capacity and is even more resilient. And we have executed the first transaction on our Digital Markets Infrastructure, which is positioned to become an important new capability for trading and settlement. We're preparing to launch our Private Securities Market. More on that at the Innovation Forum. And in Risk Intelligence, we have launched World-Check On Demand with all our critical data and insight now updated in real time.

That takes me appropriately to our innovation forum in a couple of weeks. In the first part of the event, MAP and I will cover our unique positioning, our end markets and execution to date. Irfan Hussain, our CIO; and Emily Prince, our Head of AI, will cover our AI strategy and engineering transformation. And then Ron Lefferts and Gianluca Biagini will talk about product strategy and monetization in DNA. We'll then have specific product walk-throughs and demos across the group. We're looking forward to showing you both the present and the future. And just to be clear, this is not a traditional Capital Markets Day. Don't expect any new guidance or anything along those lines.

So with that, let me hand it over to MAP to talk about our Q3 performance in more detail.

Thanks, David. So just a few words on our financial performance. We have delivered another quarter of strong growth across the group. Organic growth for the quarter was 6.4% with all divisions contributing well. We had a benefit of 30 bps from the ICD acquisition of last year and a headwind of 190 bps from FX, which together translates into our reported growth of 4.8%. Within D&A growth of 4.9%, Workflows and Data & Feeds saw very similar growth to Q2 with only a slight impact from the new UBS contract that I mentioned at the H1 results.

Analytics continued to grow strongly. The competitive environment is stable, and we are excited about the product pipeline. Our expectation for pricing into 2026 is for the yield to be similar to the last 3 years in the 3.5% range. FTSE Russell, as I indicated at H1, saw slightly slower growth in subscriptions with fewer account reviews in the period. But on the other hand, asset-based fee growth was strong as we lap the loss of a contract last year. Risk Intelligence had another strong quarter, driven by both World-Check and Digital Identity & Fraud. So overall, the subscription businesses delivered 6.5% growth in Q3, ahead of our expectation of 6% for the second half of the year.

ASV growth came in at 5.6%, a bit ahead of the 5.4% we had anticipated. Good sales momentum partially offset the expected impact of the final Credit Suisse impact wrapped into the new long-term partnership with UBS. As I have said before, I expect this to pick up again to 5.8% as we exit the year. The Markets business continued to grow well, though at a slightly slower pace than H1 as volatility was lower and comps got tougher. Looking at the 2 main lines, OTC derivative was up 9.2%, driven by continued strength in client clearing volumes in SwapClear, and fixed income was up 9.9% as Tradeweb continued to drive growth through its innovative trading protocols and an uncertain macroeconomic outlook. Elsewhere, we have seen the IPO pipeline pick up in the Equities business with more to come heading into 2026. And we are seeing the final headwinds to growth in Securities & Reporting from the Euronext exit.

Moving now to our delivery against guidance. We are absolutely on track and in some respects, ahead of our original plan. Year-to-date organic growth is 7.3%, comfortably within our guidance range, and this remains unchanged. On margin, the natural operating leverage in our business gives us confidence to raise our margin guidance to the top of the range at around 100 bps improvement year-on-year. This is a big step-up for a GBP 9 billion revenue business, and it factors significant ongoing investment in AI and new products.

We are very confident of hitting our 2026 guidance of 250 bps over 3 years, taking us to 50% plus, obviously, before the impact of the Post Trade transaction, which I will cover in a moment. On CapEx, we will invest at a rate of 10% of revenue this year as planned and expect that intensity to come down in future years. One or 2 in the market have asked whether we will need to invest more in an AI future. The answer is clearly no. We have been investing at a double-digit CapEx intensity for several years, and we are now switching the mix over time from technology debt payback towards more investment for growth, obviously, including AI. And finally, we have good visibility of hitting our free cash flow guidance of at least GBP 2.4 billion.

And finally, let's look at how we are allocating this cash flow. Overall, we are deploying more this year than what we are generating. That reflects the opportunities we see in front of us. So we expect to spend around GBP 3.5 billion versus free cash flow of GBP 2.4 billion. We are financing the difference with new borrowings of GBP 1.1 billion. Total dividends for the year are just over GBP 700 million, representing a 35% payout of adjusted earnings. In addition, we are deploying GBP 700 million net on the Post Trade transaction announced today, where we expect returns to be very attractive.

And finally, as David mentioned, you may have noticed that over recent weeks, we significantly accelerated the GBP 1 billion buyback announced with the H1 results, and we have nearly completed it. Given our strong cash generation, low leverage and the enhanced returns we believe we will generate at this share price level, we are today committing to a further GBP 1 billion. This will start shortly and complete by the full year result in February 2026. We plan to execute GBP 500 million of this GBP 1 billion in year. This is a further demonstration of the flexibility and optionality our strong cash flow generation gives us and our very active capital allocation decision-making. Taking all this together, our leverage at the end of this year should be around 1.9x EBITDA, so in the middle of our 1.5x to 2.5x net debt-to-EBITDA range.

Let's now look at the rationale of the transaction in our Post Trade business that we announced this morning. First, a group of 11 leading global banks is taking a 20% stake in our Post Trade Solutions business. The perimeter of PTS includes the recent acquisition, Quantile and Acadia, plus businesses we have grown organically, mainly SwapAgent. This transaction deepens our partnership with institutions that can benefit significantly from PTS services and allows them to help share its future and share in its growth.

Second, we have agreed to alter the terms of the revenue share paid to the partner banks from SwapClear. Historically and up to 2024, this sat at 30%, reflected in our cost of sales. We are taking this down to 15% for 2025, applied across the whole year and 10% for 2026 and beyond. And finally, we are extending it from 2035 to 2045. Again, this is strategically important, and it improves our economics at a fair valuation and extends the deep relationship with our partner banks into the long term.

Daniel will cover the strategic value in more detail in a moment. But the financial effects of this transaction are very positive. The impact of reducing the revenue share from 30% to 15%, which again is retroactive across the whole of 2025, will add around 250 bps to the Markets' divisional EBITDA margin and 100 bps to the group margin this year. While obviously, there are some financing costs, overall, this transaction is 2% to 3% accretive to EPS this year onwards. But beyond these financials and even more importantly, we expect this transaction to accelerate the long-term growth in PTS.

Let me hand over to Daniel to recap on the playbook that has been so successful.

Thank you, MAP. So I just want to take a couple of minutes now to highlight how and why SwapClear has grown over the last 15 years, and touch on the opportunity we see forward in Post Trade Solutions. So through partnership, both through the shareholdings a number of our key members have held in LCH and the revenue share in SwapClear that continues, we have built a deep and wide global network that delivers significant value to all of its constituents.

The scale shift in 15 years is extraordinary. The number of members, i.e., the banks has increased by 3.5x and the number of clients, i.e., the buy-side firms has increased by 200-fold, clearly demonstrating the network effect. Notional value registered per annum is up 10x at nearly GBP 2,000 trillion. And we have become the global destination of choice for interest rate swaps in all currencies for clearing. And this is why we are now inviting our partners into Post Trade Solutions, because we believe we can do the same again, but for the uncleared market.

We built a near GBP 1 billion annual revenue business based on cleared OTC instruments across SwapClear, ForexClear and CDSClear, all of which are leaders in their markets and all of which are built on the strong foundations and the model of industry partnership. The uncleared opportunity is basically the same size as the cleared space. Our members and our clients want to manage the whole book in one place, bringing efficiency to their capital, the margin requirements and materially simplifying and standardizing processes. We are uniquely placed to do that given the assets that we've built and brought together under one roof and with our proven track record of delivering real value through long-term partnership.

Acadia and Quantile give us collateral and margin workflow tools and compression tools, respectively. And SwapAgent and TradeAgent, both developed in-house, complete the current suite of services we call Post Trade Solutions. And we've got very good momentum to build on. Revenue in PTS is growing at double-digit pace. Volumes are up 70%, and the network is expanding at pace. So bringing these 11 major partners closer and giving them a role in shaping the business as well as a share in its growth sets us up for long-term success.

I'll now hand back to David.

Thanks, Dan. So just to recap, we have had another strong quarter of growth with year-to-date organic growth at 7.3% and all of our businesses performing well. We're executing at pace on our AI strategy of LSEG Everywhere as the AI partner of choice for financial markets data. And we are allocating capital effectively and proactively with an attractive strategic deal in Post Trade and a further big step-up in our buyback program. And now MAP, Dan, and I are happy to take your questions. Peregrine?

Thanks, David. [Operator Instructions] And with that, I'll hand over to Pauly to manage the queue.

Thank you, Peregrine. [Operator Instructions] And your first question comes from the line of Arnaud Giblat of BNP Paribas.

Could I start with the Post Trade Solutions? So banks are paying over 50x EBITDA, 9x sales for their stake. Clearly, as you said, that comes with a significant commitment to put more business through that division. I'm just wondering, I mean, you gave a bit of detail, but if you could flesh out a bit more what sort of commitments, the time frames, what specific milestones we should be looking at for that business to grow, and what perhaps give us an indication of the potential size of that business in the medium term, from a revenue perspective?

And my follow-up would be on the distribution agreements with third-party providers. Quite a lot going on there. I'm just wondering how we should think about this? Because clearly, there is a bit of a usage model you've talked about. So probably this increases significant usage and therefore, gives revenue upside. At the same time, if clients are accessing your data through a third-party vendor, then how does pricing in the long term look like if you're being -- I mean, if it interfaces somebody else?

Thanks, Arnaud. Let me turn it over to Dan to answer the aspects of your first question. We're not going to get into a lot of detail on what the revenue looks like over the medium or longer term, but you can talk a little bit about how we're thinking about the construct. And then I'm happy to talk about the distribution agreements.

Okay. Yes. Thanks, Arnaud. Look, we're very strong believers in the industry partnership model, as you know. We've been using that, building that for a number of years on different services, and I think you can see the outcomes of that. Ultimately, we build core critical infrastructure for our major customers here over a long-term basis and around the basis of trust. So we're very, very pleased that we've got our major partners around the table with us and aligned not just on economics, but also on the product road map, the governance and the product adoption, of which we have a pretty high rate of adoption for all the products we build because of this model.

I can't really be drawn on revenues. What I can point to is when you look at the -- which we shared in the slide that the gross market values, which essentially is a proxy for the scale of market risk and derivatives, if you look at the -- these numbers come from the BIS independent annual surveys, the gross market value is about [ USD 17.6 trillion ] and just over half of that is in the cleared space, but over half of that is in the uncleared space. So if you think about the level of risk of derivatives being transacted and risk transferred, they are very similar size. So we see the size of this opportunity very similarly as a result of that.

And then in terms of milestones, we've got, as you can see from the press release, 11 major firms and important people at those firms making clear commitments to work with us to build out and deliver and adopt those services. So I can't be drawn on specific road maps and revenues today, but very confident that we've got the right support from the right firms and the right people. And the network is much bigger than those 11, and we've already got very good momentum in that. So pretty confident on that.

And then your question around these partnerships or distribution arrangements. And the first point to make is that we've been doing this for years. And we have been providing our data through partners, and in some cases, as I mentioned, competitors for many, many years. And it's key when we do that, and this is a practice that we will, of course, maintain is that we protect our own relationships with our customers. And so in these kinds of partnerships, basically, the way they work is that although the initial origination of the relationship might come through one of the partners, the customer is then directed to us to establish the direct customer relationship with us. And we do that in a number of different situations and circumstances. So that protects us from being disintermediated through these kinds of arrangements.

The other really important aspect that we're very focused on in these kinds of partnerships and distribution arrangements is protecting our data and making sure that our rights, our IP are protected even through any of these distribution channels. So obviously, the AI world is a little bit different, but we're still in a position to protect our data.

And let me just give you one specific, I'll say, technical example. When we're distributing our data through an MCP server, because of that construct, we can control and monitor the access to our data. So in that construct, we're not at risk of a customer downloading all of our data, training their models on our data and then not needing us anymore. This MCP server construct allows us to control that in a very successful manner. So maintaining the relationship, protecting our data and data integrity, these are the kinds of relationships that we have managed very successfully for a long time, and it's great to see these new entrants and these new ecosystems, because we think it will actually expand the market and the customer base that we will be able to access our data. So we're really looking forward to this and excited about it.

Your next question comes from the line of Andrew Lowe at Citi.

Thanks very much for the color on the revenue split by product in Workflow and Feeds. My question is on the Data & Feeds business. Specifically, how much of the historical revenue growth has been driven by pricing versus volume? Could you please also comment on the historical pricing trends across these different groups? So for example, it would be great to know how pricing growth in real-time data compares to the other segments, including the 10% from public data sources. And it would be great if we could hear a bit more about how much visibility you have on future pricing? And I've got a follow-up, but I'll wait until you've answered.

Yes. Thanks, Andrew. So I'm not going to break it down product by product. But as we've been pretty clear over the last few years, you've seen our pricing yield on an annual basis be in that sort of 3% to 3.5% zone. And then you've seen our Data & Feeds business grow usually more than twice that. So that gives you a sense of what's going on here in terms of pricing relative to just volume growth. And we've been doing a lot of innovation in this area as well in terms of new products, new distribution channels as well. But hopefully, that gives you a sense on that.

Great. Okay. And then as maybe a follow-up to that. So are you seeing a pickup in demand for your tick history now that you've got sort of LLMs which are cheaper and more widespread? And how important is that when you're sort of selling your forward-looking real-time pricing data?

So interesting question. and tick history, for everyone's benefit, is a great data set that we have that goes back to the '90s and has tick-by-tick history for millions and millions of securities and no one else has it. It was all public data when it was released by the exchanges, but we are the only ones who have stored it, maintained it and made it easily consumable. I would say the technological changes make it easier to consume and access now than it has been over the last 20-plus years. And we certainly expect to continue to see it being a very valuable content set. Historically, it has been mostly used by quant shops back testing their algorithms. But your question is a good one in terms of recognizing that with these models, you could see a lot more potential users accessing this huge data set to look for historical correlations and help that inform their trading on a go-forward basis.

Your next question is from the line of Russell Quelch of Rothschild.

I'd also like to focus these questions on the Data & Feeds business. Thanks for the extra disclosure on the revenue breakdown. So you disclosed that 55% of the Data & Feeds revenues come from pricing and reference services. And I believe you've gone from #6 player there to #3 player in the last couple of years, just behind ICE and Bloomberg. So my questions are, firstly, number one, how have you done that? And what's your view on the main points of differentiation in your offering, which is helping you to take share?

My second question is, do you believe you can be a #2 player here? And if so, how? And the third question is a bit of a follow-on from Arnaud's question, but asked in a bit more of a direct way. Can you talk to your expectations of the size and cadence of the growth uplift from the recent and future data distribution partnerships that you mentioned relating to LSEG Everywhere?

Sorry, can you say the third part again?

Yes. Sorry, a bit of a mouthful. So I was thinking about the data distribution partnerships relating to LSEG Everywhere, both the current ones you disclosed and then you said about future partnerships. So I was wondering how we should think about the size and the cadence of the growth uplift that comes from those partnerships, both the ones that have been announced and potential future ones.

Got it. Okay. So your first question, how have we moved from #6 to #3. It is investing in our content and investing in our distribution. And you have seen us over the last few years do a number of, I would say, pretty significant steps in a number of different areas. So for example, when we took on the Refinitiv business several years ago, it was very clear to us that, for example, talking to customers, they made it clear, fixed income evaluated pricing was a weak area.

Corporate actions was a weak area. We have invested meaningfully in both of those areas and addressed those gaps, and we're now highly competitive in those areas. And so that has helped us move up the ranks. We have added new content in terms of a number of different areas, ranging from -- I guess, a good example is our inclusion of Dow Jones content alongside our exclusive Reuters News alongside thousands of other news sources. So constantly investing in content in a number of different areas.

And then on the distribution side, over the last few years, we have made our content available through a number of different distribution channels. And whether that's in different cloud providers, whether that is -- there are some of our data sets, for example, they were only available in the U.S. for technology reasons. And we have now made those available on a global basis. So it's a number of things like that. But really, if I boil it down, content and distribution.

Could we be #2? Sure. And we aim not to stop there. We're continuing to invest in this business. We have great data, great content, adding to that content, expanding our distribution capabilities. And then in terms of -- I'm not in a position to give you any specific guidance on the growth uplift. What I can say is that we're not done yet in terms of the different partnership arrangements. We think this is a really exciting time in terms of new ecosystems, new AI functionality that will provide lots of distribution opportunities for us. And as I mentioned earlier, into customer segments that might not have otherwise accessed our data. And for those customers that have historically accessed our data, this AI functionality enables them to access it in a, I'll say, a much deeper way.

I mentioned earlier the 33 petabytes of data that we have. Historically, our customers have really only scratched the surface of the data and the content that we have. And the AI functionality is much more powerful in really consuming substantial amounts of our data. And then as we shift further down this road, we've talked in the past about evolving our model more towards usage-based and consumption-based pricing. So you put all that together, we are excited about what this opportunity holds.

Okay. And maybe just as a follow-up to that, you've just seen S&P buy With Intelligence. You've seen BlackRock buy Preqin. You've seen MSCI buy Burgiss. So just wondering how you're thinking about your competitive position in private markets data? And is this something you might look to add inorganically to the offering?

Yes. So we already have a lot of private market data, and that includes what we have ingested organically. It includes what we provide from Dun & Bradstreet. The Dun & Bradstreet data, by the way, currently available on the Workspace platform, but soon will be available through a feed, which I think is unique in the industry. We have our partnership with StepStone, which is enabling us to create, again, unique private asset product in our index business. And maybe the last thing I would say is we are not done in this space, and there's more to come in terms of our ability to provide incremental value-add and, in some cases, unique private markets data. So I can comfortably say watch this space.

Your next question is from the line of Ian White of Autonomous Research.

Well, there's been a lot of discussion around the accuracy of general intelligence LLMs in financial services applications. And I guess sort of what advantage can you derive here from your privileged access to your own data when it comes to the training and development of more accurate models? Or kind of put differently, is it realistic that general intelligence tool can match a model that has been trained on your specific data set when it comes to generating accurate results derived from your data? That's essentially my main question.

And just as a follow-up, on the Workspace rollout, which is now complete, what's the latest evidence you have regarding levels of customer satisfaction with Workspace versus the legacy desktop products, please?

Yes. Thanks, Ian. So on the accuracy question, there has been a lot of discussion in the industry about a bunch of the product that is out there really maybe having some nice user interface, but not being remotely close to what this industry demands in terms of accuracy. And so I think that's probably right at this point for a bunch of the products that are out there that we have seen. We expect them to get better over time. I think in terms of our own approach, the advantage that we have is that we have the data. We have the highest quality and broadest data set that allows us to do the necessary training. It is scrubbed data. We're not training our capabilities on the Internet. And so we avoid the garbage in, garbage out problem that you see with a lot of these other models.

And this gets back to the point I was making earlier that through the MCP server construct, we are able to control the access to our data. So we sometimes get questions from people worried about the fact that our data will be made too available and others will be able to, without compensating us, train their models on our data. That's not the case in terms of the way that we make this data available for AI usage or AI consumption.

In terms of the Workspace rollout, we are very pleased with the outcome there, and this was a big exercise over the past couple of years. So we are seeing really good views on the simplicity, on the kind of change in the user interface, on the speed. And there are some aspects in terms of making some of the charting even better. There are a few different things that we're continuing to work on, as I mentioned earlier, sort of week in, week out. And this is going to continue. And it's one of the advantages of this product and the technology stack that we have moved on to.

We've talked about how we've implemented 500 or so changes in each of the last 2 years, and that pace is continuing. So even though we have basically completed the migration, we still have more releases coming. I think we have 2 more releases coming, big broad releases coming this year. Yes, more coming early next year. So it's a continuous improvement exercise, which I think is a great opportunity to continue serving our customers better and better and better.

Got it. If I could just sort of playback and make sure I understood the first point. If anybody wants to sort of train a model on your data, that's kind of a licensable activity that you can kind of control through MCP and a model that's not trained on your data specifically probably won't be very effective or will be less effective than something that's been specifically curated for that purpose. Is that a fair reflection?

I think that's fair. I don't want to claim that we have exclusive financial sector -- in other words, I don't want to claim that in the financial markets, we're the only ones who have financial markets data. There is other data available out there. Ours is the broadest, the deepest, the highest quality. And so we are in an advantaged position. But you've seen companies train their models on public data coming off the Internet. That's on the other end of the spectrum in terms of quality and accuracy. And then there are other data sets out there that you can use. They're just not as extensive and high quality as ours.

Your next question is from the line of Mike Werner of UBS.

And just 2 questions here, one main one and then one follow-up, please. I was just wondering, I mean, you talked a lot today and very helpfully about the new partnerships and LSEG Everywhere. Just stepping back and when we think about the partnership with Microsoft and OpenAI and what you guys are doing there, what's the level of that engagement today versus 12 months ago? I think you used to talk about the number of software engineers that were operating on site on LSEG's premises that came from Microsoft. I was just wondering if you can give us an update there.

And then as a follow-on to a couple of my colleagues' questions. When we think about these partnerships, particularly with the new ones with the AI engines and AI partners, is there any delta or any difference in how you think about the pricing? I know you said you protect the IP, but when you're thinking about these new partnerships, is there any change in the way that users who want to consume that data, would they see any difference in pricing than your traditional customers?

Yes. Got it. Thanks, Mike. So in terms of our partnership with Microsoft, if anything, the level of engagement is higher, and I would say meaningfully higher today relative to where we were a year ago. I know what you're referring to. We've talked in the past about having hundreds of our people embedded with their teams and vice versa. That continues and, if anything, higher level of engagement.

And we talked today about a few other things that the market hasn't really focused on, but that we're building with Microsoft, our Autex Routing Network, our Digital Market Infrastructure. These are not the areas that the market has really focused on, but we are actively building them with Microsoft. And then, of course, our Data as a Service, our analytics, Workspace being embedded in Teams, all the interoperability with Excel and PowerPoint. We have lots of teams working across a lot of different areas with the Microsoft team. So couldn't be happier about the level of engagement there.

And then just with respect to the pricing, in some cases, it's really simple. So for example, we talked about the partnership with Rogo. If you want to access our data in Rogo, you have a Workspace license. It's very straightforward. It can be a little less straightforward if we are providing our data sets, our Data & Feeds data sets through some of these channels, but we have standard pricing for a lot of these. There may always be some negotiations around particular data sets or things like that, but we have standard contractual arrangements for these and standardized pricing for these.

[Operator Instructions] And your next question comes from the line of Hubert Lam from Bank of America.

I've got a couple of questions. Firstly, on D&A, how should we think about revenue acceleration in the next year? So just given the upward momentum on ASV, should we think 6% or more could be achievable for revenue growth in D&A next year?

Second question is, I guess, last results, there was concerns about intensifying pricing competition from a couple of your biggest competitors. Just wondering if you've seen any normalization in terms of pricing? Or was the competition we saw a few months ago a bit of a one-off?

Sure. MAP, why don't you take the first question? I'm happy to take the second one.

Yes, sure. So on D&A, we indeed forecast a revenue acceleration next year. We haven't given precise numbers, but we have given one precise number, which is for our subscription business altogether, reaching 6.5% -- circa 6.5% next year. And obviously, D&A in this number is playing its part, and it will be accelerating '26 and '25.

And then on your second question, Hubert, first, just to remind people, when we talked about some of the competition dynamics at the half year, that was a very small number of cases, a couple in each of the different business areas. And I would say where we are today, we're not seeing that kind of dynamic. It feels a very stable market environment at this point from a competition perspective.

Your next question is from the line of Ben Bathurst of RBC Capital Markets.

My questions are on Post Trade. Firstly, could you help us better understand how interrelated the 2 transactions announced this morning are, if at all? For instance, how different is the list of the founding members of SwapClear from the investing banks in Post Trade Solutions? And then secondly, how significant is the decision to extend the revenue surplus share from 2035 to 2045? Was there always a presumption that, that would be extended? Or was that kind of an incremental sweetness in the deal?

Thank you. Yes. So in terms of the construct of the overall deal, there are 13 banks involved in the swap business today. And in the investment in PTS, there are 11 investing banks, just to be clear around that. Decisions to invest in the new business ventures very much down to sort of individual circumstances of each of the banks there. So not really appropriate to speak on behalf of those in the 13 that aren't in the 11. But what I'll say is super strong engagement across the industry, level of participation in this and interest is very material from all the material players there. So we're very, very happy with that.

And in terms of the extension that you asked about, yes, I think may be different opinions on whether that would have been extended or not, but the fundamental point is this is something that's been in place since 2001. We're here in 2025. It was rolling to 2035. And as part of the overall structure, those 11 banks that are investing in PTS will be extended for a further 10 years to 2045. So a 44-year enduring partnership between the major players in the OTC derivatives space on the sell side with ourselves there. So I think it's part of the overall construct rather than breaking it down into the exact sort of elements of the negotiation.

Okay. Great. So if I understand it rightly, it's just those that are participating in Post Trade Solutions that will have the extension for 2035 to 2045?

That's correct.

And just to be clear, '25 to '35 remains already existing 13. So existing 13 until the maturity of the existing arrangement and the extension of 10 years is to the 11 that are also investing in the Post Trade Solutions franchise business.

Your next question is from the line of Julian Dobrovolschi of ABN AMRO.

I have 2. Maybe the first one regarding the Microsoft product development such as Open Directory and Analytics API and some other things that you're trying to roll out together with Microsoft. Just wondering, are they offered broadly across all the tiers or restricted to premium users and as such as an upsell vector?

And then the follow-up is on ASV growth. Just wondering how confident are you regarding the, let's say, reacceleration of this in the Q4? I think you've been hitting towards 5.8%. And can you please elaborate on the impact of the UBS multiyear contracts and the Credit Suisse revenue crystallization? And perhaps if you can see some leading indicators suggesting a bit of a rebound in ASV growth in the Q4.

Thanks, Julian. So I'll take your first question, and MAP can touch on your question on ASV. So on each of these different products, some of them -- the different products that we have built in partnership with Microsoft, some of them are separate products that have separate pricing, separate licenses, separate arrangements. Some of them are embedded in existing products. And so if we talk about Open Directory and we talk about what's coming in Workspace, you'll see us charge for that over time really through price realization in the core product.

I think then in some of the products that we have rolled out in analytics, the Analytics API, for example, that's a new product, and there's separate charging for that. And we've seen some of that in the uptick in the growth rates in analytics, for example. And let me just -- I'll mention one other example where you can see this very clearly. The arrangement that we announced with Microsoft 1.5 weeks, 2 weeks or so ago, where we are making our data -- we are making some of our data sets available to all users of Microsoft Copilot. So if you have a Copilot license, you can be outside the financial services sector, you have a Copilot license and you're doing something in Copilot, you will get access to certain of our data sets. And that's an arrangement that we have with Microsoft.

And then we have other data sets that you can license directly with LSEG and then have access to them through Microsoft Copilot and Copilot Studio, if you are building, for example, agents using our data. So that gives you an example where some of them are embedded -- some of the pricing arrangements are embedded in existing products. Some of them are new, and we are charging incrementally for them. Let me turn it over to you, MAP.

Yes, sure. So first of all, before addressing your question, I'd like to point out that we have outperformed our previous guidance on ASV. And remember, in H1, we were expecting that the Q3 ASV would fall to 5.4% with 40 bps of impact of UBS. So excluding UBS 5.8%, so comparable to Q2, and we posted 5.8% in Q2, 5.4% was what we were expecting in Q3. We actually outperformed this to 5.6%. So ex UBS, 6%, an acceleration from the 5.8% we were at the end of Q2. And when I look forward for the end of this year, we're very confident into accelerating again to 5.8%. And here, it's the same thing. It's 5.8%, including of the 40 bps for UBS. So actually, excluding it, 6.2%. So 5.8%, 6%, 6.2%. That's basically the message today.

Your next question is from the line of Enrico Bolzoni of JPMorgan.

I wanted to ask you, you now revised your EBITDA guidance a couple of times, even excluding the newly announced deal. So I just wanted to ask you, what are you doing particularly well or better than you expected that basically drove the consecutive revision in guidance? So that's my first question. And partially related to that, just some small clarification. So one, you are clearly now spending just over GBP 1 billion to in-source this additional revenue from SwapClear. Can you just clarify whether this will be capitalized and whether the amortization of that will be above or below the line? So that's one question.

And another related question to numbers. You're clearly issuing some debt, you're guiding for EPS accretion in 2025. What about 2026? I know you talked about margin expansion for EBITDA in 2026. Can we say that we will also see a similar EPS uplift for next year?

All right. So I begin with EBITDA margin. So yes, just to remember for maybe those of you who didn't see it, we began with 50 to 100 bps of EBITDA margin guidance for this year, we then improved it to 75 to 100 bps. And finally, we are now confident to reach 100 bps. It's really an acceleration. So what we have implemented in the last 2 years at LSEG is a full cockpit of cost discipline, addressing all the different components of our cost base. So mostly people, we're talking a lot of people, obviously, but it's true for cloud costs, on-premise costs, travel expense and so forth and so on. And basically, this acceleration is coming from the fact that what we have put in place is more efficient and is producing more results and quicker, if you want, than what I expected at the beginning of the year.

The second reason, which is maybe -- so that's an acceleration. Second reason which is more structural is -- and maybe you remember what I was telling you at the earnings of 2024, the different automation solution that we have put in place at different places in the company. So in QAS, meaning our customer service, in our content ingestion, we were putting it in place, and I was expecting to see the first materialization into savings next year. And actually, it's happening as early as this year. So that's the combination of the 2.

Now to answer your second question about the GBP 1.15 billion, that represents the alteration of the SwapClear revenue share. So we're considering this as an acquisition. So we are creating an intangible asset exactly as we would do as a traditional acquisition. And we are going to amortize it over 10 years below the line as the rest of our acquisition.

And then your final question, which is the accretion. So accretion of 2% to 3% in 2025, because I want to be clear on the fact -- I hope I was clear in my script that this revenue share alteration is retrospective to the 1st of January of '25, okay? So it means that we benefit from the full accretion in terms of EBITDA margin that I have mentioned of 100%. And in terms of EPS taking into account the financing cost. We said 2% to 3% in '25, and we'll have pretty much the same thing, 2% to 3% in '26.

And your next question is from the line of Tom Mills of Jefferies.

I think we've skirted around it a few times on the call. I just wanted to clarify that you are sort of reiterating you're expecting to deliver around 3.5% price increase on the 1st of January is kind of [indiscernible].

Absolutely. Absolutely. We've just sent -- the price letter was sent in September. On the basis of the first reaction from this price letter and our experience, we are confident we will derive the same type of yield around 3.5% in '26 as the one we had this year in 2025.

And your next question is from the line of Oliver Carruthers of Goldman Sachs.

Oliver Carruthers from Goldman Sachs. Thanks for a lot of the incremental KPIs around D&A. I just have one quick modeling question on the FTSE Russell subscription revenues. I think you're calling out the more modest growth in subscription growth here in Q3 was to do with this mandate renewal cycle that you think is going to normalize next year. So just what's reasonable to assume in terms of the pickup in growth rate? I think you're running at around 5% on a constant currency basis year-over-year for Q3. And the reason I ask is if we go back to 2024 levels of around 10%, on my math, this adds something like 70 basis points to your ASV. So just any parameterizing of that would be very helpful.

Yes. Thanks, Oliver. So you're right. This year, a much quieter period in terms of renewals during which we would typically see incremental revenue associated with either regular price rises or bigger, broader business relationships and broader engagement. I think hard to give you specific numbers as to what that's going to look like in '26 and beyond. You've seen how this business has performed in years past in that kind of higher than mid-single-digit zone. So I think I'm probably pretty comfortable, and MAP, feel free to weigh in here as well. I think we're pretty comfortable in that zone, but I don't want to be giving you any sort of specific guidance on what that looks like at this point.

And there are no further questions on the conference line. I will now hand the presentation back to David Schwimmer, CEO of LSEG, for closing remarks.

Great. Well, thank you all. Thanks for joining us today. As I said upfront, a little bit more substance in this one rather than a typical Q3 update. We hope you all have found it useful. And if you have any questions, you certainly know where we are. We'd be happy to take any further questions through Peregrine and the team. Thanks again.