Pony AI Stock

Ladies and gentlemen, thank you for standing by, and welcome to Pony AI Inc.'s Second Quarter 2026 Earnings Conference Call. [Operator Instructions] As a reminder, today's conference call is being recorded, and a webcast replay will be available on the company's Investor Relations website at ir.pony.ai. I will now turn the call over to your host, George Shao, Head of Capital Markets and Investor Relations at Pony.ai. Please go ahead, George.

Thank you, operator, and hello, everyone. We appreciate you joining us today for Pony AI's Second Quarter 2026 Earnings Call. Earlier today, we issued a press release with our financial and operating metrics, which is available on our IR website. An earnings presentation, which we will refer to during this conference call can also be accessed and downloaded on our Investor Relations website.

Joining me on today's call are Dr. James Peng, Chairman of the Board and Chief Executive Officer; Dr. Tiancheng Lou, Chief Technology Officer; and Dr. Leo Wang, Chief Financial Officer of the company. They will provide prepared remarks followed by a Q&A session.

Before we begin, please refer to the safe harbor statement in our earnings release, which applies to this call as we will be making forward-looking statements. Please also note that we will discuss non-GAAP measures today, which are more thoroughly explained and reconciled to the most comparable measures reported under GAAP in our earnings release available on our IR website and filings with the SEC and the Hong Kong Stock Exchange.

I will now hand it over to our Chairman and CEO, Dr. James Peng. Please go ahead.

Thank you, George. Hello, everyone. Thank you for joining our earnings call today. We delivered another fantastic quarter highlighted by multifold expansion across the board. First, strong top line growth. Total revenue surged by 69% year-over-year, driven by a close to 8x jump in robotaxi revenue and over 9x surge in fare charging revenue. Second, rapid fleet scaling. Our robotaxi fleet expanded to 2,000 vehicles, putting us on track to deliver 3,500 vehicles by year-end. Third, expanded deployment. Domestically, we reinforced our leadership in Tier 1 cities as we surpassed 1.5 million registered users. We also improved our network density with more deployed vehicles and operational coverage.

Internationally, we unlocked demand by scaling our joint deployment model. Currently, we have secured over 4,000 vehicle commitments with Uber and other overseas partners. The expanded deployment in both China and overseas markets clearly shows that our dual engine strategy is turning into robust top line growth. Looking at our domestic operations first, the L4 industry in China is entering a new phase where higher standards are required to keep the industry on a sustainable healthy trajectory. For any robotaxi company to enter into large-scale deployment, it now needs proven driverless capabilities, positive user satisfaction and verified safety records.

We are perfectly positioned to capitalize on this shift because we have already been operating well ahead of this curve. These rising standards will only widen our competitive moat and solidify our leadership in China. Our confidence actually is grounded in solid results from commercial robotaxi operations. We have 3 Gen-7 robotaxi vehicle models in our daily services, including the GAC Aion V, the BAIC ARCFOX Alpha T5 and the Toyota bZ4X. We are continuously improving user experience, which is a key driver for our organic user growth as our total registered users have surpassed 1.5 million.

In Guangzhou, we extended our robotaxi services into the city center, now adding over 300 square kilometers since the beginning of this year. The operational area spans across Haizhu, Tianhe, Huangpu and Panyu districts, covering a population of over 7 million. As a result, our driverless fleet is positioned to capture highly concentrated urban mobility demand. Shenzhen, known as China's Silicon Valley serves also as a great showcase of our capability to navigate highly complex traffic scenarios. Our operational resilience was rigorously validated by corner cases such as the high-demand holidays such as the Dragon Boat Festival, the peak rush hours and heavy rainstorms.

Despite these demanding conditions, we effectively met high-frequency commuting demands. In addition, by seamlessly integrating 3 major transit hubs, including Bao'an International Airport, Shenzhen Bay Port and Shekou Cruise Port, we further expanded our network to provide users with greater convenience and more mobility options.

Now turning into our global expansion. To meet the ever-increasing demand of L4 mobility in overseas markets, we have entered more international markets with huge consumer demand and commercial potential. We are using our joint deployment model to form global alliance, fulfilling autonomous mobility demand in these international markets and creating values for our partners. To that end, we collaborated with multiple partners to accelerate our international pipeline. Currently, we have secured over 4,000 vehicle commitments, led by over 2,000 robotaxis across 5 European cities with Uber, alongside commitments from some other partners. Meanwhile, we continue to deepen our operations in existing markets. In Luxembourg, our deployment with Bolt and Stellantis keeps moving forward. And in Singapore, our service is now officially live for the general public on ComfortDelGro's ride-hailing app called Zig. These international demands are a direct endorsement of our Gen-7 robotaxi operations in China's Tier 1 cities, where we have proven our superior driving capability, reliable 24/7 operations, high user satisfaction and positive UE. I'm confident that this proven model will continue to win partners with more vehicle deployment commitments and drive user adoption globally.

Now let me elaborate a bit more on our joint deployment model. As we expand our fleet across China and overseas, we leverage existing local ecosystems and our partners' on-the-ground expertise to drive capital-efficient expansion. I am very pleased to share that the model is already delivering strong tangible commercial results. First, look at the strong monetization was validated in Q2. By broadening our partnerships, we delivered significant quarter-over-quarter growth in revenue contribution. That's a direct proof point of this JDM model's financial viability.

Second, the joint deployment model is an asset-light one where our partners fund the fleet. This fundamentally enables faster scaling, lower unit costs and superior capital efficiency for our fleet expansion. Third, with fast scaling, JDM essentially unlocks massive commercial value for years to come. For example, we recently expanded our partnership with Uber to target at premium markets. This creates a highly repeatable growth engine, allowing us to attract more partners and deliver even higher growth trajectory.

Now let's move to our robotruck business. Our robotruck business delivered outstanding results in Q2 with revenue jumping more than 40% year-over-year. We actually expect this growth momentum to persist and even strengthen in the second half of this year. We continue to expand our long-haul operations with Sinotrans through our joint venture. At the same time, our Gen-4 robotrucks have entered mass production and already begun commercial operations. Working with China Merchants Port, we launched our commercial deployment of robotrucks at Shenzhen's Mawan Port, where our fully driverless robotrucks operate together with other human-driven trucks. This success highlights our unique cross-segment synergies. We have leveraged our rich operational experience from urban robotaxis and long-haul robotrucks to enable our trucks to seamlessly navigate traffic interactions at the ports. As we pass the midpoint of the year, our acceleration across both domestic and international markets puts us well on track to surpass our 20 city goal by year-end.

In China's Tier 1 cities, we will continue to deploy more vehicles to our fleet to widen our competitive moat and advance our scaling edge. And at the same time, we are on track to enter multiple domestic new markets. Internationally, the joint deployment model will contribute to top line growth with great capital efficiency. This dual momentum gives us greater confidence in beating our original robotaxi revenue outlook, which is exceeding 3.5x last year's level. Looking ahead, our focus remains clear, delivering long-term value creation and driving the commercialization of autonomous driving with capital efficiency.

Now I will hand it over to our CTO, Tiancheng, to go over the technology progress. Tiancheng, please go ahead.

Thank you, James. Hello, everyone. This is Tiancheng. To start, our strong Q2 momentum is driven by our unique tech stack. This foundation allows us to scale rapidly and adapt seamlessly across both domestic and international markets. Starting with our domestic market. This is where we validate our technology in most challenging scenarios and translate this mastery into commercial value. Tier 1 cities such as Guangzhou and Shenzhen are clear examples. In older urban cores and major transit hubs, roads are narrow, residential neighborhood are dense and roadside parking are common. Our World Model and the Virtual Driver proved to be more agile and precise in navigating these extreme conditions, ultimately delivering high commercial returns than regular scenarios. We also rapidly replicate this success to more high premium urban market globally. Traffic rules and driving habits vary significantly across China, Europe, the Middle East and Asia. Despite these fundamental region differences, our robust generalization enables rapid deployment.

Our proven technical track record, especially in Tier 1 cities of China and Zagreb of Croatia is exactly why top-tier partners are choosing to scale with us through our joint deployment model. Beyond the driving capability, another key engine behind our expansion is efficiency. Let me now elaborate on how our unique technical and operational capabilities deliver these efficient benefits. As I shared in previous quarters, the key to enabling our robotaxi to seamlessly navigate diverse urban environment lies in our World Model precision. This is what bridges gap of so-called sim-to-real in physical AI area. In autonomous driving, closing the gap comes down to more than the probability distribution of different behaviors among traffic participants. For example, the probability of a pedestrian standing on the roadside suddenly jaywalking varies from city to city.

A high precision World Model accurately captures these dynamics, enabling the Virtual Driver to handle such scenarios with confidence. Our current upgraded PonyWorld 2.0 brings this precision alignment into loop engineering. The system automatically isolates deeply hidden issues, generates targeted solutions, and validates them for real-world deployment, reducing the need for human engineers to analyze cases one by one. This also dramatically accelerates our development timeline. The old way of entering a new city takes dozens of engineers doing manual work to review local driving issues, analyze root cause of these issues, upgrade the World Model and retrain the onboard models and then deploy and validate the new model on the road. However, with PonyWorld 2.0, our system leverages AI to resolve these local changes automatically. This turns City's expansion from effort that used to take dozens of engineers into a human-in-the-loop automatic process that just a few people can run.

So for example, when we went to Zagreb, we noticed local drivers almost never slowed down when they hold the right of way, even near blind spot. PonyWorld 2.0 caught this difference automatically, and we quickly trained a new version of Virtual Driver that fits local habit perfectly, with very few engineers involved. As a result, we can now launch in multiple cities with completely distinct driving environment all at once. This scalability ensures we efficiently achieve our target of 20 cities by the end of this year. This gives us the unique efficiency advantage to scale our footprint far more rapidly.

On the operational side, we are also using technology to redefine efficiency. For example, we don't need a closed dedicated parking lot to charge our cars. Our robotaxis can share normal parking lot with human drivers, driving themselves to find an open charging spot without human intervention. This means a tiny ground team can easily manage charging and service for a large fleet. This optimizes personnel allocation and lower our UE cost. The vehicle to staff ratio for our ground supporters and remote assistant team has improved significantly. More importantly, it also boosts the willingness of industry partners to adopt our joint deployment model. So as James mentioned, multiple partners such as Uber are clear examples. In short, our tech-driven efficiency give us a unique operational leverage as we scale across new markets. This not only reinforce our competitive moat, but also positions our technical innovation as a core engine driving the entire industry forward. This concludes my prepared remarks.

I will now pass the call over to our CFO, Dr. Leo Wang, for a closer look at our financial results. Leo, please go ahead.

Thank you, Tiancheng. Hello, everyone. This is Leo. I will focus on year-over-year comparisons for the second quarter and the first half of 2026, unless otherwise noted. For detailed financials, please refer to our earnings release. This quarter, total revenues reached USD 36.2 million, representing a remarkable 69% increase from USD 21.5 million in the same quarter last year. Breaking down the strong top line growth by business segment. Most notably, our robotaxi revenue are growing 691% and the robotruck revenues growing 40%. Our phenomenal triple-digit robotaxi growth is a strong demonstration that our commercialization strategy is translating into good financial numbers.

Looking deeper into robotaxi, we delivered a very strong growth this quarter. Robotaxi revenues reached a record high of USD 12.1 million, growing 691%, a further acceleration from the 395% growth compared to the first quarter. Our fare charging revenue delivered an exceptional growth rate of 849%. These rapid growth rates show that robotaxi continues to serve as our core growth engine. This acceleration was driven by several factors. First, our fare charging fleet continued to expand across more regions and specifically into core downtown areas with high economic values.

Second, our joint deployment model gained significant momentum and our commercial robotaxi launched in Zagreb, Croatia has served as a powerful showcase as the first of its kind in the city center of a European capital Zagreb, has proved our high-quality service in a demanding international market and enabled us to secure additional overseas contracts. Under the joint deployment model, we are currently recognizing upfront vehicle delivery revenues, which established a solid foundation for us to have high-margin recurring revenue sharing income going forward as our fleet operations scale. What is particularly encouraging is that this acceleration is broad-based, not concentrated in a single market. In China, in this quarter, we continue to strengthen our leading position in Tier 1 cities with fast-growing scale and a strong user base. Overseas, we are building an alliance that accelerates our global footprint. For example, we have secured over 4,000 initial vehicle deployment commitments with Uber and other overseas partners. Our continuous expansion in China and overseas were translating into a rapidly increasing base of recurring robotaxi revenues.

Turning into robotruck. The revenue grew 40% year-over-year to USD 13.3 million this quarter. This growth was driven by increased logistic transportation revenues. Robotruck growth is more than just about volume. It reflects the cross-segment synergies within our ecosystem from robotaxi to robotruck. As James highlighted, the Mawan Port demonstrates our ability to apply the technology and operational capabilities polished in robotaxi urban environments and robotruck long-haul routes to a new vertical. Our Intelligent Solutions segment delivered revenue of USD 10.8 million this quarter, a 4% year-over-year increase with the growth rate moderating due to the delivery fluctuation from domain controllers.

For the first half of 2026, the Intelligent Solutions revenue reached approximately USD 26.3 million growing 77%, with meaningful sales contribution from domain controllers. Moving to margins. Gross profit reached USD 6.4 million, an 83.4% increase year-over-year, translating to a gross margin of 17.5% up 1.4 percentage points from 16.1% the same quarter last year. Total GAAP operating expenses were USD 72.1 million this quarter and non-GAAP operating expenses were USD 63 million representing a modest 9.6% increase. The expense increase is significantly lower than our revenue growth rate of 68.8%. As Tiancheng mentioned, our leading PonyWorld Model 2.0 and AI-powered closed-loop R&D framework allows the same engineering team to handle far more work across different cities and a complex corner case analysis. The R&D efficiency is directly visible in our financial numbers. We are scaling globally without proportionally scaling our cost base.

We continue to see our operating loss margin narrowing and the operating leverage beginning to materialize as revenue scale. The loss from operations was USD 65.7 million, a modest 7.3% increase. The operating margin narrowed dramatically from negative 285.6% in Q2 2025 to negative 181.5% this quarter, an improvement of over 100 percentage points. On a non-GAAP basis, loss from operations was USD 56.7 million, increased by less than 5% year-over-year. Net loss narrowed significantly to USD 45.4 million, a 14.9% year-over-year decrease compared to Q2 2025. The net loss margin narrowed from negative 248.3% to negative 125.2%, an improvement of more than 100 percentage points. From a broader perspective, our revenue growth rate significantly outpaced our non-GAAP operating expense growth rate, clearly demonstrating economics of scale and operating leverage.

Turning to our balance sheet. Cash and cash equivalents, short-term investments, restricted cash and long-term wealth management instruments stood at USD 1.39 billion as of June 30, 2026, compared to USD 1.44 billion as of March 31, 2026. We continue to maintain a prudent cadence in cash management and maintain a robust financial position. Net cash used in operating activities was USD 44 million this quarter compared to USD 25.4 million in the second quarter of 2025. The increase was due to normal working capital fluctuation, especially the settlement of accounts payable during the current quarter, coupled with strategic investment in inventory and prepaids to support our fleet expansion in the second half this year.

Capital expenditures were USD 32.2 million this quarter, bringing first half CapEx to USD 44.3 million. This was mainly driven by the fleet and autonomous driving kit CapEx as we see robotaxi acceleration in both domestic and overseas market as well as increasing spending in data centers to support our greater scale deployment and continuous R&D. As we scale up our fleet, we expect to maintain capital discipline, supported by our partners' co-investment under the joint deployment model framework. Our capital allocation strategy is designed to balance disciplined investment with scalable growth. Specifically, we invest in our core technology and owned fleet in key domestic markets, while partners contribute fleet capital and the local operating capability through the joint deployment model. This allows us to expand our revenue-generating footprint across China and international markets without a proportional increase in capital intensity.

Together, with approximately 2,000 vehicles produced operating footprint across the world, more than 1.5 million registered domestic users and USD 1.39 billion cash reserve, we have the operating momentum, global opportunities and financial resources to execute our full year target and support sustainable growth beyond 2026. Meanwhile, with our recent inclusion in Hong Kong listing Stock Connect, we are excited to welcome onshore investors and maintain committed to transparent market engagement and the long-term shareholder value creation.

I will now turn the call over to the operator to begin our Q&A session. Thank you.

[Operator Instructions] The first question today comes from Ming-Hsun Lee with Bank of America.

Congrats for the good results. I only have one question. Given that Uber partners with several autonomous driving companies worldwide, what are the main reasons that made Uber choose Pony in its European rollout.

Thanks, Ming Hsun. This is James, and I'll take this one. As you can see that I'm actually quite pleased that we have signed commercial agreement with Uber to deepen our collaboration. I think the reasons Uber decided to work closely with us actually quite straightforward.

Uber always looks for autonomous driving partners whose technology is reliable at scale, and also whose cost structure brings the attractive economics. That's exactly the two reasons that we can offer on the table. We actually worked with Uber back in early 2025. At that time, our Gen-7 Robotaxis just started the deployment in China. And at that time, there were some doubts whether our autonomous driving capabilities can handle the European cities, especially the big ones where the infrastructure and road condition are typically mixed with old and new.

But after a year, now look at I think the question has been answered with resounding real-world evidences. We have already launched large-scale Robotaxi commercial operations in all tier 1 cities in China. The unit economics turned positive in Guangzhou and Shenzhen. In addition, we also rolled out Europe's first commercial Robotaxi service in Zagreb, Croatia, with Uber and Verne. So all this evidence shows that our Robotaxis can cover the most complex highest demanding scenarios. And also, what we have found out is the more places a vehicle can operate, the higher utilization becomes. So on the cost side, we also can offer is even more compelling, right? Compelling with the hardware and also the operational costs. Our total cost per mile is the most competitive in the industry.

I think another important reason is that the culture alignment has also been a hallmark of our collaboration between Pony and Uber. Both sides are impressed by one another's professionalism and dedication. The mutual appreciation and the mutual commitment really lead to -- right now what we have seen, the expanded collaboration.

For both of our companies, the strategy is to begin with the most socially and economically meaningful markets, and then we'll even extend our mobility services to additional geographies. So what we announced about the 2000 vehicles is under the current contract. With these contracts, we become Uber's largest autonomous driving partner in Europe. Going forward, as the performance and also the economics continue to validate at scale, we'll see substantial room to expand the fleet size even further. So back to the operator.

The next question comes from Tim Hsiao with Morgan Stanley.

Congratulations on the strong quarterly results. Could you please elaborate on your strategy going forward for the joint deployment model? And also, can you share more color on how the commercialization model works and operate under asset-light model?

Thanks for the question. This is James again. Probably let me begin with high level, and I'll probably -- regarding the details, I'll hand over to Leo. So the joint deployment model will actually accelerate our fleet expansion with high capital efficiency, both domestically and internationally. You can think of this as -- in this model, we are building a win-win model across the value chain. The success of our Gen-7 Robotaxi operations across the Tier 1 cities, it's really a showcase. It proves that our superior safety record and operational efficiency and then ultimately positive UE margins. By delivering this top-tier driving capabilities and user experience and at the same time, at very low hardware and operational costs, we can achieve high margins than our peers.

Therefore, partners in our ecosystems, whether it's a mobility platform or a fleet operator they can share the most economic value per deployed vehicle. So on the high level, we can think of the joint deployment model, they give partners are naturally incentivized to commit a large portion of their fleet shares to Pony because in this model, they can maximize their total value generated together with us.

Regarding the details of this business model, I'll now hand over to Leo.

Yes. Thanks, James, and this is Leo. Yes, Tim, you mentioned is correct. This is an asset-light model for Pony to expand our fleet. And in most cases, there are three parties and each plays a different role.

For Pony, we supply our Gen-7 Robotaxi with our Virtual Driver capability that is AI driver, a mobility platform that can introduce user demand and an operating company who can deal with fleet management and maintenance. We, of course, acknowledge in different markets, the consumer can have the choice on mobility platforms. And there are existing operating companies. So we don't want to disrupt this ecosystem in these markets.

But instead, our joint deployment business model is trying to bring values and form a win-win alliance. For example, we leverage Uber and Bolt as mobility platforms to attract demand. And we are also partnered with Verne in Croatia and ComfortDelGro in Singapore as local fleet operator. So from a financial perspective, this model could generate sharing-based revenue or technology licensing fee for Pony. And this has not only broaden our revenue base but also introduce higher-margin recurring income across the entire Robotaxi operating life cycle.

And as we expand our footprint into higher premium international markets, for example, in Europe, in Middle East and in other parts of Asia. We definitely think that this could lift our long-term financial outlook. And just to be clear, these 4,000 vehicle commitment from Uber and other partners will serve as a multiyear growth catalyst for 2026 and beyond. I will now turn the call back to the operator.

The next question comes from Paul Gong with UBS.

I have one question regarding on the PonyWorld 2.0. I think Tiancheng has mentioned about its self-evolution and loop engineering. Can you please provide more color on what makes self-evolution different in autonomous driving? And how does it improve your R&D efficiency? And if we think in the future, if someone open source the World Model, would your moat be affected?

Thank you. This is Tiancheng. I will take this one. To start, I would say autonomous driving is a physical AI. So training the onboard model or improving the world model are both built on real-world feedback. A general purpose open source world model is basically just a 3D video generator. We can generate data, but that's nowhere near enough to train our autonomous driving system. So we use world model to train the onboard model through reinforcement learning. To do this right, it is not just simulating what people do. It's about how often they do it. Take a pedestrian steadily jaywalking as example, the chance isn't 99%. It's not 1% either. So precision means matching the exact real-world probability. That level of statistical accuracy is what we mean by precision of the world model.

So the probability distribution of traffic participants varies from city to city. Although our model generalized capabilities strong enough to handle extreme scenarios worldwide, we still need to fine-tune it for local driving styles. So for example, in both China and Croatia, there are drivers who change lanes without checking behind them. It happens with different probability in different places. So that's where Ponyworld 2.0 comes in. It is a fast-evolving system that is continuously improving the world model precision.

In the past, our workflow was human-led. So when we enter a new city, we check data from that region then engineers will determine which scenario the current world model lack precision. Now AI drives the whole process. Human are still involved, but mostly for verification and validation. So as a result, we significantly reduced engineering resources to enter a new city. In other words, without adding R&D resources, we can either enter many new markets at the same time, quickly achieving safe and smooth L4 autonomous driving. This ability to scale is a very large moat, and I do not think it will be affected by any open source generative world model. With this, back to the operator.

The next question comes from Jeff Chung with Citi.

This is Jeff. My question is about the domestic market. And how should we think about Pony's new outlook for the domestic market heading into the second half of the year.

This is James. I'll take this call. As you can see that China is our home base. I believe that domestic fleet expansion remains a significant part of our vehicle roll-out. China itself represents a massive mobility market with over 10 million taxis and ride-hailing vehicles. So it's a highly -- the reality is that also the mobility demand is highly concentrated in Tier 1 cities and Tier 2 cities. As a result, our strategy remains the same. We will start our focus from the highest valued markets and then expanding into other cities and regions.

The Tier 1 cities alone account for a significant share of the national ride-hailing demand. These cities are also the ones that offer the most mature regulatory framework to support autonomous driving. Today, our scale and the commercial model in these Tier 1 cities remains industry-leading. In our larger operational hubs, such as Guangzhou and Shenzhen, we are already seeing strong growth momentum. Expanding the fleet size in these markets shortens users' wait time and boosts user retention. And then as a result, it directly translates into higher daily revenue per vehicle, even as we scale up our fleet size.

So this virtuous cycle not only drives paid order growth and margins, but also reinforces our regulatory trust and also the brand recognition. At the same time, scaling allows us to amortize the operational costs, driving down our daily per-vehicle costs. So what we have seen is really a continuous improvement of our UE margins. Therefore, we will proceed with deploying more and more fleets in the Tier 1 cities to widen our competitive moats. Meanwhile, of course, second-tier and even third-tier markets are strategically vital.

This year, we plan to enter key cities, such as Changsha, Hangzhou, and many of the additional Greater Bay Area cities, and potentially some other cities and regions. This will establish the foundation for these markets, essentially become a new growth engine for us to go forward.

With this, back to the operator.

The next question comes from Xiaoyi Lei with Jefferies.

This is Xiaoyi from Jefferies. My question is on robotaxi operations. You've mentioned that operational efficiency is crucial for running the fleet at scale. Could you maybe give us more color on how is that actually being achieved? For example, on the remote assistance side, vehicle utilization or charging and maintenance perspective. And then how those efficiency gains are helping you accelerate deployment, both in terms of like expanding existing cities and entering new ones?

This is Tiancheng. Thanks for the question. So regarding the operational efficiency, I will start saying, based on our experience across the Tier 1 cities, we now have developed a deep understanding of the complexity of operating the fully driverless fleet. It is a completely different game from managing traditional taxis. At the end of the day, efficiency comes down to one thing, the fleet-to-staff ratio. With traditional taxis, it is always one-to-one. 100 cars need 100 drivers to handle everything from cleaning, charging, to daily maintenance. For us, it is not just about managing people better, but even critically on whether technology can minimize need to human involvement.

For example, when all of our robotaxis return to a depot, they require zero human assistance. Autonomously navigating, locating available chargers, and executing self-parking, even in a very tight space. Because of that, we need three people for every 100 robotaxis to keep daily operations running smoothly. That is true whether we run them by ourselves or work with partners. This is directly translated into significantly lower operating costs per vehicle and advanced unit economics. Therefore, without inflating management overhead and cost, we can still expand into new cities and deploy more vehicles rapidly. We have developed this know-how into standardized operating procedures and automation tools. That is why more and more partners are joining us to adopt our joint-deployment model, making Pony's robotaxi the most efficient and profitable choice available. With this, back to the operator.

The next question comes from Kai Xiao with CICC.

Congratuations for the quarter. Could you give us an update on your new business initiatives, specifically the progress with your L4 light truck business.

Thanks, Kai. This is James, and I'll take this one. The new business initiatives, especially the L4 light truck, I think fits very well with our vision and ambition, which is autonomous mobility everywhere. The L4 light truck has a great synergy among our current product offerings. Think about it can leverage the Robotaxi's driving capabilities and cost-efficient hardware. And at the same time, the light truck also shares the same customer base with our robotruck. The light truck almost shares 100% of our Robotaxi's technology and operational infrastructure. So essentially, the development and operation can slash our costs. The light truck extends the logistics portfolio from long haul into urban delivery. It essentially unlocks a new TAM. In China alone, the active light truck fleet on the road exceeds 8 million vehicles.

Also look at the current already on the ground, the low-speed robovan. Compared with that, our light truck offers 3 to 4 times the cargo capacity, and also the speed is 2x faster. As a result, it can open up heavier loaded commercial applications across the full urban supply chain. If you think about typical usage, those from distribution hubs to the shopping malls, to the supermarkets and also the convenience stores. As you recall that we actually unveiled the L4 light truck in the Beijing Auto Show, since then, it has been 4 months. And in that 4 months, we have already built a strong commercial ecosystem. The vehicle chassis are jointly developed with CATL. The vehicle is the world's first automotive-grade, fully redundant light truck, purposely built for L4 autonomous driving.

Currently, we also have secured partnerships with SF Express and China Post Technology, two leading logistics operators in China. With the orders and the deployment schedules already in place, this partnership can create a strong pipeline for the autonomous urban delivery.

Looking at the remaining of this year, I believe that the collaboration pipelines with even more OEMs and the fleet operators will still in the pipeline to drive scaling up. We will also integrate with urban logistic network platforms to capture even further demand. So I'm actually very excited about this new initiative. With this, back to the operator.

The next question comes from Anne Ni with Everbright Securities.

We know that Waymo's management recently said that "The Demo is Only 1% Of The Work." Could Pony's management share your views on this comment, please?

Thank you, this is Tiancheng. I will take this one. First of all, this is an interesting framing, and I think it captures something real. Building an impressive demo and scaling are two entirely different games. Autonomous driving is really a probability problem. If you get into one accident, every 1,000 kilometers, sure, you can do a demo -- because a demo only covers a few kilometers. But at scale, this accident rate is a deal breaker.

A typical ride-hailing vehicle drives about 300 kilometers a day. So if you have a fleet of 100 cars in 1 city. So that is tens of thousand kilometers every day. The fleet will see 10 accidents every single day, then no regulators will tolerate this and the public definitely won't. So because autonomous driving is a probability problem, risk evolves differently at scale.

Proving safety takes time and mileage, and you cannot just shortcut by dumping thousands of cars on the street overnight. Fleet size and time are not interchangeable. This is also why regulators everywhere takes exactly same approach. They go step by step. A small fleet first, prove safety at that scale, then to the next level. So technically going from a demo to full scaling takes multiple 10x jumps in performance. And every jump is harder than the last. It's not just about fixing the remaining 10% of problems, but also systematically resolving 90% of the issues without creating new ones.

For example, hard braking to avoid a collision may solve a problem, but it may create more rear-ended collisions. And if the underlying technical approach is wrong, safety has a hard ceiling. Therefore, proving safety to regulators is just only one of the bar. From a technical standpoint, new players have to prove they can iterate very fast, because the leaders are already miles ahead by several order of magnitude of safety. So long story short, if all you have today is a demo, you still need to prove that you can achieve multiple 10x performance jumps. And on top of that, you need time to build the trust with regulators before you can scale.

So for Pony, we have already checked both of these boxes. That's why our focus for today is on expanding into more cities and deploying larger fleets. With that, back to the operator. Thank you.

As there are no further questions now, I'd like to turn the call back over to the host for closing remarks.

Thank you once again for joining us today. If you have any further questions, please feel free to contact our IR team. We look forward to speaking with you in the next quarter.

This concludes today's conference call. You may now disconnect your lines. Thank you.

Ladies and gentlemen, thank you for standing by, and welcome to Pony AI Inc. First Quarter 2026 Earnings Conference Call. [Operator Instructions] As a reminder, today's conference call is being recorded, and a webcast replay will be available on the company's Investor Relations website at ir.pony.ai.

I would now turn the call over to your host, George Shao, Head of Capital Markets and Investor Relations at Pony AI. Please go ahead, George.

Thank you, operator. Hello, everyone. We appreciate you joining us today for Pony AI's First Quarter 226 Earnings Call. Earlier today, we issued a press release with our financial and operating results, which is available on our Investor Relations website, and earnings presentation, which we will refer to during the conference call can also be accessed and downloaded on our Investor Relations website. Joining me on the call are Dr. James Peng, Chairman of the Board and CEO; Dr. Tiancheng Lou, CTO; and Dr. Leo Wang, CFO of the company.

They will provide prepared remarks followed by a Q&A session. Before we begin, please refer to the safe harbor statement in our earnings release, which applies to this call as we will be making forward-looking statements. Please also note that we will discuss non-GAAP measures today, which are more thoroughly explained and reconciled to the most comparable measures reported under GAAP in our earnings release available on the Investor Relations website and filings with the SEC and Hong Kong Stock Exchange.

I will now hand it over to our Chairman and CEO, Dr. James Peng. Please go ahead.

Thank you, George. Hello, everyone. Thank you for joining our earnings call. We kicked off 2026 with an amazing first quarter. This strong start defines our growth momentum for the whole year. Let me start with the highlights. I'm proud to report that in Q1 2026, our total revenue grew by [ 145% ] year-over-year, and we also achieved record high quarterly robotaxi revenue. Specifically, our robotaxi revenues grew nearly 400%, powered by more than 50% surge in fair charging revenues.

Our operational momentum is accelerating across the board. We have scaled our robotaxi fleet to exceed 1,700 vehicles and amplified this expansion to a massive surge in user adoption. Now our registered users grew more than 200% year-over-year in China. In fact, our weekly average paid orders so far in May grew more than 100% compared to the beginning of the year. Lastly, we continue to expand our operating area currently broadening our service footprint into an [indiscernible]. Globally, we have been advancing our operations in the capital of Croatia, totalizing Europe first commercial robotaxi service.

Looking at our overall Q1 results. I'm thrilled that our strategic and execution mode translated directly into our exponential growth in robotaxi and fare charging revenues by scaling our fleet, user base and pay the order volume. We have achieved consistent month-over-month growth this year. This is a remarkable achievement as spring typically is a low season for ride hailing. Our dual-engine strategy that is focusing on both China and the global market and the joint deployment model started unlocking new and diversified revenue streams. China market remains our primary growth engine, where we have secured a dominant lead. We are steadily ramping up our domestic suite while simultaneously broadening our operational footprint.

We expanded our operations in Guangzhou from [indiscernible] districts into [indiscernible] district, which is the heart of Guangzhou that covers high-demand areas like Canton Tower, the part CBD and the Canton complex. In Shenzhen, we have been continuously increasing the size and the density of our fleet in [indiscernible] District, the city's 2 business areas. For key transportation hub, we are now providing comprehensive airport transfer services across Beijing, Shenzhen and Guangzhou.

Our international expansion is also gaining traction. We have now established a presence in 9 countries and started the services to the public in 4 overseas markets including Croatia, Qatar, Singapore and South Korea. In the capital of Croatia Zagreb, we realized the first robotaxi commercialization in Europe. In the Middle East, our footprint in Dubai and Qatar continues to expand, currently initiating driverless deployment in Dubai. These achievements serve as proof that our model can be applied smoothly across multiple regulatory and operational environment. Ultimately, creating solid revenue streams.

As for the joint deployment model, we consistently make a significant stride because of our technology leadership, our operational success and our commercial maturity, partners increasingly recognize us as their preferred collaborator. We have seen more partners from both domestic and the international markets join forces with us starting to contribute sizable revenue in Q1. Our robotaxis success is continuously driven by our innovation and execution, which helps us to achieve a large-scale fleet excellent technology and operation and a superior user experience.

Moving forward, we will focus on reinforcing these areas to expand market share and cement our industry leadership. Operating scale fleet with consistent stability is a powerful testament to our technology and operational maturity. As we scale, we are supercharging our growth engine. We continue to build competitive barriers and trust from policymakers and fortify our brand position at the forefront of user main share. Currently, we are accelerating the rollout of Gen 7 vehicles across Toyota, Beijing Auto and Guangzhou Auto, exceeding 1,700 robotaxi vehicles.

At the Beijing Auto Show last month, we debuted our 2027 version of the robotaxi for domestic market. This upgraded version will achieve further BOM cost optimization to less than RMB 230,000. This competitive pricing facilitates rapid scaling of the robotaxi fleet for the years to follow.

Safety has always been the foundation of our company, which is ensured by our technological and operational advantages. Our industry-leading L4 technology vehicle level intelligence and resilient fleet management help us to maintain an unpremised safety. This proven mastery of highly complex scenarios enable our robotaxis to navigate, pick rush hour sense urban areas and bad weather conditions, satisfying surgeon user demand. We have moved beyond a novo experience into a go-to daily transportation choice.

The results speak for themselves, our robotic taxi fares maintain a premium over the entry-level ride-hailing services. Despite this premium service pricing, demand remains exceptionally robust, particularly during peak hours. Notably, our weekly average paid orders so far in May increased by more than 100% compared to the beginning of the year. significantly outpacing industry-wide growth. Beyond that, we are continuously optimizing ground operations from charging efficiency to dispatching agri This, in turn, boosts our fleet utilization and reduced operational costs.

Now let me move to our [ Robotruck ] business. Our Gen 4 Robotruck is slated for mass production in the second half of the year with preproduction vehicles currently rolling off the production line. I'm also pleased to share that in Q1, Robotruck revenues were up 31% year-over-year. This was driven by scaling up long-haul operations. We also strive to expand our addressable market across multiple fronts, particularly into city urban logistics. To this end, we launched our L4 autonomous light truck in April, leveraging our fully automotive grade and fully redundant Level 4 robotaxi architecture.

In terms of Intelligent Solutions, our business we recently renamed from licensing and applications to better reflect our expanding business in this segment. Q1, the ADC, essentially the autonomous domain controller shipments in this segment surged by over 500% year-over-year. This was mainly driven by domain controller deployment in low-speed delivery application.

2026 is off to a strong start for Pony. We have achieved supercharged revenue growth in all 3 business lines without any compromise in safety. Since the first day of our funding, we have been committed to provide safe and reliable autonomous driving services. It is our deepest moat, as it's now the perfect stage for Pony to demonstrate what a decade of rigorous engineering looks like. Our fourfold global taxi revenue growth fueled by accelerated user adoption in domestic Tier 1 cities and revenue contributions from our joint deployment model, both domestically and globally.

Reflecting its powerful commercial momentum, I am now raising our 2026 annual targets that we forecasted earlier this year. First, upgraded fleet target. We are now on a clear path to surpass a fleet size of 3,500 vehicles. which is an upward revision from our initial 3,000 targets. Second, accelerated revenue growth. We are now lifting our robotaxi revenue target higher to more than 3.5x from our previous target of triple. Third, scaling our domestic and overseas process as we continue to accelerate the scaling up in our existing markets, we are firmly confident to expand our footprint to over 20 cities, both domestically and globally.

As an industry leader, our mission goes beyond our own growth. We are here to lead the development of autonomous driving that has sustainable societal benefit. By providing a safe driverless technology that is safe and profitable at scale. We are building the future of mobility that the world can trust.

With that, I'll hand it over to our CTO, Tiancheng Lou, to go over the technology that's powering our leadership. Tiancheng, please go ahead.

Thank you, James. Hello, everyone. This is Tiancheng. Our strong start in 2026 fully proves our solid technology foundation. Looking at our scale, our [indiscernible] fleet now surpasses 1,700 vehicles. and the Q1 robotaxi revenues skyrocketed by nearly 400% year-over-year, hitting all-time high. Building on this robust growth momentum, we are raising our full year target to over 3,500 vehicles and revenue grows to 3.5x from the level of last year. This scaling up is driven by our proven capability to expand rapidly in high-value markets. But the factory interdowntown core of Guangzhou Hydro District and launching Europe's first commercial robotaxi service. We demonstrated our true technology leadership.

Only a tech leader can deploy feed so quickly into these high-value ultra-complex urban areas. Because our technology navigate this environment safely, more users choose to call our robotaxi and more partners want to collaborate with us. It is this demonstrates the capabilities that give us the confidence to upsize over scale. Another clear testament was our performance during a series of concerts called in Guangzhou earlier this month. This event attracts tens of thousand attendees around the stadium. I'm very proud that Pony's robotaxi officially become a government-recommended transportation choice for the peak post concerted cloud.

We put this out because we can master this level of extreme localized demand seamlessly, being integrated into official local traffic plan, proves that the authorities highly trust our cities and operational capability. So ultimately, mastering this stands high-traffic environment demand that sleep in engineering by orders of magnitude. It comes down to 3 core technical pillars, an exceptional training paragraph, robust operational redundancy as well as safe and efficient fleet management.

Many years ago, we realized a critical truth. The public demand a much higher safety standards for L4 drives robotaxis than procurement drivers. This means when human drivers make mistake to try to accept as a normal part of daily life. But if an AI driver makes a mistake, the public trust will be negatively impacted. This understanding shipped over years ago, we could not achieve true L4 by simply learning from human driving data. More importantly, knew we could not sell L4 through a simply scaling law like that large language model, meaning just increasing parameter size and data win.

Turning from human driving data and scaling up parameters can give you a decent L2 driving system, but that level of AI is only good enough for L2 assistant driving when a human acts as the backup. It can never work for large-scale L4 robotaxis because they cannot significantly beat human safety level. Driving is very different from AI coding. In coding, the AI does not need to make decisions without a low latency and the first output does not need to be perfect. The AI can try, fail and stitched errors multiple times during our agent freemwork competitors and test environment.

Human expect to see a final result except the must run of trial and error. But for AI lever, the model output must be instant and correct on the first track. Therefore, we started using reinforced learning and word model years ago. Today, this approach allows our robotaxi to drive much safer than humans, especially in complex areas. This early gives us a massive first move advantage allowing us to rapidly deploy our robotaxi in high-value market globally.

However, for true L4 vehicle, achieving safety just at the algorithmic level is not enough. It system downgrade and cost accident or simply stop that on a high-speed road to wait for rescue the public will not accept it. That is why every single pony's robotaxi features a full-stack multilayer redundancy architecture for both software and hardware. This gives us through operational capability, if any component fell during that trip, the season stay fully functional, the car will continue to drive safely to the secure spot and over avoiding traffic conjection and rare and crashes.

Furthermore, overcut can drive normally even when there is no network or GPS signal both of which can in drop in urban environment. We also do not rely on high-definition map. For example, even when road layout or markings change significantly or even if we need to drive in the opposite lane our system adapt and navigate safely based on real-time load detection.

We also take any event instantly over cars encrypted with impact sensors to the system note immediately if collision occurs or stop the vehicle right away. We also detect hardware for software failures and network instability instantly to ensure driving safety. We even have specialized water waiting sensors to make sure our costs do not enter the part that could cause damage. At over operation expense, keeping the entire fleet stay becomes just as critical as a single [indiscernible]. To achieve this, with scale or intelligence into citywide [indiscernible], protecting our larger operations through 3 strong lines of defects.

The first line of prevention. We have a dedicated safety team to systematically eliminate risk from the very beginning, we technical design to stop safety issues before they happen, including risks from human errors of cyber attacks. For example, over remote season only provide high-level guidance, they do not control the car. The onboard module on the vehicle responsible for any conditional accident avoidance. This ensures our remote assistant cannot call the accent through wrong input or network delay.

The second line detection. If demand expects and our vehicles end up heading in the same direction. Our smart dispatching system ensure they don't arrive at the same section or one, but rather ride 1 after another. If I love this block are conducted, over system will also detected it instantly and notify the whole fleet to avoid making the traffic worse.

The third line is response. We established dedicated ground sport team. If a vehicle encounter any issue on the road, overrate personnel will arrive at the same within minutes to handle the situation immediately. Being short, our technology makes our operations safe and this last operate build over ultimate mode because we choose the right foundation from day 1 and we now have a unique capability and a first mover advantage to rapidly expand in high-value markets. By the end of this year, we target to expand our fleet to over 500 vehicles across more than 20 cities.

This massive scale will allow us to unlock even greater commercial value while continuing to deliver the first most trusted alpha-rated service both domestically and globally. This concludes my prepared remarks. I will now pass the call to our CFO, Dr. Leo Wang. Leo, please go ahead.

Thank you, Tiancheng. Hello, everyone. This is Leo. I will focus on year-over-year comparisons for the first quarter of 2026, unless otherwise noted. For detailed financials, please refer to our earnings release. 2026 is the year where our commercialization strategy translates into remarkable financial performance. This quarter total revenues reached a record of USD 34.3 million, representing a 145% increase from USD 14 million in the same quarter last year. The triple-digit top line growth was driven by robotaxi revenue growth of 395% and intelligent solutions growth of 246%.

We are also capturing compounding benefits as we extend our Atom driving technology from robotaxi into robo truck and other partners along the value chain. Diving deeper into robotaxis. This segment continues to serve as our core growth engine. This quarter, we reached a record high robotaxi revenue of USD 8.6 million grew by nearly 400% compared with USD 1.7 million in the first quarter of 2025.

As James mentioned, 3 key elements have helped the Pony to achieve a leadership mode in robotaxi operation. These are still fleet, excellent technology and operations as well as superior user experience. one's robotaxi has become a popular service that has captured user mind share. And this is now reflected in our financial numbers. Specifically, our fare charging revenues delivered an exceptional growth of 46%. This impressive increase was driven by several compounding factors. We continue to add more vehicles and expanding into more regions, especially to core downtown areas with higher economic value.

Operating metrics reflect our growing capacity and strong user demand. For example, our weekly average paid order so far in May grew more than 100% compared to January, registered users increased more than 200% year-over-year, and our daily order growth rate continued to outpace the industry average. We'll make this strong growth trajectory even more remarkable is our pricing power. Even after this comp, our effective sale rate per kilometer remains above entry-level pricing on wide heading platforms and is on par with the standard [indiscernible].

Our demand remains robust and is growing at a very fast piece. We believe this is a clear reflection of the superior ride experience and the robust technology we deliver, especially during peak hours and in traffic have downtown areas. On the cost side, we continue to make good progress on both operating costs and the bond cost front. Pony's combined depreciation and operating costs per vehicle are already among the most competitive globally. And this is achieved while operating in the business downtown area during the morning and the evening peak hours and their under most demanding traffic conditions.

By leveraging operational efficiency, we continue to drive operating costs even lower, and are also on track to bring robotaxi bond costs below RUB 230,000 by mid-2021 in the domestic market. Together, these 2 levers, declining operating costs and the lowering bond costs will further enhance our robotaxi margins as we scale the fleet. Aside from their charging revenues, our joint deployment model has started to contribute meaningful revenues with both domestic and overseas partners. Such a model will enable more efficient use of capital in fleet deployment, specifically as a global technology enabler. We successfully launched the first commercial robotaxi service in the city center of [indiscernible] with our local partners, combined with our expanding operations in China, this is a strong testament to the execution of our dual-engine strategy.

Turning to robotruck. Robotruck services revenue grew 31% to USD 10.2 million this quarter. up from USD 7.8 million in the first quarter of 2025. This growth was driven by the addition of more trucks and the expansion of our diversified client base, reflecting increasing demand from downstream logistics clients in the long-haul business. We continue to see our industry-leading auto driving technology, expanding into wider use cases. for example, long haul trucking and intracity logistics. Looking ahead with the launch of Level 4 atoms light truck and again for robotruck we are firmly on track to deliver even better auto driving trucks with lower cost, superior driving performance and wider use cases, expanding into a wider addressable market.

Our Intelligent Solutions segment, formally the Licensing and Applications segment delivered a remarkable growth of 246% and reaching USD 15.5 million in the first quarter of 2026, up from USD 4.5 million in the first quarter last year. This exceptional performance was mainly fueled by strong sales of atoms domain controllers. Such strong growth is yet another testament to the opportunities of our account driving technology. as we empower other customers along the value chain. Moving to cost and margin. Total cost of revenue was USD 28.7 million, translating to a gross margin of 16.2%.

Total operating expenses were USD 63.9 million, increase of 9.5%. On a non-GAAP basis, operating expense were USD 59.3 million, representing a 20.2% increase. Such commitment, especially in R&D, have helped us to maintain our technology leadership and will effectively drive down our bond costs. Loss from operations was USD 58.3 million, remaining relatively flat compared to USD 56 million in the first quarter last year. Net loss was USD 53.5 million compared to USD 37.4 million in the first quarter last year. The increase was mainly attributable to the realization of investment income that occurred in Q1 2025, coupled with the modest increase in operating expenses excluding the impact from this investment realization, the underlying loss amount remained broadly stable.

It's worth noting that the loss from operation margin nallowed drastically, from negative 401% in the first quarter of 2025 to negative 170% this quarter. Similarly, our net loss margin narrowed from negative 267% to negative 156% year-over-year. The narrowing loss margin trends demonstrate our operating leverage driven by the rapid revenue growth and the gradual realization of commercial scale benefits.

Turning to our balance sheet. Cash and cash equivalents, short-term investments restricted cash and long-term debt instruments for wealth management stood at USD 1.4 billion as of March 31, 2026, this compares to USD 1.5 billion as of December 31, 2025. We continue to maintain an exceptionally robust financial position with ample dry powder to execute our strategy. Net cash used in operating activities was USD 74.2 million this quarter compared to USD 54.2 million in the first quarter of 2025, the increase was primarily due to an increase in the accounts receivable resulting from substantial sales revenue increase of auto-driving domain controller, along with the increase of non-GAAP loss from operations.

Capital expenditures were USD 12.5 million this quarter, compared to USD 4.9 million in the first quarter last year. The increase was primarily due to Gen 7 vehicle production for the quarter. and the procurement of vehicle components for future manufacturing and investments in data centers and servers. We believe 2026 will prove to be a defining year for the industry. and we are confident in our ability to outperform the industry in operational and financial execution.

With our solid robotaxi operational excellence, continued strong cost optimization, increasing partner interest and a strong cash reserve. We are highly confident in accelerating our path towards sustainable profitable growth for our shareholders.

I will now turn the call over to the operator to begin our Q&A session. Thank you.

[Operator Instructions] The first question today comes from [indiscernible] with Jefferies.

Congrats on the strong quarter. Just 1 from me. I'd like to ask about the regulatory environment. We've seen quite a bit of movement on the policy side for the robotaxi sector, both in China and overseas. So I was hoping you could share your perspective on how this evolving regulatory landscape is shaping up. And more importantly, how you see it impacting Pony AI's business or your competitive positioning going forward?

This is James, and I'll take this question. So as far as I know, most of the policy discussions, both domestically and globally, are actually centered on the safety operation of robotaxi. As you all know, safety is the cornerstone of the [indiscernible] driving industry. Therefore, I would consider the safety discussion and the result of standardized safety or even higher safety measures are beneficial for the long-term stable development of the industry.

As Pony, we have had many years of experience of successfully operating a large fleet and have the experience working with regulators to have a healthy, more transparent environment. especially in China, we have built a deep trust with regulatory authorities, and we consider that we will continue to work hand-hand with the regulators to safely bring autonomous driving to the public. Back to the safety sales. As Tim mentioned, we have established a full life cycle safety management for both autonomous driving vehicle itself and also the fleet operation. Every vehicle features a fully redundant architecture with fail operational capability, that is our vehicles actually will always sit pro over even during an extreme case of system failure.

Additionally, our fleet management has the capability to detect and respond to any unforeseen issues on the road. The whole system actually serves as a city-wide CTNet to prevent traffic jams and handle real-time road changes. This is actually how we ensure safety and scale. This highly sophisticated and robust safety system and also the safety track record have given us confidence to scale our business quickly. The current policy discussions and policy updates do not have any direct impact on our business. In contrary, we are, as you see, during the prepared remarks, I have actually raised our business targets for the whole year of 2026.

We are continuing to push forward with our Gen 7 deployment, and we are making smooth progress towards our target in fleet size, revenue and operational area expansion. So as I mentioned, there's no immediate impact. And I believe that in the mid- to long term, actually, the trends standardized regulatory environment will play directly to our advantage as we already established as the industry leader. It highlights once again that the complexity of operating robotaxis at scale in dense uber environment, which is exactly we have proven our capability. I think ultimately, these high standards will consolidate the market, filter out the unqualified players and further read entry barrier for the new players. And as a result, it will help the long-term growth of the industry.

With this, I will hand over to the operator.

Next question comes from Ming-Hsun Lee with Bank of America.

Given you raised your robotaxi fleet size to 3,500 by the end of the year and also you raised revenue. Could you elaborate more on the key drivers behind your upward revision for these 2 numbers?

This is Leo. I'll take this one. So the upward revision is definitely showing that we are encouraged by our strong commercial momentum and especially the result of Q1. To be honest, this is actually moving faster than we expected, and it's reflecting many all areas in our robotaxi business. For example, we are seeing our domestic operations are accelerating. We are seeing the pickup in revenue in paid order volume and also in the user basis in all Tier 1 cities in China. This is really a reflection that we are providing a qualified service nonstop in Shenzhen and in Guangzhou, and we are attracting more and more repeated users because we can provide a service even during peak hours with consistency even during complex scenarios and that eventually translates into more revenues.

And the other point is how we make the UE breakeven milestone in Guangzhou and Shenzhen. This also serves as a proven case for future possibility. And that's why we are seeing many of the potential partners. Now they have the real interest domestically and internationally to really participate in our joint deployment business model. this could be more efficient use of our capital, but it also means we could deploy more vehicles in different markets. So given all these facts and encouragement, that's why we have the confidence to push our robotaxi revenue growth target even higher to be 3.5x of 2025 and also our [indiscernible] to be 3,500 vehicles by this year-end. And now I'll go back to the operator.

The next question comes from Wei Hong with Deutsche Bank.

So I have a question on [indiscernible] recently lost or electric, the resin over show, and we'll explain the strategic considerations for launching this platform in the commentation for [indiscernible].

This is James, and I'll take this question. As you consider the company vision since our founding has always been autonomous mobility everywhere. And to us, the work everywhere actually has 2 implications. One, expanding our presence across both domestic and overseas markets. And the other is scaling our technology across different vehicle platforms for both the passengers and the fleet transportation. So the launch of our L4 autonomous light truck actually aligns perfectly with our vision and our ambition. In the logistics sector, the value chain actually sends long-haul trucking, urban logistics, and the last mile delivery.

We already established a robotruck division that working on the long-haul logistics. And for the last mile delivery, we are not directly working on it, but we actually have already becoming the leading ADC provider. So the current -- the recent launch of Level 4 light truck is actually serves the purpose of completing 1 key segment in our full logistics portfolio. The platform for the Level 4 light truck also shares a nearly identical software stack as our robotaxi. It can also fully utilize our existing operational infrastructure, such as remote assistance, the ground support networks and even the cleaning charging facilities. This unified architecture creates a powerful synergy, it can actually further flesh out our light truck operating cost by half compared with the human-driven light truck fleet.

Also, we can actually lower the operational overhead of our robotaxi service because we can share a lot of their background support. In terms of the current status, we are developing the level for light trucks, and it's already well underway. For example, we codeveloped this level electric light truck with CATL, and we are establishing a solid pipeline with some of the leading logistics companies for the future application of those trucks. In addition, we all started discussing with the regulators on the licensing front and also on the fleet management. So we expect the Auto Light truck to begin scaled operation early next year.

With this, I'll get back to the operator.

The next question comes from Ting Song with Goldman Sachs.

My question is on the technology part. So regarding the VLA visual language action model in autonomous driving, could you please share more on Pony's strategy and your future expansion technology path. Do you think the language part is still necessary as we recently noted some supply chain there start to remove the language from their models.

This is Tiancheng. I will take this one. Let me start from saying that the core of driving its understanding of the intention of other road users and respond appropriately. By [indiscernible] our intention layer into our onboard model training, we generate different intention combinations, and we evaluate the possibility of all other traffic participants. This design ensure our onboard model always like the [indiscernible] and have a plan ready for any event, even for low possibility adjucatons.

We believe the language is not the essence of driving. Also language models take too much computer power for a car. Instead, we believe intention is the real core for driving. When human drives they think about the intention of other cars, not natural language. Crucially, this intention data is hard to get from simple road testing we make generate it by 1 model. We believe large language models or language layers do not help on the [indiscernible] side, where old model and the generative data essential for training. In fact, top driving and large link models to very different tasks.

Our large language model agent like a coding tool does not need to have very low latency. It does not need to be perfect on the first trial. It works in a low-cost environment where it can try fail and fix mistakes inside the testing box. But the driving has the room for mistakes. If you make a mistake, it is an accident. Therefore, our tolerance for air health nation is 0. To solve this, we build a virtual driving environment in our model. This allows the system to try and fail during the training stage. not on real load. During the real-world inference stage on the car, our model does not pick the single highest possibility paths. Instead, it chooses actions that ensure safety under any our ability.

With this, back to the operator.

The next question comes from Jeff Chen with Citi.

How should we think about the balance between sustaining its high-growth trajectory and your increasing strategic investments, especially when you are revising of the full year target.

Jeff. This is Leo. I'll take this question. Yes, we have a very good Q1 results. which prove that our robotaxi commercialization strategy dual engine strategy is translating into accelerated top line growth. And as you can see that our top line growth is actually outpacing our expenditure, which resulted in our operating loss margin narrowed quite a lot this quarter. Given all this momentum, we are confident to raise our full year business target so that we can achieve even higher growth trajectory, which I think is really important for any growth company. In the meanwhile, we need to make strategic increased investment in certain areas, tends to keep our advantages in the industry. using example is we are actually on track to decrease our total bond cost to be less than RMB 20,000 in the domestic market by mid-next year.

Through our R&D works and deepen collaboration with our OEMs. And we think this definitely will be a payback for our future deployment and will attract more joint deployment business model partners. So I think that this is a balancing regarding, again, the expenditure and also investment versus the trajectory of our growth. We are definitely putting the growth trajectory as our highest top priority. But we will always follow a value-driven and disciplined approach for these front-loaded expenditures. Thank you.

I'll go back to the operator.

The next question comes from Purdy Ho with Huatai Securities.

Management, congratulations on the solid results. I'd like to focus on your international expansion strategy given the recent commercial traction we are seeing overseas. Could you provide more colors on your road map for global fleet expansion? Specifically, as you are evaluating different markets such as the Middle East, Europe and Asia, what it takes your prioritization across this region.

This is James. Let me take this one. as my answer to the last question, our company vision is autonomous mobility everywhere. And you can see that global expansion has always been part of our strategic efforts. Our dual-engine strategy is rapidly accelerating our global expansion. As more international countries introduce regulations in supporting autonomous driving and also there's many more partners who want to work with us. Because of these 2 factors, we are seeing actually tremendous growth opportunities abroad.

In fact, several international markets have already started contributing to our -- contributing sizable revenues to us in Q1. So it's actually, we're capitalizing on this window because our technology and commercial operations in China's Tier 1 cities have already given us extensive experience in handling the most complex urban environments, and also we have already achieved UE breakeven in Shenzhen and Guangzhou. So this proven technical capability and cost advantage. Also because of the overseas policy opening. This is actually the underlying driving force for our accelerated global efforts.

In terms of our international footprint, we are actually, as you mentioned, scaling quickly across all these key regions. We have now established our presence in 9 countries and started local taxi services to the public in 4 overseas markets, including Croatia, Qatar, Singapore and South Korea. In Europe, we partnered with Uber and [ Vern ] to launch the region's first commercial robotaxi in [indiscernible]. In the Middle East, where advancing fare charging services in Doha and initiating fully driverless operations in Dubai.

In Asia, we have deployed public robo taxi services in Singapore and currently are conducting robust testing in South Korea. So certainly, moving forward, we'll continue to collaborate closely with all the local regulators and our trusted partners to accelerate our commercialization. We'll certainly double down on our investment and fully committed to expand our footprint to over 20 cities worldwide by this year. With this, get back to the operator.

The next question comes from Eugene Hsiao with Macquarie Capital.

In the earnings release, some of the CapEx in Q1 was for stock building of [indiscernible] I'm wondering if you could please update us on if there's any material input cost impact for rising component costs. And I think Neil mentioned earlier that we're still on track for the [indiscernible] cost reduction to reach 230,000 by next year. So what areas are we targeting to reach this target?

Yes, I'll take this question. This is Leo. Thank you for asking this question. In terms of bond cost reduction, we have always been using a holistic approach, meaning we are looking into all aspects regarding the base vehicle regarding the auto-driving hardware to get the overall bond costs down along the road. So I think several factors will drive down our future bond costs.

First of all, we are deploying more and more vehicles. Our vehicle total fleet size will increase with a larger volume, especially with more and more deployment partnership coming in, we could give a more quantity order to our suppliers definitely they'll help us to negotiate with the pricing from our suppliers. Second is now we already have our Gen 7 vehicle on the street and accumulating millions of kilometers giving us real data showing where we can refine our system where we can simplify our system where we can optimize our system.

Based on these real data, definitely we can do our R&D work to further cut down our bond cost. Of course, the supply chain itself has a certain uncertainty. However, Pony has been dealing this uncertainty along the years. So for this year, for example, the memory, of course, there is certain shortage. However, we act quickly last year to secure the supply for memories. So again, this is showing our capability on handling these shortages. That's why we are very confident to hit that bond cost target by mid-next year. Thank you.

As there are no further questions, I'd like to turn the call back over to the host for closing remarks.

Thank you once again for joining us today. If you have any further questions, please feel free to contact our Investor Relations team. We look forward to speaking with you in the next quarter.

This concludes today's conference call. You may now disconnect your lines. Thank you.

Hello, ladies and gentlemen. Thank you for standing by, and welcome to Pony AI Inc.'s Fourth Quarter and Full Year 2025 Earnings Conference Call. [Operator Instructions] As a reminder, today's conference call is being recorded. and a webcast replay will be available on the company's Investor Relations website at ir.pony.ai under the News and Events section. I will now turn the call over to your host, George Shao, Head of Capital Markets and Investor Relations at Pony AI. Please go ahead, George.

Thank you, operator, and hello, everyone. We appreciate you joining us today for Pony AI's Fourth Quarter and Full Year 2025 Earnings Call. Earlier today, we issued a press release with our financial and operating results which is available on our Investor Relations website and an earnings presentation, which we will refer to during the conference call can also be accessed and downloaded on our IR website.

Joining me today on the call are Dr. James Peng, Chairman of the Board and CEO; Dr. Tiancheng Lou, CTO and Dr. Leo Wang, CFO of the company. They will provide the prepared remarks followed by the Q&A session.

Before we begin, please refer to the safe harbor statement in our earnings release. which applies to this call as we will be making forward-looking statements. Please also note that we will discuss non-GAAP measures today, which are more accelerated explained and reconciled to the most comparable measures reported under the GAAP in our earnings release available on our Investor Relations website and filings with the SEC and the Hong Kong Stock Exchange.

I will now hand over to our Chairman and CEO, Dr. James Peng. Please go ahead.

Thank you, George. Hello, everyone. Thank you for joining our earnings call. 2025 is an amazing year for us. This was actually defined by multiple remarkable milestones. First, our top line growth significantly accelerated. Looking at Q4 last year, our Robotaxi revenues surged by 160% year-over-year and fair charging revenues, sky rocketed by over 500%. And Second, since our Gen 7 Robotaxi's [ reveal ] last April, we moved straight into mass production and commercial deployment. Our fleet has now surpassed 1,400 units. Third, we are expanding our footprint, launching services in new cities in both China and globally. This has massively broadened our reach. In fact, we've now crossed the 1 million user mark in China alone. Fourth, we've proven our business model actually works. We achieved new breakeven in both Guangzhou and Shenzhen, and we will replicate this success into more markets.

Looking ahead at 2026, it will definitely be a year of hyper growth for Pony. We are writing a perfect wave of industry momentum built on five pillars fully driverless technology, policy support, mass production, large-scale operation and ecosystem maturity.

Last year, we used China's Tier 1 cities as a strategic blueprint to deploy Gen 7 Robotaxis. From official [ reveal ] to mass production, regulatory validation and rigorous testing, we achieved commercially fully driverless operations within just 6 months. Last quarter, we set our Robotaxi fleet target to over 3,000 units for this year, boosted by the financial fire powder from our successful Hong Kong IPO and also with the Toyota bZ4X Gen 7 model already in SOP. We now have greater visibility and are confident in actually even exceeding this target.

Our excellent virtual driver is the key to support this confidence, proven mastery of highly complex urban scenarios and a superior safety record have earned deep trust from the policy makers and the partners driving rapid user adoption. This directly translates into positive UE. Since we hit the UE breakeven in Shenzhen last month, growth momentum continues with this March, we are seeing peak daily revenues of RMB 394 per vehicle and daily orders at 25 per vehicle.

We will certainly replicate this success globally. By year-end, we plan to deploy Robotaxis in over 20 global cities. As a go-to partner, we have forged strategic alliance with industry leaders like Tencent and Uber. Together, we will propel global expansion, powering accelerated top line growth and more than tripling our Robotaxi revenues for 2026.

Let me elaborate how we will drive this type of growth. We are executing dual-engine strategy. That means we are all in on both China and the global markets. Our proven business model in China gives us a solid foundation to replicate the success internationally. And we are already seeing great results that provisions us for our next growth phase.

In China, we have earned a clear leadership across Tier 1 cities scaling further and pushing deeper into busy downtown areas. Take Shenzhen for example, our Robotaxis satisfied surgeon demand in traffic hubs such as [ Nansen and Bauan ] during Chinese New Year. The paid orders in the first 2 months this year alone have already surpassed that of the whole year 2025 in Shenzhen.

We also entered university town in Guangzhou, the business campus zone in Southern China. This sets the stage for more launches in multiple cities across the Greater Bay Area.

In March, we also entered [ Hangzhou and Changsha ] to driving top Tier 2 cities. This is just a start, and we will have more cities to follow soon.

Now turning to overseas markets. Our presence in Europe, the Middle East, East Asia and Southeast Asia now covers a population of 100 million. We're aiming for nearly half of our 20 [ CD ] target to come from overseas by the end of this year. Recently, we teamed up with Uber and Verne, which is a [indiscernible] [ MAC ] group company to enter Croatia working together to launch Europe's first commercial fair charging Robotaxi service. In the Middle East, we rolled out our first fair charging service with Mowasalat Karwa in Doha, and we are gaining up for fully driverless operations after the approval later this month in Dubai, UAE. In Singapore, we have launched public reveal of autonomous driving services with ComfortDelGro. We are confident overseas revenues will grow rapidly in 2026.

Ecosystem maturity is a critical pillar in executing our dual-engine strategy. Our successful business model makes us a go-to partner partners are now actually lining up to join our joint deployment model. Essentially, it's a model that they will fund vehicles and we can share success together. This will empower us to achieve fleet acceleration, reduce cost and capital efficiency.

We have got a robust pipeline of new partners ready to jump on board. Toyota is the first to adopt our joint deployment model. Their [indiscernible] bZ4X Gen 7 Robotaxi will account for a significant portion of our 3,000 vehicle target in 2026, and we have already secured 1,000 units. As a long-standing strategic partner, our collaboration with Toyota extends far beyond just manufacturing. Together, we will commercially deploy Robotaxi to drive market penetration.

By leveraging our OEM partners, mature supply chain and extensive after sales service networks, our enhanced partnership with both Beijing Auto and Guangzhou Auto further reduced our vehicle cost. In addition, we will jointly deploy Robotaxi vehicles into more overseas markets. To reach a broader user base, we also partnered with Tencent by integrating with WeChat Mobility, unlocking access to hundreds of millions of users to call our Robotaxi services. We are also deepening strategic partnerships with On-time mobility in Guangzhou and ATBB in Beijing to accelerate adoption of our joint deployment model.

Overseas, our global partnership with Uber enables us to access users across multiple continents, starting from Europe. Our regional alliance strengthened our market penetration with partnerships established with ride hailing platform both and also auto manufacturers, Stellantis.

Now let me turn to Robotruck. Over the past few years, we've made huge technological leaps by using our proven L4 tech stack. It has been translating into commercial breakthroughs. We are now covering major logistic routes connecting industrial hubs, ports and consumption centers across China. To seize the opportunity, we introduced our Gen 4 Robotruck trucks in 2025, reducing the ADK BOM cost by 70%. We target mass production of Gen 4 Robotruck and deploy them this year.

In 2025, we have deployed fully drive Robotrucks at the [ Jiangmen ] port in Guangdong province and tested the [ 1+N ] driverless [ platoning ] in extreme weather conditions in the Northwest China. With this proven pack, we will deploy Robotrucks in more ports and mine haulage scenarios.

Lastly, our licensing and applications business delivered a robust growth. Last year, Autonomous Domain Controllers, the ADC sales actually reached sixfold to the level of 2024. We have also expanding our application scenarios to low-speed deliveries, global sweepers, logistics and human noise robotics. Strong customer demand and growing market recognition of our technology will continue to drive growth.

In summary, we have hit a major inflection point. As we validated business model through 2025 achievements such as fleet expansion, new [ CD ] launches and breakeven. 2026 is posed as a year of hyper growth. We are super confident to triple our Robotaxi revenues, roll fleet to over 3,000 vehicles and deploy Robotaxis in more than 20 global cities.

Powered by our dual-engine strategy, we are speeding towards autonomous mobility everywhere. I firmly believe every effort we make today, we are not only reshape the future of human mobility, but also drive a revolution in transportation. This will be a revolution where safety, efficiency and accessibility redefine how the world connects, commutes and lives.

With that, I'll hand it over to our CTO, Dr. Tiancheng Lou, who will go over our technology strategies. Tiancheng, please go ahead.

Thank you, James. Hello, everyone. This is Tiancheng. Looking back at our journey in 2025, that was a landmark year. We improved the commercial viability of to mobility, achieving positive unit economics in Guangdong and Shenzhen. Today, our fleet has surpassed 1,400 a large growth throughout the year.

Robotaxi is the first of few application of physical AI validated by real-world operations and user adoption. Over four native tech architecture built down years of R&D has earned the trust of policymakers and establish first-mover advantage to capture multiyear growth.

As highlighted in the previous quarter, word models are now the widely recognized a path a domain where we hold a firm leading position with ponied model. But technology is only the foundation. The key to success is who can deliver reliable drives Robotaxi service at scale. I will walk you through how our technology drove commercial result in 2025 across three dimensions: scale, efficiency and user tiers.

First, scale. Through strong execution on production, we surpassed over 2025 fleet target, and this momentum to listen to reach over 3,000 units by the end of 2026. Since middle 2025, we began for the mass production of 2 Gen 7 models with Guangzhou Auto and Beijing Auto, both now ramping up to full capacity.

In February this year, the bZ4X Gen 7 Robotaxi codeveloped with Toyota roof the production line. The strong realization of overall contriving stack enables us to efficiently adapt across different vehicle platforms. This multi-OEM network enables rapid scaling, while strengthening local partnership and broadening of Robotaxi vehicle offerings. This scale is backed by a comprehensive [ ODD ] that validates our technology's ability to generalize our cross diversity of urban environment.

Today, our fully dropped fleet 2/47 in many cities across the group. Serving the public during peak rush hours and severe weather conditions. Achieving this requires regular engineering validation and a breadth of over fleet deployment and ODD reflects the majority and the robustness of our autonomous driving stack.

Overseas expansion further validates its generalization capability. In Croatia, we are operating across a large area in the center [indiscernible], the Capital's urban core handling complex urban traffic rather than limited low-complexity rocks, such ability to deploy in demand environment from day 1 demonstrates both the robustness of our technology and the commercial potential of our global expansion. This gives us strong confidence in reach of the target of more than 20 cities worldwide.

Second, efficiency. We will establish a clear cost advantage, driving a twofold improvement on hardware, over optimized design in Gen 7 Robotaxi effectively lower [ BOM ] cost through adopting more cost-effective components. Our operation over city record creates significant leverage, dramatically reduced insurance fees and improving remotes efficiency of this enable us to scale with positive unit economics.

Beyond that, [ overtech ] is building a powerful operation mode. We have developed a highly generalized AI driving capability to build a comprehensive and scalable operational workflows. This deep know-how makes us to the go-to partners across the mobility ecosystem. It perfectly positions us to execute our joint development model allowing us to scale fleet much faster with better capital efficiency.

The third, user tiers. [indiscernible] enables Robotaxi to serve consistently in high-value, high-difficulty scenarios...

Pardon me. It seems like we've lost connection with our speakers. One moment while we reconnect.

Ladies and gentlemen, thank you for your patience. We've reconnected with our speakers.

Yes, I will continue with the third factor use shares. So our technology enabled Robotaxi to serve consistently in high-value, high-difficulty scenarios exactly in high-frequency rideshare hotspots where demand peak and users are willing to pay a premium. This differentiates our service supports our price strategy and drive improvement. For example in Shenzhen, over 24/7 [indiscernible] Robotaxi covers high-traffic urban zooms such as [ Nanshan ] hi-tech area to fulfill daily community needs. During rushovers, or AI virtual driver navigates not only major main load but also narrow street where commuters actually need to pick up and drop up, providing convenient product coverage that truly address real-world community needs.

In Beijing, during a heavy snowstorm in the early March, getting a ride become a major pain point for users, such as long waiting time and limited availability. Despite the extreme conditions, snow-covered sensors reduced visibility and unpredictable road conditions, all demanding significant high driving capabilities. Our Robotaxi fleet continued our operations throughout the snowstorm, capturing substantial order volume growth during that period.

Beyond handling extreme conditions, we have significant full riding comfort. Over Gen 7 Robotaxi deliver smoother acceleration, breaking and cornering significantly reducing motion sickness, a pain point that user came about most. This underscores the fundamental point technology leadership is not just an engineering milestone, it is the core engine of our commercial success. Outstanding user shares earns user preference and repeat usage organically without rely on discount.

Beyond our Robotaxi business, or proving L4 technology enable us to capture commercial opportunities across the broader automotive industry. Over platform standardization enables 80% of the tax stack share between Robotruck and Robotaxi. We have achieved two full scenario or whether several operational capabilities. Over operations now span from complex highway segment to unique scenarios like port logistics with accumulated managing exceeding 60 million kilometers.

We also unlock synergy for the license application segment through leveraging our advanced autonomous driving domain controller design, capitalizing on the rapid growth trend in low-speed delivery robust reapers, logistics and robotics. We effectively fulfill our customer demand on [indiscernible].

Looking ahead to 2026, we will increase our investment in R&D and AI talent to assess our competitive position. Specifically, we are focused on advancing our PonyWorld model to further strengthen our auto-driving capabilities, reducing bond costs through continued hardware, software optimization and improving operational efficiency to lower per vehicle operating costs. This improvements are designed to fuel faster commercialization, expanding our robotic operations to more than 20 cities globally by year-end and delivering faster revenue growth.

We will over triple Robotaxi revenues in 2025. As we demonstrated already, technology leadership directly drives commercial performance, and this investment will further widen our advantage.

This concludes my prepared remarks. I will now pass the call over to our CFO, Dr. Leo Wang, for a closer look at our financial results. Leo, please go ahead.

Thank you, Tiancheng, and hello, everyone. This is Leo. I will focus on year-over-year comparison for the fourth quarter unless otherwise noted. For full year 2025 and fourth quarter detailed financials, please refer to our earnings release.

2025 marked an inaugural year of large-scale commercialization for our Robotaxi operations. Robotaxi segment continues to act as core growth engine for the group, delivering exceptional top line growth. In the fourth quarter, Robotaxi revenue surged 168% and to USD 6.7 million. For full year 2025, Robotaxi revenues reached USD 16.6 million, growing more than doubled at 129%. This remarkable acceleration was primarily driven by our fare charging service, which we saw Q4 fare charging revenue skyrocketed by 501% with a full year growth rate of nearly 400%.

More importantly, within just 4 months of Gen 7 Robotaxi launch, we are thrilled to see consecutive UE trend positive in both Guangzhou and Shenzhen, two most valuable cities in China. This milestone was built on to unique pillars that are exceptionally difficult to replicate from others.

First, our clear cost advantages in both vehicle and the Robotaxi operations. Second, our exceptional AI driving capabilities, being capable of navigating highly complex urban environments 24/7, we deliver a consistent, reliable and high-quality service, which helped us to capture the robust user demand.

With the foundation of positive UE and as vehicle density improves, we are seeing a clear network effect, improving fleet density shortens wait time, boost utilization rates and drives the number of orders per vehicle. This in turn enhances overall passenger experience and further stimulate ride hailing demands. Specifically, year-to-date of 2026, our users have nearly tripled year-over-year and reached 1 million. In February, we successfully delivered unique economic positive in Shenzhen with an impressive average daily orders of 23 and RMB 338 average daily net revenue on a per vehicle basis.

As a matter of fact, this strong upward trend is continuing right now. In March, we hit a new daily peak of RMB 394 net revenue and 25 orders per vehicle. More excitingly, our paid orders in the first 2 months of 2026 in Shenzhen, have already surpassed the entire order volume for the full year of 2025.

Looking ahead, we remain highly confident in the growth trajectory of our Robotaxi business. As James mentioned, our 2-engine strategy will drive rapid expansion into more than 20 cities in China and overseas. We are confident that our Robotaxi revenues will at least triple this year.

Simultaneously, we are enhancing our revenue quality by adding high-margin recurring revenue streams through Robotaxi joint deployment with our partners such as on-time mobility.

This model will lower the CapEx requirement on initial fleet deployment from our end and can also give us leverage to expand faster and more efficiently into new regions.

From a technology perspective, as Tiancheng mentioned, our advanced AI driver capability directly empower a premium user experience, providing safe, reliable, smooth and efficient rights for passengers. This superior experience strengthens our pricing power and deepens user mind share, which can further boost top line growth.

On the cost side, we have proactively secured procurements for critical vehicle components and hardware, including high demand memory modules. Therefore, we expect the minimal impact from supply chain pricing fluctuation. Meanwhile, with greater fleet scale continues a reiteration and the deepening OEMs collaboration. We have high confidence in continuously reducing our vehicle bond cost and further improving operation efficiency.

Together, the high gross margin profile of the Robotaxi segment is fundamentally elevating our revenue quality and actively contributing to the group's future profitability.

Now let's move on to Robotruck. By leveraging our proven Robotaxi stack. Our next-gen robot truck achieved a 70% cost reduction. Furthermore, our transition to EV trucks also will continue to drive down per kilometer operating cost.

Looking ahead to 2026, our shared expertise in Robotaxi will accelerate our Robotruck mass production, enabling us to begin deployment within 2026.

As we aggressively deepen our route coverage across major logistic corridors and expanding into more scenarios such as dedicated lines and port operations. We expect to see accelerated growth in revenues beginning in the second half of this year. With lower structural costs, continuous tech breakthroughs towards driverless operation we are perfectly positioned to capture the free markets immerse value.

Thirdly, our licensing and application also delivered impressive growth in 2025. We are seeing strong client demand from our Auto Domain Controller, ADC product, which the ADC volume grew to 6x of the level of 2024. Looking ahead, we are seeing a solid order pipeline from existing customers and are actively expanding into new use cases.

On the overall profitability front, we achieved a historical financial milestone in the fourth quarter by achieving a first ever quarterly GAAP level net profit. This historical revert to profitability was primarily driven by the gains from our strategic equity investments, which strengthens our broader ecosystem positioning and unlock business synergies. In 2025, our expense were slightly widened. This was a deliberate front-loaded investment to accelerate Gen 7 mass production, expanding into new cities and strengthening our tech stack. Such investments are already starting to drive strong top line growth.

In the area of AI, we anticipate continuous investments into AI technology and the talent to help us secure a long-term competitive edge. Beyond the technology benefits, our joint deployment model will also be a powerful lever for CapEx efficiency. By collaborating with partners to share the initial investment, we are able to scale our fleet rapidly while maintaining a lean balance sheet.

Looking ahead, we expect the revenue growth to outpace the growth of operating expenses as we capitalize our fleet scale and capture the virtual cycle of positive UE.

Finally, we closed the year with a highly robust balance sheet with substantial cash reserves of over USD 1.5 billion, following our successful Hong Kong IPO. This solid capital position gives us the firepower to invest decisively into R&D, SG&A and go-to-market capabilities. We are confident that the stepped-up investment will accelerate our pace on large-scale commercialization and deliver faster revenue growth in 2026.

Looking ahead, we are crystal clear on our strategic priorities, tripling our Robotaxi revenue. expanding our fleet target to over 3,000 vehicles and deploying Robotaxis to more than 20 cities globally by 2026. We have ample dry powder to support these initiatives, and we will drive progress through our dual engine growth strategy. Combined with our joint fleet deployment model, that optimize capital efficiency, we are well positioned to accelerate these targets and turn our operational momentum into sustained profitable and long-term growth for our shareholders.

I will now turn the call over to the operator to begin our Q&A session. Thank you.

[Operator Instructions] The first question today comes from Ming-Hsun Lee with Bank of America.

This is Ming from Bank of America, and thank you for giving me the opportunity to ask a question. So my only question is that since you already have a target over 3,000 Robotaxi fleet by the end of 2026. Can you share your production ramp-up and the deployment plan. Now that you have achieved the UE breakeven in Shenzhen and Guangzhou, how do you think about the future UE trajectory? .

This is James. I'll take this one. So in my opinion, I think the hitting the UE breakeven is a huge win for the whole industry, not just for us, improves that our technology actually works in the real world. It also shows that Robotaxi is not just feasible but profitable at scale. Right after the UE breakeven in Guangzhou, we did it again in Shenzhen. This shows that our model is actually replicable and we achieved this breakeven by focusing on service value, not discounting.

What we have seen is that in the regular front -- regulatory front, currently, we do see there's a policy tailwind to support the whole industry. In China, there's coordinated efforts between the central and the local governments to bring Robotaxi services to many cities. And in our existing markets of the Tier 1 cities, I also have seen their small licenses were issued to facilitate a larger fleet.

Globally, many countries learn from the progress of -- in China and the U.S. to clear the policy hurdles and come up with regulations to support accelerated deployment. So the regulatory momentum gives us confidence to actually to replicate our current success in many more markets, both globally and in China. Therefore, to capture the market, two things we are focusing this year. One is ramping up production and the other is launching Robotaxis in many more markets. On the fleet ramp-up over the last 2 months, we've been focusing on producing the Toyota bZ4x. And then we're also continuously producing more vehicles with Beijing Auto and Guangzhou Auto, with all three vehicles, we're confident we'll hit over 3,000 units by the year-end.

Then on the fleet expansion, we're pushing deeper into the downtown hubs. And secondly, we are also expanding into new cities such as Hangzhou, [indiscernible] and many cities across the Greater Bay Area.

Thirdly, in terms of UE, since the fares in China is relatively low compared with many of the global markets. And in China, we already delivered positive UE and we're continuing improving the UE. We expect better earnings in our existing markets and definitely a lot better margins overseas. We plan to expand this year into 20 cities, which can give us a very strong first-mover advantage.

Our joint deployment model will also lower our CapEx expenditures, which can help us to accelerate the fleet growth and at least triple our Robotaxi revenues this year. With this, get back to the operator.

The next question comes from [ Tim Shau ] with Morgan Stanley.

This is Tim from Morgan Stanley. I just have a follow-up question about pointing the latest operational footprint based on the dual engine strategy that James just mentioned. So regarding Pony's expansion strategy to enter over 20 cities this year. Could you share details about which CDs you plan to enter? And what's the split between China and the overseas market?

And separately, with geopolitical tensions as grading in millions, are you seeing any challenges or headwinds to your operation? That's my question.

This is James again. I'll take this one. Strategically, we're definitely using our success in China as a blueprint for our global expansion because our technology and the business model are proven, we can replicate quickly broadly in the global markets.

In fact, we expect nearly half of the 20 cities we're targeting in this year to be overseas, spending in Asia, Europe and the Middle East.

In terms of go-to-market strategy, we are teaming up with industry leaders to improve our joint deployment model, which can greatly reduce our CapEx expenditure. This helps us scale efficiently, while at the same time, building a strong local networks. We are already launched in [indiscernible], Doha, Dubai and Singapore, partnering with global giants like Uber, [ Boat ] and [ Stellantis ]. One example is the -- together with the Uber and [ Verne ], we actually have launched the first commercial Robotaxi services in Europe.

Looking ahead, we're certainly exploring more European cities and also doubling down in Asia, such as South Korea and Singapore.

Regarding the last point of your question on our efforts in the Middle East First and foremost, risk rating remains to be our high priority. So far, we haven't seen any material impact to our business from the current geopolitical tensions. We are still charging along with our efforts in the GCC region. We expect to roll out fare charging services with Mowasalat in Doha, Qatar, and we are getting ready for fully driverless operations in Dubai after approval later this month. I'll get back to the operator.

The next question comes from Leo You with CLSA.

My question is on technology. So road model and autonomous driving fact are now operating in multiple cities across different countries. So could you please tell us more how does the technology generalize new environments where the conditions would be very different from China? And what role does the growth model play in accelerating your expansion plan?

This is Tiancheng. I will take this one. First and foremost, the key insight is that driving is about the interaction and negotiation with the agent around it. There is no difference whether you're in Guangzhou in Shenzhen or in [indiscernible], differently to the country is essentially a different combination of similar scenarios. What varies is the probability distribution, not the fundamental nature of the challenges. Some example [indiscernible] are reckless [indiscernible] without checking mirrors for [indiscernible] in the road. This corner cases occur everywhere. Our technology has been already validated in the most demanding conditions, operating at scale across all peak hours, all weather in dense urban courts in China major cities. This means that when we enter a city like [indiscernible], we are not starting from scratch. We are now deploying a system that has already mattered a superset of the scenarios that we encounter. This is why we can operate directly in that [indiscernible] urban [indiscernible], which carries significant commercial value.

Regarding the second part of your question, over [indiscernible] model plays an important role in accelerating this process. It enable us to model the interaction and the negotiation dynamics between our vehicle and surrounding agents and to generate large-scale simulator scenarios that reflect the specific traffic patterns of our new market. By reinforcement learning within this stimulating environment over system continuously improve, it's driving policy, allowing us to validate and fine-tune efficiency without leading to collect massive amounts of data in a new city.

The enablers for reaching 20 cities are clear. Our market OEM network provide locally suitable vehicle platforms. Our operational playbook from remote system to fleet management is highly standardized and reputable. Our technologies brought ODD coverage means we can operate in compact urban environment, not just limited low difficult rocks. Together, this gives us strong confidence in achieving our target of deploying Robotaxi services in more than 20 cities worldwide by year-end 2026. With that, back to operator.

Next question comes from Xinyu Fang with UBS.

Congrats to the solid results. My question is about the drilling deployment model. For vehicles only plan to add this year. Can you elaborate a bit more on how you will apply the deployment model? And how should we think about the benefit of this model for the company and our value chain partners.

Yes. This is Leo. I'll take this question. So as you can see that we have hit a critical milestone of UE breakeven in Guangzhou and Shenzhen. After that, we've been seeing a lineup of partners in the whole ecosystem that wants to join the Robotaxi market, and we are there go to choice. So in this joint deployment model, our partner fund the vehicle CapEx and start tapping into the whole Robotaxi value chain.

For example, ground operation vehicle maintenance and charging. We consider this is a win-win situation for both of us. Our partner gets growing revenue from deployed vehicle, and we essentially are in the asset-light model to expand our fleet rapidly.

In this year, we expect nearly half of our new vehicles are coming through this model and which is led by Toyota. Not only we improve our capital efficiency in this -- in our expansion through this model, but also it will create an additional revenue stream through recurring rent income from the form of revenue sharing or AI driver license fee.

So this revenue stream, combined with our self-owned fleet fair charging revenues will help us to achieve more than triple revenue -- Robotaxi revenue in 2026.

And beyond the current lineup of our partners such as total on-time mobility and ATBB, we actually expect even more partner will jump on board this year. And I will get back to the operator.

The next question comes from Purdy Ho with Huatai Securities.

Regarding the 1,000 over taxes already contracted with Toyota. How are you trying to deploy these vehicles? And do you expect any future scaling up or strategic initiatives with the Toyota down the road?

This is James. I'll take this one. So in terms of Toyota, I consider is as not just a partner. They've actually been with us since the 2019 as our largest strategic shareholder. The relationship between us goes way beyond just an auto supplier. It's actually a deep strategic long-term collaboration.

In terms of the mass production of Robotaxi vehicles, we have jointly launched several Robotaxi models on the Toyota platforms since 2019. And in this year 2026, we are adding 2,000-plus new vehicles and nearly half will be the new Toyota bZ4X Gen 7 vehicles.

This model is jointly developed with Toyota Motor company and GEC Toyota. The best production is already live on Toyota's assembly lines. There certainly is a great synergy between us, their manufacturing capability and top line platforms are blend perfectly with our L4 technology and operational know-how. So considering our collaboration besides the jointly developing vehicles, Toyota is also the first partner to adopt our joint deployment model. Funding the fleet to help us scale capital efficiently. This shows their incredible confidence in Pony. And together, we are rolling out commercially starting from China's top-tier cities. With this, I'll get back to the operator.

The next question comes from [ Hu Jian Ding ] with HSBC. .

I've got two. The first question is about competition dynamics. How do you see the other makers getting into the Robotaxi segment?

And second question is about the competitive edges. So market narrative is shifting towards more into scaling with more entrants getting in with Pony's most unique leading advantage?

Sure. I'll take the first one, and I'll hand over to Tiancheng for the second part. Certainly, we have seen that there -- especially lately, there are many announcements about new players already entered or plan to enter the Robotaxi business. Those new entrants, including automakers, ride hailing companies, tech giants or even startups. In general, I think these new entrants to the Robotaxi space validates the long-term potential for our industry. And I certainly very much welcome the new players. Essentially, they can make the whole ecosystem to be even larger.

But in reality, L4, especially Robotaxi is such a complex systems that requires integrated solution. As I mentioned in my prepared remarks, there are five pillars for the Robotaxi industry, namely technology, policy, mass production, operation and partnerships. These five pillars are so intertwined that simply throwing resources and then will not accelerate the development process.

So we -- over the years, we actually have developed a unique advantage on all aspects of the Robotaxi industry. So regarding some of the -- our competitive mode, I'll hand over to Tiancheng to elaborate.

Technically, I do not think the automakers have an advantage in L4 Robotaxi just because they are strong in manufacturing or in L2 systems. The key point is that L2 and L4 are fundamentally different. They are not just 2 point on the same path. L2 as the mouse for intervention increases, the actual rate can increase. Partial automation can create a force sense that the system is almost good enough until it fails in a situation where the human is no longer ready to take over. That is why the L2 part does not naturally lead to L4, especially when you're talking about the driverless fleet at scale.

One unique advantage we have at Pony AI is over long-term investment in word model and L4-native virtual driver training approach. The reason this matters that L4 robotics need to be significantly safer than human drivers. And that cannot be achieved by simply imitating equipment driving behavior. To reach that level of safety, the system has to keep improving through large-scale trials and ours in a virtual environment. Which is why award model essential. In other words, the key to train our L4 virtual driver is building a virtual environment with strong enough time capability especially when it comes through interaction between vehicles. That is also why the L4 requires many years of investment in AI, and it does not improve mainly by collecting more real data.

The second unique advantage is that we have real robotic fleet operation. And those fleets continuously help us see where the word model is still different from the real world. The hardest part of L4 is not the first 90%. This is the last 1%, the long tail and of rare but critical coronary cases. We handle those cases simply it is not enough to look at on recorded trajectory. What really matters is understanding how other vehicle productions may behave across many different intentions with the AI driver. This exactly why award model is so important. Only award model can give you enough coverage of the full combination space of different intentions in chronic cases. And that kind of coverage is what L4 safety ultimate requires.

At the same time, only fully drive this robotic can keep narrowing the gap between the world model and the real world. Real-world Robotaxi operations let us observe and understand actual behavior pattern of vehicles and protection in those scenarios where human driving data and not observe [ induction ] with Robotaxi. This is also aligned with the fact that new players typically can only start with very small fleets.

Regulators understand this logic as well. So they are naturally very cautioned about granting permit at early stage.

To summary, automated interest space confirms the size of the opportunity for L4 Robotaxi is not something you get by L2. Pony AI's unique advantage comes from two things: award model built for L4 and a real Robotaxi feed that continuously help to improve it. Together, they create a close look that keeps both the model and the product moving forward. So with this, back to the operator.

The next question comes from Joe Yang with Nomura.

I would just like to understand how do management view the impact of media launching the open source model for Smart Driving of L4. [indiscernible] this year also [ GTC ].

This is Tiancheng again. For this question, I think the key is to distinguish between our model and the real product. at open source, our driving model can be a good start point, but not the end product. There is still a very big gap between model and the Robotaxi that is commercially deployed safety proven, government approved and operating scale. Closing that gap is exactly where over care advantage lies. To add Pony, overstrain come from years of full stack in-house development and real L4 deployment at scale, nothing includes not just the software or the model but also the vehicle architecture, sensor redundant design, domain controller, operational system validation and the [ compacialization ] capability need to run our actual Robotaxi service.

Yes. For example, sensor consolidation redundant design directly at facilities with manufacturability and BOM cost, OEM partners are also very different from a [ plat-play ] approach, that gives us better system integration, higher reliability and the lower overall system cost.

So we view progress from NVIDIA help move the ecosystem forward. At the same time, we believe the real barrier to entry remains very high. And of course, NVIDIA is an important partner of ours on domain controller, and we maintain a strong collaborative relationship. So with this, back to the operator.

The next question comes from Tianyu Lu with Citic Security.

I have a question. How you plan to allocate your Hong Kong appeal proceeds. And given your accelerated development targets, do you expect any onward regions to the 2026 cost and expenditure.

Yes, this is Leo. I'll take this question. As I mentioned earlier, we have ample cash reserve about [ 1.5 in billion ] as of December 31 in 2025, which was driven by our Hong Kong IPO proceeds, which we get proceeds of more than [ 800 million ]. So this definitely secures long-term capital for to fuel multiple year growth.

And for us, the 2025 achievement in terms of Gen 7 production deployment and UE breakeven made us a clear industry leader. So looking forward, we look for accelerated top line growth to widen our leading position and also push the whole industry towards the next stage. Hence, we need to strategically increase our investments. We've significantly scaled up the number of Robotaxi operating across China's Tier 1 cities, especially in Shenzhen and Guangzhou.

We also recently entered into new cities in China, such as Hangzhou, [ Changsha ] and international in Croatia. And the plan to deploy in more than 20 cities by this year. So in order to support multiple market expansion, we will invest in business development, operation and marketing. And as we scale our Robotaxi deployment, we are expanding our Robotaxi fleet through joint deployment models as well as investment in self-owned vehicles.

And we would also recruit AI talent and invest in AI infrastructure to further improve our virtual driver capability. We think this would allow us to consistently meet the public high expectations for safety, reliability and quality to offer a trusted Robotaxi service.

James already mentioned that this is a critical period to expand market share, which we think it requires necessary investments to solidify our technological and operational modes. We believe the strategic increase in the investment is a value-driven trade-off to secure long-term market leadership. And we believe with disciplined capital allocation and also the benefits from joint deployment model. It will be paid off with much faster growth city expansion and the fleet size. And I also believe this will also lead the whole industry into a much advanced phase. Let me now get back to the operator.

The next question comes from [ Kai Sha ] with CICC.

I have one question regarding the raw materials inflation as memories. So could you share your view how does the inflation impact your production plan and core tagline?

Yes, this is Leo. Again, I'll take this question. Thank you for asking this important question. As I mentioned in the earlier speech, the impact on both vehicle and ADK BOM cost is very limited. So we think this is -- this resilience is driven by our proactive supply chain strategy and inventory synergy with our ADC domain controller business. Through this approach, we actually have secured our memory supply even before the market went into price inflation and the shortages.

So we are very confident that we can fully support this year's Robotaxi production target of over 3,000. And thanks to the supply chain measures and our continuous scaling . We remain on track to achieve a 28% reduction in ADK BOM costs for 2026 compared to 2025 levels. We will carry out an ongoing hardware software optimization, and this will further reduce our overall BOM costs down the road.

There are no further questions at this time. I'd like to now turn the call back over to the company for closing remarks.

Thank you once again for joining the earnings release -- earnings call today. If you have any further questions, please feel free to contact our IR team. We look forward to speaking with you in the next quarter.

This concludes today's conference call. We thank you all for attending today's presentation. You may now disconnect your lines.

Hello, ladies and gentlemen, thank you for standing by, and welcome to Pony AI Inc.'s Third Quarter 2025 Earnings Conference Call. [Operator Instructions] As a reminder, today's conference call is being recorded, and a webcast replay will be available on the company's Investor Relations website at irpony.ai under the News and Events section.

I will now turn the call over to your host, George Shao, Head of Capital Markets and Investor Relations at Pony AI. Please go ahead, George.

Thank you, operator, and hello, everyone. We appreciate you joining us today for Pony AI's Third Quarter 2025 Earnings Call. Further today, we issued a press release with our financial and operating results, which is available on our Investor Relations website. An earnings presentation, which we'll refer to during this conference call can also be accessed and downloaded on our Investor Relations website.

Joining with me on the call today are Dr. James Peng, Chairman of the Board and Chief Executive Officer; Dr. Tiancheng Lou, Chief Technology Officer; and Dr. Leo Wang, Chief Financial Officer of the company. They will provide prepared remarks followed by a Q&A session.

Before we begin, please refer to the safe harbor statement in our earnings release, which applies to this call, we will be making forward-looking statements. Please also note that we will discuss non-GAAP measures today, which are more thoroughly explained and reconciled to the most comparable measures reported under GAAP in our earnings release available on our Investor Relations website and filings with the SEC and Hong Kong Stock Exchange.

I will now hand it over to our Chairman and CEO, Dr. James Peng. Please go ahead.

Thank you, George. Hello, everyone. Thank you for joining our earnings call. I'm excited to share that we have successfully completed the due primary listing on the Hong Kong Stock Exchange under Stock Code 326 on November 6, just 1 year after our NASDAQ listing. With strong support from both international and domestic investors, we secured the largest IPO in the global autonomous driving sector this year raising more than USD 800 million. This significantly strengthens our balance sheet and provides the dry powder to accelerate mass production and large-scale commercialization.

We now expect stronger growth, surpassing 1,000 robotaxi fleet plan by year-end and expanding to more than 3,000 vehicles for 2026. We have already seen the flywheel in action, expanding fleet is driving higher user adoption, shorter wait time, more orders and strong revenue growth. After launching Gen-7 Robotaxi, we have already seen a citywide unit economics breakeven. This, in turn, gives us more room to increase fleet size. The capital we raised also fills our business development, research and development making strategic investments in new markets, new applications and attracting world-class AI talent. All this are set to further propel our technology leadership and the long-term growth.

Our Hong Kong IPO also powers our core mission, bringing autonomous mobility to everyone around the world. We are firmly delivering on this commitment. Earlier this month, we officially launched fully driverless commercial service for Gen-7 robotaxies taxes across Guangzhou, Shenzhen and Beijing. Today, our management team, including myself, actually arrives at our Shenzhen office in a fully driverless Gen-7 robotaxis to host this conference earnings call.

This is more than just a normal ride for us. It actually marks a giant leap in autonomous driving advancement. We are making Level 4 autonomy more accessible than ever to a much broader user base. I'm excited to share a critical milestone our Gen-7 robotaxis have reached city-level UE breakeven in Guangzhou, shortly after their official commercial launch. This is pivotal to validate our viable business model. It not only gives us strong confidence to further scale our fleet, but also attract more and more third-party partners, enabling them to fund our fleet and support our asset-light model.

The scaling up of a fleet is key to our growth as large-scale operational footprint drives efficiency through the economy of scale. Our robotaxi vehicles are moved -- essentially moving billboards. In fact, many new users discover and download our Pony pilot app after spotting our visuals on the road for daily operations. Fleet expansion serves as a highly efficient self-reinforcing marketing engine, facilitating user adoption and strengthening brand recognition.

This creates a powerful upward spiral, more vehicles generate greater visibility, which attracts more users and establish network effects. The results are already evident. Building on that momentum, new registered users nearly doubled within just 1 week of launching Gen-7 from late October, reflecting robust user demand and effective go-to-market strategy.

Now let me highlight some key advances we made in recent months in executing our scale-up strategy. First, we have ramped up production at an accelerating pace since the start of production in the middle of this year. By November, more than 600 Gen-7 robotaxis had rolled off our assembly lines, bringing the total fleet size to be over 900 vehicles. Thanks to the streamlined production process, we now expect to outperform our full year target of 1,000 vehicles, delivery ahead of schedule. This gives us increasing confidence to sustain robust momentum, driving fleet size to surpass 3,000 vehicles in 2026.

Second, in Q3, our robotaxi revenue surged by 90% year-over-year with fair charging revenues delivering over 200% year-over-year growth. This was fueled by rising user adoption across all 4 Tier 1 cities, improved fleet operational efficiency and tailored pricing strategy for diverse user segments. We have seen that the higher order density leads to lower users' average waiting time and, in turn, higher vehicle utilization rate. This allows us to continuously optimize our pricing strategy. Third, we'll continue to expand our operational footprint. For example, in Shanghai, we became the city's first company to launch fully driverless commercial robotaxi operations earlier late July, covering the Jinqiao and Huamu areas of Pudong.

In Shenzhen, we extended commercial fully driverless operations to more and bigger city areas, including Shekou and Overseas Chinese Town. Fourth, we're taking major steps towards scalable mobility...

Excuse me, I believe there has been interruption. Just one moment, please. Excuse me, I've rejoined management. Please continue. Thank you.

Sure. I was talking about the scale-up strategy. So following our collaboration with Xihu in June, we recently forged another partnership with Sunlight Mobility. This alliance reflect growing market recognition of our business model with an increasing number of third parties wanting to fund fleet deployment. This actually enables us to speed up further fleet expansion.

Now let me turn to our global expansion. We are deeply dedicated to advance robotaxi services while strategically expanding our international fleet. Now we have robotaxi presence established in 8 countries across China, the Middle East, East Asia, Europe and in the U.S. We entered a new market in the Middle East, Qatar, through a partnership with Mowasalat in the third quarter. Mowasalat is the country's largest transportation service provider. As part of this collaboration, our robotaxis have recently begun testing on public roads in Doha, the capital of Qatar.

We have also advanced our presence in South Korea by securing nationwide robotaxi permits enabling operation across the countries autonomous testing and operational zones. Our collaboration with local partners continue to deepen. We're collaborating closely with the country's largest transportation service provider to begin road testing. In Luxembourg, we plan to deploy testing vehicles based on the e-traveler through our alliance with the Stellantis. It's a European leader in light commercial vehicles.

This effort will initially focus on vehicles designed for European diverse mobility needs to enable a range of use cases. In addition, we have partnered with global ride hailing platforms that also participated in our Hong Kong IPO. Those platforms include Uber and Bolt. Bolt is Estonia-based mobility company operating in over 50 countries and 600 cities. Built upon our collaboration with Uber, we aim to leverage Uber's robust ecosystem to enter the Middle East and then scale into additional international markets.

Last but not least, we recently released our fourth generation robotruck with production and initial fleet deployment effected in 2026. Featuring fully automotive-grade components, optimized software hardware integration and the transition from internal combustion engine vehicles to electric vehicles. The Gen-4 robotruck delivers a significant more efficient cost structure and a greater energy CV. The new platform fully leverages the technological foundation and operational expertise developed through our Gen-7 robotaxi vehicles.

In addition, we deepened our collaboration with SANY Group and added Dongfeng Liuzhou Motor as a new partner to have multiple vehicles to support our further operations.

To sum up, 2025 is a critical year of mass production and commercialization for Pony AI. We take pride in the progress we have made and are steadily delivering on the promise we have made to our shareholders at the time of our U.S. IPO last year. Our recent Hong Kong listing not only marks a major milestone for our company, but also underscores the promising future of the industry.

Moving forward, we will drive technological innovation and create lasting values by scaling fast, efficient and comfortable autonomous mobility services towards our mission, autonomous mobility everywhere.

With that, now I'll hand it over to our CTO, Dr. Tiancheng Lou to share more about our technology strategies. Tiancheng, please go ahead.

Thanks, James. Hello, everyone. This is Tiancheng. Let me first share my thoughts on our autonomous driving technology stack. From day 1, we believe that full stack integration across software, hardware and operations was the only way to build a truly scalable autonomous mobility. That conviction has been validated again and again, especially for this critical year of getting up.

With the achievement we made, it is clear the over early technology bets help us achieve the leading position and it will further accelerate our future growth. Our deep foresight into tech stack is what is positioning us as a leader in the industry today as we become one of the few companies to operate large-scale fully driverless services. So as early as 2020, we recognized the importance of training based on reinforced learning unit simulation. In that year, we transitioned over tech stack into our work model, which is what we call PonyWorld today.

Through years of R&D effort and real-world validation, our top driving model has evolved into a closed loop training. We achieved unsupervised self-improvement integrations. In recent years, we are seeing the broader autonomous and robotic industry coverage converge on 1 model, validating the approach we adopted today. This tech stack has given us a meaningful headstart, and we are confident that we will stay ahead for multiple years.

Then let me dive into the 3 criteria that put us the frontier forefront of automotive development. First, the high fidelity impact is simulation. This is far beyond the ability to just generate the scenarios and render sensor data. Driving is by nature interactive. The robotaxis actually directly affect how the agent to behave, such as other vehicles and pedestrians to react to over driving behavior must understand and adapt to new situations and the complex physical real time, building true onload interactions.

It enables robotaxi operations that are safe, smooth and social aware. After 10 billion kilometers of test miles that only were generated each week, more than 99% capture vehicle agent detection, while less than 1% or environments such as sensor rendering.

Okay. Second, the ability to reproduce scale and realistic cases. While this long-tail scenario don't occur frequently, they are critical to safety in autonomous driving. More importantly, every scenario must be something that could really happen in the real world, not those use cases with no basic in reality. For the third, the AI-based learning evaluator. This is the real world based evaluation mechanism, driving the multiple object organization problem what is considered as a good driving also changes in various driving scenarios.

Within the training environment, PonyWorld and our virtual driver are continuously evaluate on key driving metrics. This assessment done on real-world data, label data rules. Instead, it used AI empowered model to learn what good driving looks like directly from outcomes. Turning real and simulated experience into a powerful cycle of staff improvements. Our best-in-class word model must meet all 3 criteria to enable truly unsupervised and self-improving closed-loop training. This is critical to realizing large-scale driverless autonomous driving.

And leveraging over full-stack technology as a core strength, I will now turn to how to drive business progress during the third quarter. First, on cost and operational efficiency. We pioneered 100% automotive-grade autonomous driving kits for Gen-7 robotaxis with optimized the design reducing on cost by 70% compared with approval generation. The Gen-7 have been officially operating for public in Guangzhou, Shenzhen and Beijing fully validating our safety standards and operational efficiency. We build on our momentum and deliver further progress, driving by scaled production and enhanced R&D we will already realized an additional 20% reduction in the auto-driving kit form cost for the Jensen platform designed for 2026 production compared with 2025 baseline.

This laid foundation for sustaining the cost fee. Our robust AR fleet management exercise had proven effective at driving operational efficiency. To better identify user demand in hotspot areas during rush of hours, and we enhance our acres for all the dispatch match and scheduling, thereby insurance sustained different -- sustained efficient robotaxi utilization. We have also improved our virtual driver to recognize more and more complex scenarios. This allow us to improve over remote assistance to vehicle ratio substantially on the track to reach 1 to 30 by year-end.

Over service has become the key reason users chose Pony AI robotaxi. After the launch of Gen-7 robotaxis, we have earned a widespread positive feedback and generated great social media platform users. As we deliver high-quality experience, users are increasingly willing to pay a premium for the enhanced effort, reliability and safety of autonomous journey. For right comfort of advanced interactive planning capability intelligence optimized for the frequency and the magnitude of acceleration breaking a selling. This delivers smooth natural motion control to the electronic vehicles and the ridesharing market offering consistent comfort experience for every robotaxi ride.

These enhancements have reflected the manageable improvement for Gen-7 emergency brakes and selling over the past few months. Additionally, has featured a software in cab experience. We also pioneered the innovative smart positioning feature. With NTAP, users can remotely adjust their vehicle position for more convenient pickup and drop. We introduced the voice active features because the purple voice assist allow users to do start chips and control air-condition, et cetera. We will continue to upgrade the cabin into AI-powered mobility terminal. Together, this upgrade creates more accessible and streamlined user experience.

So third, our tech stack is also built for generalization. The alpha native tax architecture allow us to adapt quickly to new markets and platforms. In terms of cross-region generalization, autonomous driving can quickly understand and adapt to diverse traffic conditions around the world. For example, leveraging over high fidelity training environment and valuation mechanism powered by PonyWorld, which over 40 driverless coverage in Pudong district in just a few weeks. In addition, when standing to Europe, the system intelligently identified and key difference in local road conditions, such as unique traffic signals configuration and various scrubbing patterns.

Over technology power across platforms as well. The latest generation robotruck will commence production and operations from next year. This demonstrates our capability to create synergy between robotaxi and robotruck's tech stack. Looking ahead, we will leverage our success Hong Kong listing to reinforce our technological leadership increasing R&D investment and attract top AI talent to advance our robotaxi, robotrucks and new market initiatives. We will continue pushing the frontier of the autonomous mobility and refining what is possible in the transportation

Okay. This concludes my prepared remarks. I will now pass the call over to our CFO, Dr. Leo Wang for a closer look at our financial results. Leo, please go ahead.

Thank you, Tiancheng. Hello, everyone. This is Leo. I will focus on year-over-year comparisons for the third quarter, unless otherwise noted, Q3 2025 was a landmark quarter. We delivered a robust revenue growth specifically with solid progress in robotaxis large-scale commercialization. And now we expect to outperform our full year fleet target of 1,000 vehicles. Moreover, our newly deployed Gen-7 robotaxi fleet have reached a pivotal city-wide unit economic breakeven milestone.

This lay out a solid foundation for further scaling up and the implementation of asset-light business model, which will be further accelerated by our successful Hong Kong IPO capital raise. In this quarter, revenue finished at USD 25.4 million, growing by 72%. This strong performance was primarily driven by the continuous optimization of our robotaxi services and the sustained demand in our licensing and application business. Firstly, robotaxi services revenue reached USD 6.7 million, representing a remarkable growth of 89.5% year-over-year and 338.7% quarter-over-quarter.

Specifically, fair charging revenue continued to deliver a triple-digit growth, surging 233.3%. This was achieved even before the commercial rollout of our Gen-7 robotaxis, supported by a stable commercial fleet of our Gen-5 and Gen-6 vehicles, the strong growth during Q2 and Q3 stemmed from growing user demand in Tier 1 cities in China. Our continuous effort to optimize fleet operation and pricing strategy, altogether, leading to increased fleet utilization and efficiency. This is a testament to growing user recognition and the brand royalty to Pony pilot service.

Going forward, as we follow this strong momentum towards a significant fleet expansion of over 3,000 vehicles by 2026, we expect robotaxi revenue growth to accelerate even further, driving more orders and higher operational efficiency. In Q3, another key robotaxi update is the implementation of our asset-light model for fleet expansion. As we have shown promising numbers in vehicle unique economics, we received strong interest from third parties who are willing to purchase Gen 7 vehicle to run as robotaxi operators. Such partners include but are not limited to leading ride-hailing or taxi operators, for instance, Shenzhen Xihu Group and Sunlight Mobility.

The asset-light model has contributed revenues through technology licensing fee and vehicle sales while giving us further leverage and capital efficiency for further fleet expansion. Aside from strong top line growth domestically, we are also seeing fast growth of robotaxi revenues from overseas market. Moving forward, we expect robotaxi revenues from overseas market to continue to grow. Currently, our robotaxi footprint have already expanded into a country globally serving as a promising foundation in our exploration of the international opportunities.

Secondly, moving to robotruck. Robotruck service revenues were USD 10.2 million, growing by 8.7%. Moreover, as we launched our Gen-4 fully auto-grade robotruck, we expect to reduce the bond cost of its ADK, autonomous driving hardware kit by 70% and reached 1,000 unit scale of robotruck fleet going forward. This new generation of robotruck will powerfully accelerate the progress of robotruck commercialization at scale.

Thirdly, licensing and application revenues were USD 8.6 million growing significantly by 354.6%. We continue to see robust and growing demand of our autonomous domain controllers, primarily from robo delivery clients.

Turning to gross margin. We delivered a significant gross profit margin improvement from 9.2% in Q3 2024, to 18.4% in Q3 2025, with gross profit of USD 4.7 million in the third quarter. This remarkable improvement was firstly driven by our strategic initiatives to optimize the revenue mix; and secondly, by a greater contribution from robotaxi services, which carry a relatively higher margin.

The UE, the unique economic breakeven achievement validates our due focus on go-to-market execution and optimize operational efficiency. Since the launch of Gen-7 commercial operations in Guangzhou, the daily net revenue per vehicle has reached RMB 299. The net revenue refers to the total RMB value generated from ride-hailing services after deducting discounts and refunds. Notably, daily average orders per vehicle have reached 23 fueled by a robust widespread user demand and our operational optimization.

Meanwhile, we have also optimized hardware depreciation as well as operational cost including charging remote assistant, ground support, service and maintenance, insurance, parking and network costs. This will further improve our margin down the road. The total operating expenses were USD 74.3 million, up by 76.7%. Excluding share-based compensation expenses, non-GAAP operating expenses were USD 67.7 million, up 63.7%. The increase primarily reflects the one-off R&D investment in Gen-7 vehicles and the expansion of our R&D personnel critical to securing and extending our technological leadership.

Specifically, approximately half of the increase in research and development expenses stemmed from onetime customized development fee of USD 12.7 million for Gen-7 vehicles. Net loss for the third quarter was USD 61.6 million, compared to USD 42.1 million in the same period of last year. Non-GAAP net loss was USD 55 million compared to USD 41.4 million last year. Looking ahead, we expect to sustain disciplined investment to accelerate large-scale commercial deployment.

Turning to the balance sheet. Our cash and cash equivalents short-term investments, restricted cash and long-term debt instrument for wealth management were USD 587.7 million as of September 30, 2025 compared to the balance as of June 30, 2025, of USD 747.7 million. Around half of this decrease comes from one-off cash outflow, including capital injection to our joint venture with Toyota to support the Gen-7 mass production and deployment. All of the capital commitment in has been completed.

The remaining cash balance reduction primarily reflects our mass production and the large-scale deployment status, including firstly, ongoing operational cash outflow; and secondly, capital expenditure for the procurement of Gen-7 vehicle in Q3 to support our goal of 1,000 vehicle fleet by year-end. For the 9 months ending September 30, 2025, we have an accumulative free cash outflow of USD 173.6 million. With the completion of our recent Hong Kong IPO, we have over USD 800 million cash newly added, providing us with substantial fuel for the next phase of growth. The IPO proceeds will help us accelerate fleet expansion into key addressable markets, further optimize our platform for scale and deepen our R&D investments to further solidify our technology moat. Looking ahead, our mass production momentum continues to strengthen, and we are on track to exceed our full year vehicle target of 1,000, achieving this milestone ahead of schedule.

This acceleration reinforce our confidence in scaling rapidly, and we now anticipate to grow our fleet to be more than 3,000 vehicles by 2026. In addition, we've already transitioned to an asset-light model for a meaningful portion of our new vehicles. This will enhance our capital expenditure efficiency and provide greater leverage for scalable fleet expansion. With the proven operational model and the financial runway from the recent Hong Kong IPO, we are uniquely positioned to accelerate our business plan, turning momentum into sustained profitable growth.

I will now turn the call over to the operator to begin our Q&A session. Thank you.

[Operator Instructions] And the first question comes from Ming-Hsun Lee with Bank of America.

So I just have 1 question. So could the management team give us more update on the fleet size for this year and also the outlook in 2026. For the new vehicles added, what is the fleet deployment plan across different cities?

This is James. I'll take this one. So as you can see that since the launch of our Gen-7 robotaxi, we actually have seen a much faster-than-expected production and the deployment. So for this year, we certainly expect to outperform our previous target of 1,000 robotaxies by the year-end. We certainly expect this strong momentum to continue into 2026, now with a conservative target of over 3,000 vehicles. This is mainly because we have already seen upward spiral with the launch of our Gen-7 vehicles. Essentially, the fleet density creates a much shorter wait time for the passengers. And then that creates a better user experience, and the user experience leads to much higher utilization for our vehicles.

And then we can actually then charge a better pricing. So this spiral really created a strong momentum for us to expand much faster. In addition, we also started experimenting with the asset-light model by collaborating with fleet managers such as Xihu, Sunlight and, certainly, we'll add more partners. This asset-light model allows us to deploy at a much larger fleet with less CapEx. So this is our growth plan.

Then in terms of the fleet deployment plan, we'll build deeper on our existing markets; and at the same time, we'll go much wider to explore some new opportunities. The city-wide UE breakeven for the Gen-7 in Guangzhou, in my view, it's a pivotal milestone to validate our business model. This gives us huge confidence and allow us to deepen our collaboration and our operation in the existing markets, which are the Tier 1 cities in China. This is because, as I already mentioned, expanded fleet size creates upward spiral. But at the same time, we also expand into many more domestic cities and also the overseas markets. We see those for our future growth. Our go-to-market strategy on those markets is that we'll collaborate deeply with the local partners and the local government agencies to establish presence and prepare for our future growth. So stay tuned. I think we'll have great news ahead of us.

With that, back to the operator.

The next question comes from Bin Wang with Deutsche Bank.

I just have 1 question, which is about fare-charging. I'd like to know fare-charging revenue from and another growth in 3Q '25. So what is the outlook for fare-charging revenues as we deploy more vehicles?

Yes. This is Leo. I'll take this question. Yes, in Q3, our fare-charging revenue actually surged even faster. It was growing about 233%. Though at that time, our fleet were still with the Gen-5 and Gen-6 vehicles. So we believe such growth was driven by both the demand side as well as the operational side. On the demand side, we have been continuously to do our effort to improve the whole writing experience and also the user experience. So with this effort, we've seen robust and organic user demand in Tier 1 cities. This is also a signal of strong consumer adoption of our robotaxi service.

Giving you an example that the total registered user was more than doubled year-over-year in Q3. And on the operational side, we have also been optimizing the fleet operation to improve our vehicle utilization and order fulfillment, as Tiancheng already mentioned in his remarks. So for example, we enhanced our fleet dispatching and the deployment, this has consistently reduced our wait time. It's approximately 50% shorter compared to the same period in 2024. And we also continue to expand our pickup and drop-off points to create a much more smooth user experience.

For example, in Shenzhen, now we have more than 10,000 such points. more than 300% increase since the end of June this year. With all this demand side and operational side improvement, I believe we could see sustained strong growth momentum through the continuous fleet expansion with more and more Gen-7 vehicle are into our service. First of all, we expect that our fleet has been growing exponentially from 270 last year and to be more than 1,000 this year and a target of more than 3,000 next year. This scaling up would also create a better network defense, which means shorter wait time and higher vehicle utilization and higher user adoption.

We would also progressively expanding our service area in cities such as Shanghai, Shenzhen, we've already been doing so today. we would increase the population coverage and expanding to more drivable mileages, et cetera, et cetera. With all this being done, I think we can boost the average order value per chip.

Okay. Now I'll get back to the operator.

The next question comes from Kyle Woo with Citi Research.

This is Kyle from Citi Research. And congratulations on achieving the milestone of city-wide breakeven. Could you elaborate more about the assumption behind redeliver even, including the order pricing, value operating hours and the ratio of remote assistance.

Yes, I'll take this question. Like you said, we all believe the city-wide unique economic breakeven is a pivotal milestone for the company and also for the industry. First of all, we achieved this pivotal milestone in Guangzhou City since our Gen-7 vehicle has been put into commercial service. And we always believe China is the largest market of global ride-hailing market. And for the Tier 1 cities, the total TAM accounts for a huge percent of ride-hailing market in China. So achieving this milestone in this market is far more meaningful from a commercial perspective.

Then if we talk about the economics, there's the revenue side, there's always the cost side. On the revenue side, first of all, on the daily net revenue per vehicle. As I mentioned, our daily net revenue per vehicle has hit RMB 299. It's based on a 2-week daily average figures as of November 23, following the launch of our Gen-7 vehicle in Guangzhou. And this net revenue also refers to the total RMB value generated from ride-hailing service after deducting discounts, and refunds.

And in terms of daily orders from this RMB 299 number, it was average 23 orders per day. It's fueled by a robust widespread of user demand. Now let's look into the cost side. So the cost side of the unit economic basically has 2 major components. First of all, it's the hardware depreciation. For Gen-7 vehicle, the annual vehicle depreciation is based on a 6-year useful life. The other major component on the cost side is the operational cost, which includes the charging, remote assistant and ground supporting staff, vehicle service and maintenance, insurance, parking, Internet network cost.

So regarding the remote assistant, we are on track to achieve our on well over 30 And from this milestone that we achieved, we are very confident to capture the China huge TAM. Meanwhile, it also established a strategic foundation for further scale scaling up dramatically and internationally. This not only gives us strong confidence to further scale our fleet, but we also see more and more third-party companies are enabled to fund their fleet and helping us to transition into an asset-light model. So all these together, we believe, will drive our top line growth and also the cost optimization.

Okay. I'll get back to the operator.

The next question comes from Perdy Ho with Huatai Securities.

Congratulations on the results. We've observed a surge in diverse players attempting to the robotaxi operations, particularly the EV makers, right? So what's your take on these new entrants in the Level 4 autonomous driving space? And not so specifically, could you elaborate on the main technical and operational challenges, such as tackling corner cases and fleet management for these new comers?

This is James. I'll take this one. So first and foremost, I think it's definitely, as we see more and more companies announcing that they're going to enter into robotaxi industry, I think, itself is actually a great thing because it indicates increasing recognition and the confidence in robotaxi imminent potential for the large scale of commercialization. As the awareness increase more resources, more companies come in, more resources will pour into this robotaxi industry to actually accelerate its development.

So overall, I view this as a good thing. But on the flip side, the robotaxi industry is actually not a one that any new player can easily enter because as you can see, the fact is that currently, none of the new entrants being OEM maker or being a ride-hailing platforms, none of them have fully driverless vehicles deployed of the open road. So it's clear evidence, this is not an easy industry to be entered. I think there are certainly 3 huge hurdles for any new players. And those hurdles are business side, regulatory side and also technical challenges.

Let's probably look at the business challenges first. Because robotaxi, as you see, it's not just about driving itself. It also has many more aspects such as user acquisition, vehicle production, fleet dispatching, fleet maintenance, such as the cleaning, charging and everything else. So as a leader and the first mover in this industry, we certainly enjoyed the early-mover advantages as we have a much bigger L4 fleet on the road. We generated a better brand awareness. We have optimized the cost on every aspect of the business, as Leo already mentioned in his answer to the last question.

And we -- because of early mover, we also have secured more partners. I think all those are important and it creates a big hurdle for any new entrants. The second hurdle that I want to mention is on the regulatory front because L4 -- robotaxi needs various safety requirements. All the policymakers worldwide have fundamentally will require a much, much higher safety requirements for the robotaxis compared with the traditional taxi. That means in any city, a new player needs to prove its safety step by step before they can expect even into a fully driverless fleet.

Typically, a new player will start with a testing with just a few dozen or maybe even less vehicles. And then once those vehicles prove to be safe, they add more vehicles and then expand operational areas after they can accumulate the safety records. And along the way, they also need to acquire all the required licenses and permits. And this in itself is actually a lengthy process. So overall, the whole process takes time, and this co-starting process cannot be easily accelerated. So that's the second challenge.

The third challenge is certainly, in my view, is on the technical side. And probably for this one, I'll turn to Tiancheng to elaborate.

Yes, sure. So I am Tiancheng. So let me continue from a technology perspective. So as I said in my prepared remarks, we are now saying the broader industry starting to word model, such as robotaxi players and automakers. Essentially, they are all using reinforced learning based on simulation training environments. First and for most, I we started developing reinforced learning for autonomous driving 5 years ago. This gives us an early-mover advantage. Based as one of the most expert company in the world model, we believe that we will continue to stay ahead as more peers follow the same path. So as the world model mature now the human feedback and the real-world data no longer used for further iterations.

So at the stage of the word model and driver could evolve into a cycle. This means we're model training, the virtual driver and at same time, the word more improve sales through feedback of the virtual driver. This sharply reduce reliance on the real-world data. The question will touch on the technical challenge for the meeting the quantitates. Maybe an example here that one of some corner cases. So this feedback to the real-world model and the world model will improve its distribution of the coated the next generation of next version model will be able to create generation testing and also improving the capability of the virtual driving we handle the corner cases.

Okay. So looking ahead, our real advantage lies in ability to validate new technology simply and deploy that scale. So based on our proven track record of scaling robotaxi operations, so we believe we can quickly capture the next wave of innovation. Also, last but not least, our current Hong Kong IPO will further accelerate R&D and the equation cycle, reinforcing our technical leadership and widen our competitive moat.

With that, back to the operator.

The next question comes from Xiaoyi Lei with Jefferies.

I have one as well. My question is about what do you see as the main factors behind the faster expansion of your operational areas and beyond technology, what else do you think really matters? And from the technical perspective, are you using large language models? And if so, how are they helping push L4 upon forward.

Thank you. This is Tiancheng. I will continue to answer this question. I think your question consists of 2 parts. But let me answer your question on first, then we address the other autonomous driving model later. For generalization, I would technically over tech stack financial built for donation. So a good example is that our operational area expansion into new areas in Shanghai, Pudong and Shenzhen in the third quarter, in both cases, it only took us only a few weeks from verifying the safety to truly realizing fully driverless operation to public. There was no need for additional model training.

The key reason that native architecture is built for handling corner cases and to gun cases, while these cases are actually very consistent across different regions. There are really nothing more than things like small obstacles, boxes on load, adaptions causing and suddenly from other cars without looking at the vehicle behind, et cetera. So it's just about the likelihood and the probabilities of each what's happening, so hope that can help us and why the L4 tech stack is by nature build for generalization.

So at this moment, I will say the key to over new area extension, the number of robotaxi vehicles. If we expand to too many areas without adding more cars as we dilute the density. So that is the reason why the speed of operating area expansion cannot significantly faster than that of fleet size. So then let me share my thoughts on the second part, that is a large language model. First and foremost, there are 2 nonnegotiable requirement for L4 onboard value model on compromising safety and also low latency. They are the large language model don't need and that are not designed to meet as well.

So for safety, large language model generally have issues like health which is unacceptable for L4 in terms of safety. And for latency, large language models are optimized for per second. In contrast, L4 optimized for low latency and the ability to run fully driverless chips that both low power consumption and cost efficient. Moreover, large language model overly long human data, which fundamentally limit them to the boundary of the exiting home knowledge inevitably makes them pick up human errors and bad habits from human driver.

So we also extensively build out large language model in the R&D effort such as AI-enhanced human machine interaction, engineering productivity pool for coding and documentation and analysis for the rider feedback for excellent improvement. But however, due to the multiple reasons mentioned above, large language model is financially not build for driving model on board.

So with that, go back to the operator.

The next question comes from Xinyu Fang with UBS.

I have 1 question here. It is currently the only cooperate with multiple OEMs for low taxing manufacturing, including BAIC, JC and Toyota. Does management see potential for improving operating leverage through working with only 1 OEM instead.

This is James. I'll take this one. So the matter of the reality is that in the whole robotaxi industry, local governments and the local residents actually have a strong preference for the local branded taxi vehicles. So that's a reality. Typically, while robotaxi fleet is relatively small, the brand doesn't really matter much. But if we need to deploy a significant fleet size the requirements certainly is no longer true, and the local branded OEMs is much more preferred. So it is necessary for us to cooperate with multiple local OEMs in different regions.

It actually can help us to expand into different markets much quickly, and that's why we are now collaborating with 3 OEMs to produce our Gen-7 robotaxis. It is true that fitting our autonomous driving kit into different vehicles actually posts a huge technical challenge. But on the -- if you look at from the other side, the matter of fact that we were able to standardize our technology and being able to fit our setup into different vehicles, that shows our technical generalization and down the road, it actually can create a huge competitive edge.

So as a result, we can add new models much faster to accelerate our expansion into new regions. For example, in Europe, we currently added the partnership with Stellantis. So with that, back to the operator.

The next question comes from Xuxia Tang a with Guosen.

I have 1 question. Why Pony can't use remote assistant robotaxi when the car is difficulty in stand of remote control or takeover? And what is the technology difference behind that?

This is Tiancheng. I will take this one. I think one of the previous questions I also touched on the model system for robotaxi. So let me elaborate on that risk more detail. First and foremost the over remote assist never control the vehicle through the same wheel or pedal. Instead, they provide remote support and suggestions by responding to service requests. For all the time, the vehicle can independently drive independently make decisions without remote assistance. Assistance only initiatives when a vehicle requests it through the remote driving. so when vehicle receives the system response, the onboard driving system will still make timely decisions based on the actual situation. The cost of vehicle never wait for remote command to act.

So it remains safe operation without any dependence on network latency. So one typical example of remote assistance is the situation of a temporary traffic control. In such cases, the system may request remote assist, which can provide high-level suggestion to confirm the cost decisions, navigating service scenario. But also, as I mentioned, we have continued to improve the AI algorithm and also leverage the general AI capability to recognize more and more complex context. This allows us to improve remote vehicle ratio in the third quarter to reach 1 to 30 by year-end.

I hope that answers your questions. So back to the operator.

The next question comes from Serena Lee with China Securities.

This is Serena Lee from China Securities. As far as we know, some countries in the Middle East have issued fully driverless robotaxi license recently. What's our view on that? What overseas to

Sure. This is James again. Let me take this one. Our company's mission has always been autonomous mobility everywhere. So we certainly have the global ambition since our funding to actually utilize our technology to benefit the local societies worldwide. Currently, our global efforts are focused on the markets with hyper growth potential, so those are the markets with typically strong mobility demand, well-developed infrastructure and a supportive regulatory environment.

When we evaluate a potential market to enter on a high level 3 factors we will consider. One is the addressable market size, which is 10. Second is the openness and the execution of the local government to support and issue permit for the fully driverless commercial operation. Third is how strong is the local partner for their on-the-ground resources and operational capacities. So as you can see, our current global expansion status is that we have already entered 8 countries for our robotaxi. And we also, for example, in Q3, we added Qatar as a new market by collaborating with

In Q3, we have also saw a rapid revenue growth, especially for the robotaxi from our overseas markets, and we certainly expect this momentum to continue. So going forward, we will enter other global markets if we see there's a good growth opportunities. So this is our overseas strategy.

With this, back to the operator.

As there are no further questions, I'd like to turn the call back over to the company for closing remarks.

Thank you, operator. This is George again. If anyone has any more questions, feel free to contact the IR team. We will conclude our call today. Thank you, everyone.

This concludes today's conference call. We thank you all for attending today's presentation. You may now disconnect your line.