Blaize Holdings Inc Stock

Thank you. Good afternoon, everyone, and thank you for joining Blaze's second quarter 2026 conference call. Before management begins the prepared remarks, we would like to remind everyone that earlier today, Blaze Holdings issued a press release announcing its second quarter 2026 results. Earnings materials are available on the Investor Relations section of the Blaze Holdings website. Today's earnings call and press release reflect management's views as of today only and include statements related to the company's 2026 financial guidance, revenue, gross margin, competitive position, anticipated industry trends, market opportunities, products, and financing opportunities, all of which constitute forward-looking statements under the federal securities laws. actual results may differ materially from those contained or implied by these full-looking statements due to risks and uncertainties associated with blaize holdings business For a discussion of the material risk and other important factors that could cause the company's actual results, Please refer to the company's Form 10-K and Amendment Number 1, Form 10-K, for the year ended December 31, 2025, and our Form 10-Q for the period ending June 30, 2026, including the risk factor section therein and today's press release. Any forward-looking statements that management makes on this call are based on assumptions as of today, and other than as may be required by law, we undertake no obligation to update these statements as a result of new information or future events. During this call, management will discuss certain non-GAAP financial measures. These non-GAAP financial measures should be considered as a supplement to and not a substitute for measures prepared in accordance with GAAP.

For reconciliation of non-GAAP financial measures discussed during this call to the most directly comparable GAAP measures, please refer to today's press release. Now I would like to turn the call over to Deniker Monegala, Chief Executive Officer of Blaze Holdings.

Thank you and good afternoon. With me today are Harminder Semih, our Chief Financial Officer, and Stephen Patek, our Chief Revenue Officer. I will start with the outlook and where the business stands. Harminder will take you through the numbers and Stephen will cover a commercial engines. I will then have some closing remarks after the Q&A. As you saw from our earnings release this afternoon, we reduced our revenue outlook for 2026. Our full year revenue is now expected to be between 40 million and 43 million dollars. What that number does not show you is what we have already secured.

We hold a signed agreement covering 2000 servers worth approximately $70 million at current memory prices. Part of that converts into revenue this year. The rest is committed business we carry into 2027. Let me tell you what changed and what did not. First, several engagements have not converted into orders. including some where pilots were completed successfully. Other opportunities are still in progress and expected to close later than we forecast. Third, supply chain cost inflation. Memory pricing has risen materially this year and we expect that to persist.

Harminder will take you through each of them along with the backlog. expect to be holding at year end and what we have changed in how we build our expectations Behind that number, the business is broadening. Our largest customer in China continues to generate meaningful business for us. We have opened Europe with the first purchase order for several thousand units. And activity across the Asia Pacific region has increased. Let me tell you what we are seeing because it explains both the quarter and the book behind it. The market has made up its mind this year. Building frontier AI costs more than it earns, and the gap is widening.

At the same time, efficient open models are making AI cheaper to run. Value is moving from who trains the biggest model to who runs it most efficiently. The economics of inference are now the deciding factor. That is the market our architecture was designed for. And we are making real progress in it. Two market trends are converging, and we are winning in both. First is physical AI. Countries and industrial companies are putting AI into the field on vehicles, on ships, on machines, and inside their own borders. and their own sites, partly for security and control of the data, but mostly because the work demands it.

Speed of response scale places the cloud cannot reach. Next is a new generation of AI data centers built to run AI, not just to train it. Training does not go away. changes shape into constant tuning and specialization, These sites run many models on many kinds of chips. And they're built on purpose, to depend on one vendor both are hybrid and the operators have learned something important renting Creating out GPUs is not a sustainable business. Applications and AI services are. That makes the software that schedules and tunes the work, the layer that matters, and that is exactly where we sit. That brings me to what we are focused on most right now. The revenue that we produce and the margin we make on it.

Let me take each one. First, revenue. We earn revenue in two ways. One is our silicon NSDK designed into OEM's product. shipping inside autonomous systems, robotics, and ruggedized equipment. Once we win the design, we scale with that OEM into markets they already serve. And a proven design opens other platforms and markets for us. The other is our hybrid AI platform, a vertically integrated stack. It runs the industry applications that customers buy, built by us and by our software partners.

Service providers and system integrators deliver it as a managed service. Stephen will take you through where each of them stands. Second, margin. Behind both SIF's AI services are software suite for AI inference. We expect AI services to become an increasingly important contributor of our margin over time. Today, AI services includes capabilities such as facial recognition. Based on requirements from active customer programs, we are developing and integrating additional capabilities, including document processing, quality grading, compliance scoring, video analytics, small language model assistance, and industry-specific services. We are also developing model optimization and orchestration capabilities. intended to route workloads to the appropriate compute resource and optimize models for the underlying hardware.

The goal is to give customers more output per dollar of infrastructure They get more from what they run and we expect to be paid for what gets them there. We intend to price it as software per rack, per megawatt, or per fleet. That brings me to this quarter. Our gross margin was 8 percent, reflecting a mix weighted to third-party hardware. Our branded hardware and AI services is what we expect will shift that mix. Building it out is the work in front of us this year. Finally, onto the next generation. The deployments we are supporting today increasingly require a mixture of models and inference workloads.

Observing that demand pulls us deeper into the stack, both in what we build ourselves and what we integrate from others. Based on requirements that we are seeing across current customer engagements, We are working on our next generation AI inference product designed for production environments. We expect it to complement what we ship today and extend the same architecture to higher-performance workloads. We also intend to incorporate confidential computing capabilities to address requirements from sovereign customers. We view this plan as a staged investment and intend to pace development against customer requirements come commercial progress, and what the business can support. The platform comes first, and the next generation product is intended to extend that platform into larger influence workloads we expect customers to deploy over the coming years. With that, I will hand it over to Harminder to take you through the outlook and the quarter.

Thank you, Dinika, and good afternoon, everyone. Before I get into our second quarter results, I will address why we're revising our full year 2026 revenue guidance, what's driving how we're managing the balance sheet through this transition. Deniker addressed the change from $130 million to a range of between $40 million and $43 million. That is a significant reduction, and I want to walk through exactly why. There are three primary factors behind this change. First, I would like to emphasize that while pilot programs have been successfully completed, several commercial opportunities did not materialize as we expected. We had planned on fulfilling a regular cadence of purchase orders from customers already under contract.

With respect to Starshine specifically, we made the decision not to engage further until Starshine pays its outstanding balance. We have fully reserved the receivable this quarter, engaged local partners to pursue collection, and are reevaluating that commercial relationship. There is meaningful uncertainty as to whether it will progress further. Second, as we progress into the third quarter, customers are deferring follow-on orders based on their broader scaling of overall deployment of AI solutions. Opportunities are proceeding just more slowly than our prior forecast assumed. Cloud and data center customers have taken longer to qualify new technology. certain government programs are on longer procurement timelines than expected. and regional uncertainty has pushed a smart city opportunity in the Middle East into an extended field trial. Finally, memory economics have gotten harder.

DRAM and LPDDR pricing has increased materially this year. as industry capacity has shifted toward high bandwidth memory. Additionally, the requirement for advanced payments from suppliers has increased. We expect these supply conditions to persist. Taken together, we have raised the bar for what we are willing to include in guidance going forward. Stephen will walk through the pipeline in more detail shortly, but I want to be precise about how we built this specific number, because because the methodology matters as much as the figure. As mentioned, our revised guidance is between $40 million and $43 million for the full year we project backlog at December 31 this year of approximately $50 million at current memory prices. It is weighted heavily toward revenue from our largest account and based on binding non-cancellable purchase orders that we can fulfill with inventory commitments already made or planned to order.

Let me explain the difference between bookings and revenue recognition. Several of the opportunities that we're currently pursuing are intended to generate bookings during 2026, but will only partially convert to recognized revenue this year. with the remainder entering backlog for future periods. Backlog for us means a committed contract or purchase order exists that we have not yet fulfilled. This guidance reflects what we currently expect to recognize as revenue in 2026, not the full value of business that we expect to book by year end. Stephen will address where activity continues and revenue upside is in play. With continued supply chain cost inflation that we may not be able to immediately pass through and some higher margin opportunities pushed into 2027, we now expect gross margin of 17% to 19% for the second half of the year. and an adjusted EBITDA loss of $62 million to $65 million for the full year. A reduction in gross margin from lower revenue and the starshine provision are key contributors in the increase in the loss from our prior guidance.

This is offset by reductions in operating expenses and a change in the timing of payments for the next generation chip program. Now, turning to our financial results for the second quarter ended June 30th, 2026. Revenue was $12 million, a significant increase from $2.7 million in the prior quarter. For the first half of 2026, revenue totaled $14.7 million, compared with $3 million in the same period last year. a year-on-year increase of 390%. On neotensor specifically, the amount due from our fourth quarter 2025 sale has been paid in full. And the receivable associated with our second quarter 2026 sale remains within its contractual payment terms. Third and fourth quarter deliveries of our hybrid servers are being planned.

Gross profit was $0.9 million, representing a gross margin of approximately 8 percent, compared with 58 percent in the first quarter. the quarter-on-quarter decline was driven by revenue mix. The second quarter consisted almost entirely of third-party server hardware, which carries constricted gross margins for plays. The first quarter gross margin benefited from a mix of higher margin-play software and hardware plus some third-party hardware. Operating expenses for the second quarter were $31.5 million, up from $23.9 million in the prior quarter, an increase of 32%. This increase was largely driven by a 7.1 million provision for the balance of Starshine receivable, approximately $1 million additional investment into the new chip. and a 2.8 million one-time non-cash charge in a related party settlement. partially offset by a release of a prior provision in professional fees. R&D expense was $10.5 million and included stock-based compensation of $3.7 million. The increase of $0.8 million, or 7.6%, sequentially from the first quarter primarily reflected third-party intellectual property costs associated with the ongoing development of our next-generation chip.

Adjusted EBITDA loss of $20.9 million, including the Starshine provision, was up $7 million compared to a loss of $13.9 million in the prior quarter. Adjusted EBITDA is largely driven by non-cash at-backs, which include stock compensation, changes in the fair value of our financial instruments, and non-cash financing charges, such as the modification of certain warrants this quarter. Moving on to our balance sheet, we ended the quarter with $36.8 million in cash, and in an increase of $3.6 million from the first quarter. We received $9.4 million in customer payments and generated $32.8 million in net proceeds from the equity offering completed during the quarter. Our revised growth outlook reduces the near-term working capital requirements of the business, as the significant second-half bookings and revenue ramp contemplated in our prior guidance would have necessitated significant upfront commitments to purchase memory-intensive inventory at elevated prices well in advance of customer collections. We feel it is advantageous to our shareholders to judiciously manage the supply and demand imbalances of the cost of inventory. This enables us to prudently control economics and timing trade-offs for the long run, rather than maximizing revenue at any cost.

In addition, we continually review our cost structure against revenue priorities and our future roadmap. Accordingly, we are identifying further opportunities to reduce operating expenses in light of the lower guidance. The priority is to optimize cash consumption with a goal of extending our financial flexibility and preserve the core capabilities required to execute our strategy. We are exploring ways in which to leverage our customer contracts to secure non-dilutive debt financing. Additionally, we're seeking advance payments from customers to mitigate working capital demands. No financing decision has been made at this time. With that, I'll turn the call over to Stephen to discuss our pipeline and provide additional context around our output.

Steven. Thank you, Harminder. First, let me start by adding one additional comment on the quarter. Our Q2 revenue was driven by our momentum with Neotenzer, where they continue to adopt more of our solution stack as they see strong demand in their market. Second, and where I want to spend a bit of time, is the commercial picture behind the revised annual outlook Harminder described. You have to start with demand because that is the fundamental starting point. We're seeing real demand across every part of our business and it's growing. What we're working on is our ability to capture it. And we've had several changes in our go-to-market approach in order to execute more effectively.

As it relates to our revised outlook, a couple of things to point out. As opportunities move closer to signed deals, our visibility into end user demand improves, and that allows us to more effectively align supply to demand and scale the supply chain in a measured capital efficient way. already seeing this improvement and I will come back to touch on this shortly The rest is timing. Our customers ramp their deployments at the speed in which their own markets grow, and several ramps later than we had forecasted. Those that we are also getting better at reading these cycles early, which is what improves our close rates moving forward. From a region perspective, the biggest impact we had was the Middle East, which remains uncertain. What we believe is critical moving forward is that we continue to build a broader customer and partner base so that fewer individual deals determine a quarter, and there's evidence that that is working. received our first purchase order out of Europe, where we've already shipped thousands of units, Opportunities are growing across the portfolio throughout Asia Pacific, where we have very strong partner-led engagements, and we're beginning to build a pipeline in the United States. Our pipeline and customer base is broader and more diversified than it's ever been.

Next, I will touch on the two primary revenue engines for Blaze moving forward. First is aero robotics and ruggedized platforms. I mentioned earlier that our focus is on building a scalable and predictable go-to-market engine, and that is exactly what this line of business gives us. These customers have demand for much higher volumes, provide monthly rolling forecasts, and provide deposits to secure inventory. Our differentiation is our architecture. We are being selected in deployments that are constrained on size, weight, power, and heat, where the customer needs sustained real-time performance inside a fixed envelope rather than peak benchmark numbers, and where they're building our SDK directly into their own product. That is a narrow set of requirements, and it's where a purpose-built accelerator, such as Blaze, does better than a general-purpose park. We are being pulled into these opportunities across every market where we have presence.

Once we qualify our solutions, opportunities arise for reoccurring revenue. Our second revenue engine is a hybrid AI platform, which we brought to market two quarters ago, and is what our AI services runs on. It lets Cloud and data center providers deploy faster and open new revenue streams for their own customers through API services. We are actively pursuing several national scale programs, each at a different stage. facility supervision for a national food service group across thousands of sites, production quality grading for a manufacturing company, fuel retail monitoring for a national fuel network with hundreds of stations. And each of these, we are providing the platform and the models and system integrators are handling the deployment. These engagements take time to close and are not forecasted in our current guidance. However, when they do close, we get more than just the contract.

We get a proven solution that can be replicated with customers anywhere in the world. We've also found that there are opportunities to expand even further on our hybrid AI platform and to play a much broader role in new data center build-outs including providing professional services, our AI services suite and advising on the AI architecture of the build. an example there's a national government program building out a data center in the range of 150 megawatts our role is an active discussion and while there are many details to work out the intent on both sides is to move towards finding terms the second sovereign program under discussion with a similar structure and with a different government, but the same partner model. These types of engagements are much longer in nature and are also not included in any of the guidance that we've given today, but they indicate that not only does our platform work with existing customers, but also new NeoCloud build-outs as well. And although these two revenue engines may sound like opposites, they're actually designed to reinforce each other. Physical AI puts our silicon inside real deployments today. The hybrid platform is how we take that into vertical services that enterprises and governments are asking for. And over time, we expect those same customers become the consumption base for the AI data centers of the future that we intend to help enable.

That is where our focus is. With that, operator, we're ready for questions. Thank you. As a reminder, to ask a question, you will need to press star 1-1 on your telephone. To remove yourself from the queue, you may press star 1-1 again. Please limit yourself to one question and one follow-up to allow everyone the opportunity to participate. Please stand by while we compile the Q&A roster. Our first question comes from the line of Evan Cassidy of Rosenblatt Securities.

Yes, thanks for taking my question. And I guess just understand, could you help us understand a little better about the customer delaying their purchases? Is it the, you're saying their end markets aren't moving as quickly? Are they not seeing the benefits of AI, or are the price increases? of slowing them down that maybe there's not a good return on investment. Maybe just a little more detail around why are there so many delays.

Good bye. Gentlemen, your line is muted. Still not hearing a response.

There we are. Yes, sir. Please proceed. Can you hear me? Yes. Is my audio coming through?.

Yes, please proceed. Okay, okay. So let me start and then Stephen can jump in. The nature of some of these projects are tied to camera deployments. So they are about cameras being deployed in the field and then behind that pulling in boxes and then software from Blaze. So that's the delay. I let Stephen jump in on that as well. Stephen Goulding Sure.

Several of our AI service opportunities, we're working with data center and cloud service providers. And as we continue to build our portfolio on our own capacity, they've also got to drive their own demand in the markets in which they serve. So the faster they drive the demand for those type of services, and in this example being computer vision, we will continue to see more orders and more cash capacity that they will need from us. And that's just what some of the delay has been. It has nothing to do with demand, pricing, or anything else. It's just their own services ramping. And so therefore, the capacity that we sell them and they buy from us can ramp at the same levels.

And that's where we've seen some of the delay.

Right. Just to the underlying demand is definitely there, and that's where the whole backlog comes in.

that we've announced based on contracts. And then, you know, opening up Europe, that's great news. And just wonder if you could compare the opportunities that you have in Europe, you know, how do they compare to the US or, Southeast Asia that you've been winning, are they faster time to market or are they, you know, I guess, or is it just more shots on goal having another territory opened? Yes.

No, they're, this is Steven again, look, they're very similar and in line with where our revenue engines are, you know, across all three of those pillars, which we spoke about. What I will say is, is we've got to be very focused on our go-to-market, and we've got to be very intentional on the opportunities that we pursue and when we pursue them. Naturally, you mentioned age-specific, and that started showing itself a couple of quarters ago. And then naturally behind that, we had opportunities in Europe that also started showing themselves at customers that we're talking about now. We've been working with them for a couple of quarters to validate and qualify our solutions. And that's paid off, as we've seen, with our first purchase order. And as they continue to finish their qualification, And now we're going to see really that moving at scale.

So whether it's the U.S. business, Europe, Asia Pacific, they're all really being driven by those pillars we talked about from both the autonomous systems as well as the hybrid AI platform.

Okay, thanks. Thank you. Thank you. Our next question. Comes from the line of Gil Luria of DA Davidson. Your line is open, Gil.

thank you based on the gave something on starshine and new tensor based on the other business that you have and the backlog that you're looking at right now give us an early sense for what the picture looks like in terms of revenue and revenue growth for 2027?.

Sure. The $50 million that we're talking about is essentially a a commitment from NeoTensor, which you remember we announced earlier, contract of up to 50 million. So this 70 million is an amendment to that agreement. And we're expecting that $50 million to go into 2027. I'm expecting 2027 to be probably 2.5 to 3x where we are guiding this year to end at.

Got it. Thank you. Thank you. Our next question. It comes from the line of Craig Ellis of B. Raleigh Securities. Your line is open, Craig.

Yes, thanks for taking the question, guys, and appreciate all the information so far. The first question is really a clarification inside of the updated calendar year 26 guide. So our midpoint's $41.5 million, and that would imply after the revenue revenues in the first half of the year about 26 and a half million in the second half. So we're talking about good half on half growth. The question is this, can you help us with the linearity of revenues from the third quarter to the fourth quarter? And it seems like about two thirds of the second half revenues would be our new server program. Is that right? And what other programs would make up the balance of revenue?.

So hey Craig, so I would see, I'm expecting Q3 to be similar to Q2, and Q4 will be a little bit heavier. Any upside that some of the things that Steven talked about will more likely come and hit us in Q4. But you're right that the $70 million or part of the $70 million is going to be the majority of what happens in Q3 and Q4. And again, going back to the discipline that we're applying to guidance, is those are contracts in hand. There are others that are.

would expect it to follow shortly or follow in time. Maybe I can answer all. Yes, thanks Stephen. I can answer the second part of your question as well. Look, we have an incredibly high confidence level on the way that we just guided, right? We have 90% of those orders in house. The additional 10% is not single-threaded on any given deal. And so a very high level of confidence as it relates to how we get to the number that we just stated.

Yes, and I think a lot of us will find that conservatism refreshing, Stephen. So thanks for that. And it seems like there's conservatism in the way you framed up your end backlog, just given program size. My follow-up question was really related to longer-term items, and I'll phrase it this way. As the team looked at how it was going to frame this year's expectations, I think the press release indicated there were a number of opportunities that were excluded from this year's new revenue guide, but it seemed like those opportunities were still advancing. And is that the case? And as we think about the four national scale hybrid AI platforms, can you just help us understand how, Could those be things that contributed to revenue next year? What would lead to that? And if next year would be the first year of revenue, what's the timeframe we're talking about across these four different opportunities? Thanks, guys.

Yes, sure. Thank you for the question. On those, to your point, those are not included in the guide. They are moving forward vigorously through the sales stage, through customer acceptance and qualification. Right now, I would peg at least... let's call it, I don't want to ever call all four of those, but let's just say those subset of opportunities would look to be Q1 of next year. And our goal is to be able to bring that into this year and provide above and beyond the guidance. But right now, I would peg that for Q1 next year.

Yes, and just to calibrate, I think the trends that we're witnessing are important. to do with aerial autonomous robotics and hybrid AI and that's what is driving the demand so we're feeling good about how our AI services and hybrid AI is actually helping our customers and that's driving all the demand.

Thanks, guys. Thank you. Once again, to ask a question, please press star 1-1 on your telephone. Our next question comes from the line of Richard Shannon of Craig Harlem Capital Group. Your line is open, Richard.

Great. Thanks, guys, for letting me ask a couple of questions as well. I guess the first one is on me talking about about an elongated sales cycle here. I guess one thing I want to confirm is that these opportunities are still in the pipeline and not lost. Can you clarify the degree to which that is the case?.

Absolutely. All these opportunities are still in the pipeline and are not lost. And honestly, our pipeline continues to grow. But the reality is for us right now is we are focused on the opportunities that are in front of us that we can execute and close as quickly as we can. And the ones that we're speaking about now are absolutely still in the pipeline, just simply have moved from a timing perspective.

Okay. And to follow up on this general topic, you cited the memory costs, which are well known in this market here, I mean, do you need to see these come back down in order for these to re-accelerate or reignite or whatever? Or do we just need to stay flat here? Or just what needs to happen here?.

with memory costs in order to make good on this pipeline. Hey, Richard. So we're doing a couple of things. Number one, we already started conversations with the memory suppliers to see what it would take to go do some forward buying, what kind of commitments they need, and that conversation is ongoing. And really, the second thing is that we've invested probably eight and nine, eight and nine million dollars already in forward buying certain components, chips. and some of the boxes and so on that are required. And most of those are going to turn into revenue over the next six to nine months. So we're trying to manage the cost that way. The final point I'll make is that with we're reserving the right to change pricing as memory pricing changes, you know, change the pricing to the customer.

So that we maintain at least, when we're doing third-party software, third-party hardware, I beg your pardon,.

at least we're maintaining a reasonable margin. Okay. That makes sense. And one last question for me, and I'll jump out of line here. You mentioned these four national programs. Obviously, I've identified the countries and wouldn't expect you to, but some countries are very small and some are large here. Could you give us some sense of how big these programs can be, certainly in your pressure at least in your prepared remarks, you talked about some deals with thousands of servers. Is that the kind of scale we're talking about or could it be more? Just give us a sense of what these could be. Yes, I think that's fair. Look, because of our, I would say our focus in the last couple of quarters,.

Several of these deals that we mentioned are in Southeast Asia. Right? And so, you know, we're going to have to wait and see. really primarily we're seeing the acceleration of the sales process and the sales cycle because we have a very strong partner-led model there. And one we've mentioned in the past, one being, you know, one being Nokia, and that allowed to accelerate that as well. And these are very large opportunities across several countries in Southeast Asia. So I think that that scope.

that you gave is reasonable. Okay perfect that is all from you guys thank you.

Thank you. This concludes the question and answer portion of today's call. At this time, I would like to turn the call back to Deniker for closing remarks.

Thank you all for joining us. Let me close with the three things we are driving from here. conversion turning what we are pursuing into committed orders so you it lands as revenue this year and as backlog into 2027. Second, margin. It improves when full solution goes in. Our silicon, our software, with AI services on top. We expect the first revenue from AI services this year. Third, cost. We have reset our revenue outlook this quarter, and aligning our spending to match it. I said earlier that the economics of inference are now the deciding factor. and that this is the market our architecture was designed for.

That has not changed and neither has our conviction in it. Every AI deployment being built right now will run inference for years after it goes live. Stephen showed you where the demand is coming from. Enterprises, governments, sovereign programs. We are built for it on all three fronts. The demand in front of us, the software we are adding, and the next generation product that comes next. And on current expectations, we are entering 2027 carrying approximately $50 million of that committed business. Thank you for your time and your support.

This will conclude today's conference call. Thank you, everyone, for joining. You may now disconnect your lines.

This live transcript is auto-generated without human intervention or review.

[Call has ended.]

Good day, and thank you for standing by. Welcome to the Blaize First Quarter 2026 Earnings Conference Call. [Operator Instructions]

Please be advised that today's conference is being recorded. I would now like to hand the conference over to your speaker today, Lana Adair, Investor Relations. Please go ahead.

Before we begin the prepared remarks, we would like to remind you that earlier today, Blaize Holdings, Inc. issued a press release announcing its first quarter 2026 results. Earnings materials are available on the Investor Relations section of Blaize Holdings, Inc.'s website. Today's earnings call and press release reflect management's views as of today only and include statements related to our 2026 financial guidance, revenue, gross margin, competitive position, anticipated industry trends, market opportunities, products and financing opportunities, all of which constitute forward-looking statements under the federal securities laws.

Actual results may differ materially from those contained in or implied by these forward-looking statements due to risks and uncertainties associated with our business. For a discussion of material risks and other important factors that could impact our actual results, please refer to the company's Form 10-K and Amendment #1 to Form 10-K for the year ended December 31, 2025, and our Form 10-Q for the period ending March 31, 2026, including the Risk Factors section therein and today's press release, both of which can be found on our Investor Relations website.

Any forward-looking statements that we make on this call are based on assumptions as of today, and other than as may be required by law, we undertake no obligation to update these statements as a result of new information or future events. Information discussed on this call concerning Blaize Holdings, Inc. industry, competitive position and the markets in which it operates is based on information from independent industry and research organizations, other third-party sources and management's estimates.

These estimates are derived from publicly available information released by independent industry analysts and other third-party sources as well as data from Blaize Holdings, Inc.'s internal research. These estimates are based on reasonable assumptions and computations made upon reviewing such data and Blaize Holdings, Inc.'s experience in and knowledge of such industry and markets.

By definition, assumptions are subject to uncertainty and risks, which could cause results to differ materially from those expressed in the estimates. During this call, we will discuss certain non-GAAP financial measures. These non-GAAP financial measures should be considered as a supplement to and not a substitute for measures prepared in accordance with GAAP. For a reconciliation of non-GAAP financial measures discussed during this call to the most directly comparable GAAP measures, please refer to today's press release.

Now I'd like to turn the call over to Dinakar Munagala, CEO of Blaize Holdings, Inc.

Thank you, Lana, and good afternoon, everyone. We came off a breakout growth year in 2025, and we expect 2026 to continue that trend. Q1 strengthened our commercial foundation through several new contracts and partnerships. First, we expanded our NeoTensr contract, bringing the total potential value to $70 million.

We signed a strategic partnership agreement with Winmate, a publicly traded leader in ruggedized computing with the intent to close approximately $15 million in business in the first year. We deepened our joint engagement with Nokia across Asia Pacific. Together, we stood up a joint AI innovation lab advancing hybrid AI rack scale development.

The engagement also includes a strategic partnership with Datacomm, one of Southeast Asia's leading cloud service providers. Finally, we announced Blaize AI Services and will bring our first application service to market. Q1 revenue came in at approximately $2.7 million. This reflects a global memory shortage that limited server availability from one of our trusted suppliers and delayed orders.

Customer demand remained intact throughout the quarter. We expect to secure the inventory needed to deliver over $11 million to a single customer in the second quarter of this year and we are reaffirming our full year 2026 revenue guidance of $130 million. At GITEX AI 2026, in April, one of the largest AI showcases in Asia, we announced Blaize AI Services, which we expect to turn AI infrastructure into production-ready APIs that cloud service providers, data center operators and system integrators can deploy, monetize and resell.

Today, we are going to announce the next step in execution, the upcoming launch of our face recognition AI service, the first in a series of application-level services running on the Blaize Hybrid AI platform. Why this matters? AI services will complement our hardware sales with recurring application layer revenue per query. It's higher margin, it's stickier, and it scales with our partners' growth, not just with their CapEx cycle.

Face recognition is the first proof point, additional high-demand services, including intelligent document processing will follow. We have signed a contract with NeoTensr that is expected to generate up to $50 million in revenue in the first year. This builds on more than $20 million in revenue that we recognized in Q4 of 2025, bringing the total potential value to approximately $70 million. The development uses a co-branded AI server built on Blaize Quad card. Each server handles 200-plus simultaneous camera streams with advanced AI analytics while running LLM and VLM inference on the same infrastructure. This is what our hybrid AI architecture was built for, real-time perception at the sensor layer, advanced reasoning on the same rack, no round trip to a distant cloud.

The rollout is expected to span multiple cities across Asia Pacific in multiple phases. Each phase is expected to drive higher-margin revenue as the AI services layer takes hold. Earlier this month, we entered into a strategic agreement with Winmate. Together, we will integrate Blaize AI into ruggedized systems, drones, handhelds, vehicle-mounted units and embedded devices for mission-critical operations, border security, maritime, essential infrastructure and field health care.

Beyond the contracts I just described, we are advancing a series of rack-scale hybrid AI engagements anchored by our joint partnership with Nokia. This work reaches cloud service providers and infrastructure partners. These opportunities are multisite, multiphase with hundreds to thousands of edge nodes per program. They span smart city, sovereign data center and large-scale ruggedized field use cases. The architecture is hybrid GSP plus GPU at rack-scale, orchestrated by Blaize AI Services stack.

The pattern is consistent. Customers want sovereign control of their data. They want efficiency. They want application-level AI services they can resell. Hybrid AI delivers all 3. Stepping back, the AI infrastructure conversation is shifting fast. A year ago, the industry was focused on one thing, massive centralized GPU clusters for training.

Today, the conversation moved decisively towards sovereign language model, inference at the edge, in-country at unit economics that actually work at scale. That shift is what Blaize was built for. Three pillars: number one, sovereign AI infrastructure. Governments and large enterprises across Asia, Middle East and Europe demand compute that stays within their borders under their control. Hybrid rack-scale enables this without hyperscaler economics.

Number two, smaller LLM-based AI services. Most enterprise AI workloads do not need a frontier model. They need a tightly tuned domain-specific model on infrastructure they can afford. Our hybrid architecture runs vision and language workloads on the same rack, opening the service revenue our partners can monetize for query.

Number three, programmable energy-efficient compute. This is where the Blaize GSP advantage compounds. Performance per watt, deterministic latency, a software stack that serves vision, LLM and VLM workloads on the same hardware. Hybrid rack-scale is the unit of deployment for the next phase of AI. We are building toward it, and our partners are buying in.

On May 6, we closed a $35 million equity offering, supported by a group of large institutional investors. This capital strengthens our balance sheet. The proceeds will support our commercial deal commitments, continued AI services development, rack-scale hybrid platform advancement and next-generation platform development.

Blaize is a company executing against one of the most significant opportunities in AI history. Rack-scale hybrid AI, sovereign infrastructure, the strategic path for recurring AI services revenue and partnerships that put Blaize at the center of the AI inference build-out.

Contracts are expanding, partnerships are deepening across an increasingly diverse base of AI use cases. And finally, engagements are advancing in the field.

So with that, I'll turn it over to our CFO, Harminder Sehmi.

Thank you, Dinakar, and good afternoon, everyone. I'm pleased to share our first quarter 2026 results today. First quarter revenue was $2.7 million, up 170% (sic) [172%] year-on-year and in line with the pre-release issued on April 14. As we flagged at that time, this was impacted by an industry-wide shortage of high-bandwidth memory or HBM, the specialized memory chip that is necessary for AI servers primarily used for training or running large language models. That shortage delayed an order to one customer, NeoTensr, that we now expect to fulfill in the second quarter at a value of more than $11 million. This is about a timing issue.

Customer demand remains strong and over 70% of the revenue billed to NeoTensr in Q4 of last year has been collected to date. Beyond NeoTensr, revenue in the quarter included delivery of software licenses and servers to our primarily U.S.-based customer drawn from inventory on hand. As noted on earlier calls, our road map for hybrid servers mitigates against these challenges.

Our partner-branded servers powered by Blaize cards deliver competitive AI inference performance without requiring HBM. We expect those servers to begin shipping in the second half of this year, and we have already placed forward orders for Blaize chips and cards. We're exploring ways in which to strategically procure certain memory cards now to meet our projected demand into 2027. We believe this approach helps derisk our projected revenue growth as the data center opportunities begin to crystallize.

In parallel, we are developing a comprehensive rack-scale service solution to address data center inference workloads. We will continue to deliver enhancements to the application features on our AI services platform throughout the year. Given the timing of large orders and the early stage of data center expansion, we expect revenue to be back half weighted this year with visibility increasing as opportunities convert.

Gross margin was 58% this quarter, up from 11% in the fourth quarter of 2025. Two factors drove the expansion. First, the mix shifted towards our higher-margin software and Blaize-powered hardware. Second, the HBM-intensive NeoTensr order shifted into the second quarter. As previously indicated, blended gross margins are expected to be compressed by the higher portion of third-party hardware in our revenue mix in the next 2 quarters. As we begin the transition to deliver more inference servers and recognize recurring software revenues, blended gross margins in the fourth quarter of 2026 should exceed 30%.

We anticipate further expansion in gross margin in 2027 as our partnership with Nokia opens additional data center opportunities globally. Net loss for the first quarter was $22.7 million compared to the net loss of $147.8 million for the same period a year ago. Q1 of 2025 included significant noncash items and onetime merger transaction accounting adjustments. Consistent with previous calls, I'd like to spend a few moments breaking these numbers down to provide clarity about the underlying results, including singling out quarter-on-quarter trends where helpful.

Total operating expense, including stock-based compensation of $8.9 million was $25 million in this quarter. This was a decrease of $14.7 million year-over-year. Q1 of 2025 included $11 million of stock-based compensation and $12 million in transaction expenses related to the business combination. The cleaner story is in our operating discipline. Research and development costs of $5.8 million in the first quarter, excluding stock-based compensation, were marginally lower than the prior quarter cost of $5.9 million. Selling, general and administrative expenses, again, excluding stock-based compensation were $10 million in the first quarter of 2026, up $1.6 million sequentially.

Adjusted EBITDA loss for the first quarter this year was $13.9 million, $1.5 million better than the loss in the first quarter of 2025 and $1.9 million higher than the fourth quarter of last year. We ended the first quarter with a cash balance of $33.3 million on March 31, 2026. On May 6, we announced our $35 million equity raise that extends our runway to the middle of 2027 and adds a new base of shareholders. This round drew strong participation from high-quality institutional investors with deep expertise in data center infrastructure investments. This growth capital will enable us to deliver against demand to accelerate customer rollouts, lean into the data center opportunity and invest in our product road map.

We maintain close relationships with our key vendors and continually seek to secure favorable payment terms, which is particularly important during this period of supply chain constraints. As our data center opportunities gain momentum, we also intend to explore appropriate project financing partnerships to support deployments at scale.

Finally, our revenue outlook for full year 2026 remains unchanged with the second half meaningfully stronger than the first. Our adjusted EBITDA loss guidance also remains unchanged at between $45 million and $50 million for the year.

In closing, our recent equity raise was well subscribed and drew strong participation from marquee investors with exposure to the data center infrastructure ecosystem. Our AI services platform and rack-scale hybrid AI developments are resonating strongly as the market shifts towards inference and real business outcomes from AI. And finally, we have great and growing partnerships in place to support revenue growth.

With that, I'll turn it back over to the operator.

[Operator Instructions]

Our first question comes from Kevin Cassidy with Rosenblatt Securities.

Congratulations on maintaining the $130 million for the year. When we look at that $130 million, how would you expect it to be spread across geographically for you?

So it's -- the NeoTensr contract, of course, is expected to contribute a significant portion of the $130 million. There are other opportunities in Asia Pac through the Nokia partnership. Datacomm is the one that we announced. That should start to feature towards the end of Q4. And we have other edge opportunities in Europe that are also expected to be part of that $130 million number. So it's spread around Europe, Asia Pac. Dinakar, I don't know if you want to add.

Yes, the pipeline is quite strong in North America as well. And we are beginning to discuss some commercialization via orders that in the U.S. as well as in Africa as well. As they materialize, we'll, of course, be sure to announce them.

Okay. Maybe could you also talk about the effect that maybe the war in Iran might have on some of your opportunities there for security?

We have actually received significant inbounds for our drone detection system use case that we've demonstrated. This is all about perimeter security kind of use cases. And yes, there's an increased momentum in terms of opportunities coming our way. Of course, as these materialize into POs and revenue, we will keep announcing them.

Okay. And just one more question on the supply chain. So I think in your pre-announcement, you had said that you're expecting product to be shipped in the April quarter first. Did that happen? And is it only the memory that's the long lead times? Or are you having trouble with other products also?

So these are the HBM-intensive memory sort of servers and NeoTensr is one of the early customers for the business we do there. So it's actually obtaining the server itself. One of the reasons that we explained in Q1, we could have secured supply, but we would have actually had to pay premiums that we weren't prepared to at the time. As we move forward into Q3, Q4 and our hybrid servers become available and particularly the one we're really excited about is the one with NeoTensr, the white labeled one, which has our Quad PCIe card in it, then some of those supply chain problems should diminish somewhat. But I think the macro sort of environment is still something that we all need to keep an eye on.

Our next question comes from Richard Shannon with Craig-Hallum Capital Group.

I'll ask a very quick tactical question here regarding the outlook here for the second quarter. Harminder, I think you mentioned you're targeting $11 million for one particular customer. Is that the estimate or starting point you would like us to think about? Or could it be somewhat or meaningfully higher than that?

It will be somewhat higher, but again, it depends on just getting -- maybe in our one-to-ones, Richard, we can talk a little bit more openly about that. But for now, we have good visibility on getting the NeoTensr delivered in addition to 1 or 2 others that we have in mind.

Okay. Perfect. Second question, I guess, for probably both of you, but I want to ask about the Blaize AI services. You're talking about the first application being face recognition rolling out here. I'd love to get kind of a few different questions about this. First of all, over what time period do you expect this to be rolled out and ultimately bring first revenue recognition for you? Are there any particular end markets where you expect to be first adopted?

And then the last part is, how do we think about kind of the revenue contribution over the life cycle of your equipment relative to that equipment sale? Is there a percentage we should be thinking about? Just any way to kind of provide a mental model for that, that would be great.

Sure. I can take the first part and then Harminder can jump in. AI services, certainly, it is exciting to our cloud service provider partners as well as data centers because it allows them to monetize their infrastructure that they've invested in and that's driving all the momentum. So initial application, of course, we have video-based applications that we are working on, which we're actually working with anchor partners as well as the facial recognition. And the initial target is around use cases around smart kitchens, around immigration, those class of use cases where face rec is pretty widely used.

Initial anchor customers are in the Asia region. Also things like citizen safety, elderly care, et cetera, there's some software that we've developed that is actually being well received. In addition to this, document processing is something that we will be next launching, and that's announcing -- it's already under development, and we will be releasing it to early access cloud service providers once it's complete. And this is actually quite helpful because from an economic standpoint, the cloud infrastructure that they invest will be monetized, the recovery, return on investment is much faster because they'll be able to monetize it through these services. I'll let Harminder...

Yes. Your other question was the time period. We expect from Q4 onwards to start to deliver some of the CapEx. So if you stand back, the AI services comprises of Blaize-powered servers, hybrid servers. So there's a certain amount of CapEx involved, which we recognize straight away. And then there is a recurring revenue element associated with monetizing the APIs. And that, of course, there will be some sort of a contract in place, but the revenue recognition will be monthly as usage takes place.

But Q4 is when we start to see some of that featuring in our revenue mix. I actually expect to see AI services as a whole becoming a significant feature of 2027 revenue mix and more of it being some of this recurring revenue because we have the opportunity to basically trade off some of the upfront margin that we would make on the CapEx sale in place of higher margin of ongoing software revenues.

And just to add that although we spoke about these 2 or 3 areas, there's quite a strong and compelling road map behind this that we are announcing and showing our early access partners, and it's resonating well with them. This is actually helping us significantly in terms of translating the conversations into actionable, how they place orders and become long-term partners with us.

Okay. Great. My last question, I'll jump on the line here is just a follow-up on Nokia. Obviously, a great partner to have here with worldwide reach. It seems like your first big partnership with Datacomm seems to be the kind of the champion of Indonesia here. How do we expect to see or how should we look for success in other places in Southeast Asia through Nokia? How are those developing? What should we expect to see from that during 2026?

So we started off about 6, 7 months ago with Nokia. And the initial action was to develop a joint pod, rack-scale offering that comprise both Nokia and Blaize hardware as well as AI services software. And we've demonstrated this at GITEX Asia. That was well received. And there's a pretty strong pipeline of customers behind that cloud service providers, infrastructure players, system integrators that we've been working with.

And the first conversion is Datacomm, and there are others behind it. So as these contracts start materializing, we'll start announcing them. I don't know if you want to add any more.

And just the only thing I'd add is the other thing we're really excited about is the rack-scale hybrid server work that's happening right now because as you recall, a couple of quarters ago, we introduced the whole concept of AI services platform. And what we're now starting to see is that concept resonating really well with cloud service providers. Something that Dinakar has mentioned been mentioning for a while is the faster we can help these Tier 2 players to reduce their ROI through a combination of Blaize hardware and other partner solutions, then the faster we will see the adoption of real-world outcomes from AI being utilized by customers.

Our next question comes from Craig Ellis with B. Riley Securities.

I wanted to pick up where you left off talking about AI services and just clarify, inside of the expectation for $130 million in revenues this year, what have you incorporated for AI services?

If I take a combination of the hardware and some of the software, probably about 15% to 20%.

Got it. And then another lens into the $130 million, we've got more HBM-dependent configurations and HBM free configurations. If we look at the $130 million on the systems side of the business away from services, how does the expectation split between what's dependent upon HBM and what would be HBM free?

So if you'd asked me this question maybe 3, 4 months ago, I would have said a large portion of the NeoTensr early contract that we've got would be more HBM intensive. What's actually forcing a faster adoption of -- towards our hybrid solutions is the fact that these servers are now becoming uneconomic for some of the smaller players.

So out of the $130 million, maybe 20% or so would be the HBM sort of intensive stuff. But I'd see -- I'd expect to see a migration. Our servers start to come on stream in the second half of this year and at scale. And the faster we can get that done, the faster we can make sure that our own supply chain is unencumbered, then that transition will happen that much faster.

Yes. Just to add that quite a bit of momentum around our -- the fact that we were able to demonstrate a real end business case return on investment using DDR technology. I think that's actually resonating well with customers. So majority of the $130 million is based on DDR, LPDDR kind of memories.

[Operator Instructions]

Our next question comes from Scott Searle with ROTH.

Maybe just a couple of follow-ups on Blaize AI services. I wanted to clarify in terms of the ramping recurring model, is that revenue share? Or is that going to be purely capacity driven? And then also to follow up on a couple of the earlier questions, I think you said, Harminder, about 15% to 20% would be tied to that either in CapEx or otherwise in calendar '26. Is all of that to occur in the fourth quarter? And then what's the early thought process then in 2027? You said it would be significant. Just wondering if you could frame it for us. And then I had a couple of follow-ups.

Okay, sure. So the recurring revenue is partially revenue share, but we also have developed a very rich library of AI models, which we're already monetizing with some of the sales that we've made so far. So it's going to be a combination of the particular deals that we strike with the partners that we've got in cloud service providers, the cloud service provider partners that we get through rev share, through licensing of some of those libraries that we've developed. And then in -- yes, the 15% to 20%, I expect largely in Q4. It's just a question of when those servers of ours become available at scale.

Got you. And in the past, you guys have talked about a total qualified opportunity pipeline. I'm wondering if you could give us some indication in the ballpark of where that might be. And Dinakar, there were a couple of comments that I found interesting. I think you referenced the United States. some opportunities. I wonder if you could talk about the application in the end market. And I think specifically, you said within Europe, more edge AI applications, and you've mentioned drones a couple of times. I'm wondering how small and scalable do the solutions go? Are you going out to the drones themselves in ruggedized applications? Or is it an other infrastructure that ends up being drone detection?

So the combination of both. If you see, we do have this small [indiscernible] factor, form factor that can go into a drone. So we do have a pipeline based on that. We also have the connectivity layer to a command and control center and where our servers reside. And there, you could do actions like drone detection, any kind of early drone security warning, which is actually quite an interesting use case amidst what's happening globally.

So I'd say it's a combination of both. To the earlier question about U.S., the range of opportunities are from energy-efficient data center. That's one of the initial and driving thing because our servers are inherently lower in power. And therefore, the OpEx for the end cloud service provider and the cloud and the data center operator is much lower. At the same time, using our AI services, they can monetize the infrastructure. So that's driving the U.S. business. And I don't know if you want to.

Yes. So you asked about the pipeline. So look, pipeline is constantly evolving. For us, it's a sizable number. We -- and we're prioritizing the near-term opportunities and particularly those that leverage our -- the hybrid AI services advantages. What we are transitioning to focus on, and I'll talk -- start to talk a lot more about this on the next call is about our contracts and POs, about our bookings, about backlog and revenue. I think these are much more meaningful metrics that enable folks like yourselves and investors to get a sense of where the revenue growth is -- how the revenue growth is developing.

Very helpful. And lastly, if I could, I'll throw out one more. Just the competitive landscape, it's rapidly shifting. It's rapidly evolving out there in terms of edge AI and data center hybridization. I'm wondering who you're seeing on the short list and who you're really competing against besides the large obvious guys?

So we're actually complementing quite a bit of GPU-based designs. So people look at us as a healthy way to reduce both CapEx and OpEx. So that's one. The second piece is often the discussions are around, hey, I have these enterprises, right? They really care about the use case that they're trying to solve within a certain CapEx and OpEx budget.

So really, those are the frameworks that we get in. And then having a programmable solution and the right software and AI services helps us piece together along with our system integrator partners, solutions for the business. So it's less to do with who's a head-on competitor, but more about how we deliver to a certain business value, and that's what is resonating and leading to wins.

I'm showing no further questions at this time. I would now like to turn it back to Dinakar Munagala for closing remarks.

Thank you for your questions. And let me share a few thoughts before we close. The inference market is now and Blaize is positioned at the center of it. 2025 was a breakout 20x growth year and the contracts and partners that we discussed today are extending the trajectory into 2026. NeoTensr drives our Asia Pacific edge data center expansion with $70 million in total value. Nokia anchors our rack-scale engagements and AI services engagements across cloud service providers and infrastructure partners globally with Datacomm extending our reach across Southeast Asia.

Winmate brings Blaize into ruggedized platforms for mission-critical operations and embedded edge infrastructure. Customers are validating our hybrid AI rack-scale platform and our AI services layer as the right way to address the inference economy. The momentum is real, and we're excited and we expect to continue this trajectory in the coming quarters. Thank you for your time and continued support.

This concludes today's conference call. Thank you for participating. You may now disconnect.

Good day, and thank you for standing by. Welcome to the Blaize Fourth Quarter 2025 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. [Operator Instructions] I would now like to hand the conference over to your speaker today, [ Lana Adair ], Investor Relations.

Good afternoon, everyone. Thank you for joining Blaize Holdings, Inc.'s Fourth Quarter 2025 Earnings Call. Before we begin the prepared remarks, we would like to remind you that earlier today, Blaize Holdings, Inc. issued a press release announcing its fourth quarter and full year 2025 results. Earnings materials are available on the Investor Relations section of Blaize Holdings, Inc.'s website.

Today's earnings call and press release reflects management's view as of today only and include statements related to our 2026 financial guidance, revenue, gross margin, competitive position, anticipated industry trends, market opportunities, products and financing opportunities, all of which constitute forward-looking statements under the federal securities laws. Actual results may differ materially from those contained in or implied by these forward-looking statements due to risks and uncertainties associated with our business. For a discussion of material risks and other important factors that could impact our actual results, please refer to the company's Form 10-K for the year ended December 31, 2025, including the Risk Factors section therein and today's press release, both of which can be found on our Investor Relations website.

Any forward-looking statements that we make on the call are based on assumptions as of today, and other than as may be required by law, we undertake no obligation to update these statements as a result of new information or future events. Information discussed on this call concerning Blaize Holdings, Inc.'s industry, competitive position and the markets in which it operates is based on information from independent industry and research organizations, other third-party sources and management's estimates. These estimates are derived from publicly available information released by independent industry analysts and other third-party sources as well as data from Blaize Holdings, Inc.'s internal research.

These estimates are based on reasonable assumptions and computations made upon reviewing such data and Blaize Holdings, Inc.'s experience and knowledge of such industry and markets. By definition, assumptions are subject to uncertainty and risk, which could cause results to differ materially from those expressed in the estimates. During this call, we will discuss certain non-GAAP financial measures. These non-GAAP financial measures should be considered as a supplement to and not a substitute for measures prepared in accordance with GAAP. For a reconciliation of non-GAAP financial measures discussed during this call to the most directly comparable GAAP measure, please refer to today's press release.

Now I'd like to introduce Dinakar Munagala, the CEO of Blaize Holdings, Inc.

Good afternoon. Over the course of 2025, we grew our revenue from approximately $1 million in the first quarter to $23.8 million in the fourth quarter. We exceeded the upper end of our revenue guidance, representing approximately 20x growth over the year. This reflects strong momentum across inference infrastructure, sovereign AI and public safety applications. Customers today evaluate AI infrastructure on 3 things: cost per inference, power efficiency and revenue per rack. At the same time, many enterprise inference workloads do not require the largest models. We are seeing increasing adoption towards smaller task-specific models that deliver strong results with far greater efficiency and faster time into real business outcomes. That is where Blaize is focused.

Over the past several months, we strengthened our execution. We brought on a Chief Revenue Officer, Stephen Patak, to scale our commercial efforts globally. In January, we signed an MOU with Nokia's Asia Pacific division, and we are now advancing that collaboration through an innovation hub in Singapore to build and validate our combined AI platform. We will launch this at GITEX Asia in Singapore, where we'll present to enterprises, governments, cloud providers and data center operators across Asia Pacific. We are already seeing early traction taking shape across the region, spanning cloud infrastructure, sovereign AI and real-world applications.

Many of these opportunities follow multiphase models, where our systems expand as workloads grow. One of the most concrete examples is in India, where we signed an MOU with the Government of Telangana supporting its AI cloud innovation hub. This foundational platform spans mining safety, smart cities and agriculture, where we jointly enable real-time intelligence of worker safety, equipment operations and environmental conditions. In China, we are also expanding our footprint with regional solution providers focused on AI data center build-out, driving assisted living and smart community solutions in patient safety and remote monitoring with enterprise engagements underway.

In Korea, we are working with solution partners like GSIL specializing in factory safety and industrial monitoring. Across Southeast Asia and Australia, we are working with Nokia and vertical systems integrators to explore AI use cases in urban safety, retail analytics, maritime infrastructure and airport security. In the U.S., Europe and Latin America, we're expanding engagements across enterprise and data center environments focused on AI infrastructure, public safety, industrial robotics and autonomous operations. And for the third consecutive year, we will showcase our solutions at ISC West, the largest converged security trade show. The Middle East and North Africa continues to be a strong growth market.

Governments and enterprises are investing in security, visibility and sovereign infrastructure. In Saudi Arabia, we support energy and urban city use cases. In the UAE, we support civil defense, aerial monitoring and drone detection. In North Africa, we support large-scale industrial ecosystems. Blaize enables real-time detection and monitoring, supporting infrastructure security across energy, industrial and transportation environments. Our capabilities extend into robotics and autonomous systems. These systems require low latency and efficient inference where hybrid architectures become essential. This shifts AI from centralized data centers to distributed infrastructure. AI infrastructure is no longer limited to hyperscalers.

It is now distributed across regional cloud providers, data center operators and sovereign programs. AI environments today remain highly fragmented. Thousands of vendors deliver narrow AI capabilities focused on vision, documents, identity or automation. Organizations are left integrating multiple systems before they can deliver real outcomes. The opportunity is to move from fragmented tools to integrated services. These capabilities are consolidating into platforms, and that transition is happening now. What ties all of this together is the underlying economics. At scale, this is about cost, efficiency and utilization. In our analysis, GPU-only infrastructure can scale revenue but remains constrained by high and recurring compute cost.

By contrast, the Blaize model is designed to be cash flow efficient from the start, driven by lower silicon cost and power efficiency. A hybrid configuration combining GPUs and Blaize inference acceleration can deliver roughly a 50% lower infrastructure cost with approximately 60% lower power consumption or more than 2x improvement in efficiency. To support this model, we are progressing towards the initial release of the Blaize AI services platform in the second quarter. This is not just about cost. It brings fragmented AI capabilities into a unified services layer and enables customers to move faster from infrastructure to real-world outcomes. The platform combines inference silicon, intelligent software, API-based AI services.

For AI providers, instead of relying on GPU rental, Blaize enables operators to monetize AI outcomes. Revenue comes from inference transactions, AI events and application services. As services scale, revenue grows faster than cost, driving operating leverage and margin expansion. In our analysis, traditional infrastructure models remain cost constrained over time. The Blaize AI services model enables more efficient scaling of revenue with improving economics as services grow. This is the difference between scaling compute and scaling a business. AI infrastructure investment continues to expand globally. This phase of the industry is no longer defined by larger models. It is defined by monetizing inference at scale. Platforms that combine efficient architecture with AI services are defining how AI operates today. Blaize is built for that model. Our focus remains on expanding commercial activities, scaling AI services and converting pipeline into revenue.

Thank you. I will now hand this off to our CFO, Harminder Sehmi.

Thank you, Dinakar, and good afternoon, everyone. I'm pleased to share our fourth quarter and full year 2025 results today. I'd like to begin with a few highlights. This is the fourth consecutive quarter where we exceeded our revenue guidance range since we became a public company in January 2025. Revenue of $38.6 million for the full year 2025 outperformed the upper end of our guidance by $600,000. Revenue for the fourth quarter doubled to $23.8 million from the prior quarter. And adjusted EBITDA loss was $50.5 million, an improvement of $4.5 million from the lower end of our guidance range for the year. This includes a $1 million benefit from higher gross margin and $3.5 million in lower OpEx and deferred technology costs.

Focusing on revenue and gross margin. In the fourth quarter, we delivered revenue of $23.8 million, exceeding the upper end of our guidance by $700,000. This performance was driven by customer deployments of servers in the Asia Pacific region, supporting AI solutions into the smart health space. We are seeing continued demand as customers expand into AI data center infrastructure build-outs, which we expect to contribute to future revenue growth.

Turning to the full year. 2025 marked an important milestone as our first full year of operations as a public company. We're pleased with the progress we achieved. Revenue of $38.6 million was up significantly from $1.6 million in the prior year. This reflects our success in laying the foundations to meet the rising demand for AI solutions across high-growth markets. Our growing partnerships with systems integrators and software providers is key to accelerating and streamlining the adoption of AI solutions powered by Blaize hardware and software.

Let me now address the gross margin trends. Gross margin for the fourth quarter was 11%, and it was 16% for the full year. In prior updates, I have indicated that this approach has been important to our strategic plans as we've been able to more rapidly see substantial commercial relationships. I expect the quarterly trend to continue for the first half of 2026 as we adapt to the global memory constraints. Blaize hardware and software is expected to form a higher mix in our AI solutions from the second half of 2026. This should result in gross margins of between 30% and 35% in the fourth quarter.

Turning to our fourth quarter and full year net loss and operating expenses. The GAAP net loss for the full year was $206.9 million compared to a GAAP net loss of $61.2 million in 2024. I'd like to spend a few moments breaking these numbers down to provide clarity on the underlying results. Key line items in our 2025 financials were a noncash $226 million charge arising from the change in fair value of legacy Blaize convertible notes and warrants, noncash $37.5 million in share-based compensation charge and transaction expenses of $12 million related to going public. These were offset by a $123.2 million credit, inclusive of both cash and noncash components, primarily driven by the change in value of warrants and earn-out shares, among other items.

The adjusted EBITDA loss for fiscal 2025 was thus $50.5 million, up from a loss of $42.7 million in the prior year. The key reasons for the year-on-year increase were $3.2 million in building our teams, investment in our technology road map of $2.4 million, an increase of $1.5 million in marketing and $5 million in new expenses related to our preparations to operate as a public company, some of which are not expected to recur in 2026. I will now review operating expenses on a sequential basis, fourth versus third quarter 2025. In the fourth quarter of 2025, total operating expenses of $14.5 million, excluding $9.5 million in stock-based compensation, were largely flat versus the $14.9 million in the third quarter, excluding the stock-based compensation also of $9.5 million.

Research and development expenses and sales, general and administrative costs were similarly flat quarter-on-quarter. We will invest prudently in people in line with growing revenue opportunities in 2026. Our engineers continue to develop the next-generation products, and we expect related external costs to kick in, in the second half. Our adjusted EBITDA loss in the fourth quarter of 2025 was $11.1 million, unchanged from the third quarter. We ended fiscal 2025 with $46 million in cash and cash equivalents. The available funds under our committed equity facility are $15.6 million.

Now I'd like to spend a few moments talking about our recently announced shelf before moving ahead to guidance for 2026. As is common with companies that become eligible and meet the criteria to file a shelf S-3 registration statement, we took the opportunity to do so on the first year anniversary of our merger. This shelf allows us to raise up to $250 million through a broad range of securities in the next 3 years and on an as-needed basis. Our shelf offers broad flexibility to raise capital quickly when market conditions are favorable. We believe the shelf is helpful for strategic positioning. It will provide working capital needs, fund field trials and enable continued investment in new product development.

Moving to our guidance for the current year. We operate in a dynamic environment that now includes global memory supply constraints and geopolitical tensions. We continue to monitor the supply chain closely and will invest prudently in research and development and go-to-market capability. We see demand across both edge and data center deployments. This creates an opportunity for recurring revenue as we expand our AI services platform. We're continuing our current partnerships as well as adding new customers. We believe new partnerships with recognized names like Nokia should lead to additional strategic opportunities in areas where we have not yet developed traction.

With that, our 2026 fiscal guidance is as follows: revenue of $130 million remains unchanged. I expect the first half to be lighter than the second. Flat gross margins for the first half of 2026 expected to average between 30% and 35% by the fourth quarter; adjusted EBITDA loss of between $45 million and $50 million. In closing, we delivered strong revenue growth in the second half of 2025 and continue to build momentum across our customer base. We remain focused on disciplined cost management and operational execution. Our outlook for 2026 remains consistent with what we have previously shared.

With that, I'll turn it back over to the operator for questions.

[Operator Instructions] And our first question comes from Gil Luria with D.A. Davidson.

You talked in your press release, and we've talked a lot about the different types of applications that are in front of you. I think in the release, you referred to public safety, retail, smart cities, aerial robotics. And I know there's auto coming down the pipe. How would you prioritize them in terms of what you're going to have this year and how those opportunities play out over the next 3 or 4 years?

So I think the commonality is AI inference. This is where we're seeing momentum and comprising our full stack, the silicon, the system servers and our software on top. Specific use cases that -- where we've seen momentum are around a combination of smart health, factory automation, industrial use cases, and we're also part of relationships with drones and such use cases. I don't know if you want to add any further, Harminder.

No, you've -- so Gil, the pipeline that we've got includes all of those. How we set priorities is really the pace at which the -- any POCs or pilots are getting concluded with those customers. As you know, inference requires or inference solutions requires access to data. So in short, the near-term priority is just converting a pipeline where we've got access to those customers and data. And in the medium term, it is how do we expand more business into some of those customers.

And then the second question is about gross margins. I appreciate the visibility into the end of this year. But longer term and at scale, what do we expect our long-term model to look for in terms of gross margins on the hardware side and on the software side and with more of a push to services, do we still expect software and services to be about 1/4 of the mix in a longer-term model? That will help us get the full picture.

So yes, in the longer term for us is 55% plus, and that's going to be a blend of hardware and software. I think what we are observing now with the remarks that Dinakar went through on AI services platform, which essentially becomes a combination of hardware and software. So it's not -- you're not distinguishing between the 2. And there is a revenue share type model that we can see coming our way. So 55% plus as a blend. I think software and recurring revenue, if I can put it that way, could become a larger portion of the mix, but too early to say just yet, and we'll continue to make announcements as and when some of those deployments get public.

Our next question comes from Craig Ellis with B. Riley Securities.

Congratulations on hitting the strong revenue on-ramp in the fourth quarter. I wanted to start the line of inquiry following up on the $130 million revenue guide for calendar '26. Can you help us understand the extent to which Starshine and Yotta are driving that versus other things like maybe converting the Nokia MOU into revenue or maybe even getting traction on some of the new capabilities that we identified in the press release and you've talked about the services platform and AI application delivery.

So let me start and then Dinakar can come in. So yes, Yotta and Starshine are partnerships that we developed late last year. They still remain important to us. The pace at which we deliver products to them is largely driven by their end user needs. There are other partners that we have introduced in the back end of last year. Over the next maybe 3 to 6 months, we expect to add maybe 1 or 2 more partners. So when I stand back and look at it, the revenue guidance is really supported by some of the engagements we've had and what we expect to close during the year. An important point to make is that the AI services platform and the relationship with partners like Nokia are expected to start to feature into -- towards the end of next year.

And really, when we look at the business going forward, it falls into 3 sort of big buckets. Number one is just system revenue, which is a combination of mainly hardware, but it could be Blaize and third party. And number two is there's an attach rate of Blaize software, which gets monetized. And overall, when you look at the system, when you -- particularly when you're applying it to cloud service providers, Tier 2 cloud service providers, it's giving everybody or them an opportunity to start to monetize the infrastructure that they've invested in.

Yes. And just to add, I guess, is that where we were with a couple of key relationships, I think that is actually growing in 2 dimensions. One is within anchor customers, there is a land and expand. It typically starts with one use case. And once we establish credibility, that leads to additional use cases and additional opportunity there. But also once we've developed a certain use case, it is relevant to a larger market. So we are witnessing that momentum as well, where a solution that we develop with a certain partner is required in a different geography, different customer and so on. So we're witnessing that kind of demand as well. The common theme is, of course, our combination of the 3 components that Harminder mentioned, hardware, software and API revenue that we expect will kick in with the launch of our AI services platform.

Got it. That's very helpful. And then the follow-up is really a 2-parter. In past calls and conversations, we've quantified the opportunity pipeline at about $725 million, $725 million, can you give us an update on whether that's still the right way to look at the opportunity pipeline? Or has it changed? And then on the adjusted EBITDA guidance for the year, can you clarify to the extent to which mask set costs are included? Will there be any chip-related mask set costs that we should be incorporating into our OpEx modeling?

Sure. So every pipeline is dynamic, and we've seen meaningful traction in the Asia Pacific region in particular. We have -- Dinakar talked about Stephen Patak joining us as CRO. So he's got -- he's working through -- he's got very good visibility of what's going to support 2026 revenue. And whether we use pipeline as a public measure, for us, it's really about trying to get contracts signed and converted. Having said all of that, the pipeline is still significant. It's very significant. It does change. The geopolitical tensions have had some -- a bit of an impact on some deployments where we're not quite sure when they will come back in.

The Nokia partnership and the AI services platform will add to our pipeline, which isn't in the numbers today. So let me leave that there. And then the second point was about the adjusted EBITDA. As you know, the core design of our chip is common across the road map. And the good thing is that our engineers, in-house engineers whose costs are really in the payroll, they continue to work on adding features and reacting to what's happening in the marketplace. The external costs, so when you're talking about mask sets, that's tape out, that will be in 2027 and beyond. But the early part of the third-party external costs, I see some of those kicking in towards the second half. And that's generally going to be third-party IP that we buy and some of the professional services that we pay for the third-party physical design companies.

Our next question comes from Richard Shannon with Craig-Hallum Capital Group.

Maybe a follow-up on one of the prior questions here, maybe looking at a different angle here on calendar '26. Would love to get a sense of relative to this $130 million guidance for the year, how much of this is in backlog or some sort of commitments here. I think last call, you talked about kind of visibility of $160 million in 2 of your biggest customers. Obviously, we're 1 quarter through that, but love to get a sense of what that support looks like. And then maybe to ask more specifically, how do we think about customer concentration or mix this year within that?

So yes, we do have -- we announced those 2 large contracts. They still remain -- we're still delivering against those. The -- as I mentioned in my prepared remarks, Richard, the pace at which those purchase orders come in are kind of determined by the end user, the customers, what they're going to deploy. We have added new customers into our pipeline, and they too have a -- so backlog may be a different way to look at it.

We'd like to think that if you want -- if you've got a design win and you are -- a customer has a need for either edge or as we'll find out over the coming weeks and months, AI services, customers draw down by issuing purchase orders on us. And so we stay close to them in order to manage our own supply chain and so that we can play with those. I hope that, that helps a little bit more explaining -- understanding why we are comfortable about our $130 million guidance. Yes, you said about customer concentration. I don't know, Dinakar, if you want to add.

The customer concentration is -- we're actually moving beyond our initial customers. As I mentioned previously, the use case, once it's perfected, it's relevant to more customers. So we're getting that pull. And AI inference is growing very rapidly. I mean, there was a point of time, 1 training chip -- for every 1 training chip, there were like 4 or 8. Now we're hitting numbers like 16, right? And this is rapidly changing. So the key message is having a hybridized platform with our AI services that we can deliver into these use cases within the same customer as well as across other customers is seeing quite a bit of momentum.

Okay. Fair enough. My follow-up question here is regarding Nokia, a very interesting and powerful press release you had earlier this year about an MOU here. Would love to understand what are kind of the next steps here, especially announceable steps in this relationship and when you ultimately look for it to be contributing to backlog and eventually revenues. Dinakar, I thought -- or excuse me, I think, Harminder, I think you mentioned in one of your replies maybe talking about sometime end of next year, which seemed kind of a long time process. So I just want to clarify that's what you meant there.

Right. So let me start and Harminder can add. So we are actually quite excited with the whole Nokia relationship. They started almost 6 months ago on a visit to Singapore, where we met with their Asia Pacific leadership. We showed them our platform, and they got visibly excited and then they saw how this allows for a collaboration for them to participate in the AI infrastructure build-out.

So the tangible next steps, right, we are building a joint solution, an AI platform focused on inference needs into their customers as well as customers that we can bring to the table, their networking stack plus our AI system stack, software stack. And this is the joint solution that we will actually demonstrate and launch at GITEX Asia in maybe less than a couple of weeks. There is going to be a joint go-to-market co-selling into their customers, system integrators, cloud service providers and enterprises. And that is the -- those are the near-term next steps.

Yes. Sorry, Richard, I don't know whether I misspoke or maybe you misheard. No. The revenues from AI services platform generally, of which, of course, Nokia will be part as a partner is towards the end of this year. So as Dinakar mentioned, we're launching certain aspects of the platform sooner. And as more and more APIs are developed, that just allows us to expand the population that can start to pay for or utilize these services and, of course, pay us for it.

[Operator Instructions] Our next question comes from Kevin Cassidy with Rosenblatt Securities.

This is Chris Myers on for Kevin Cassidy. I think you guys already answered my question. It was going to be about the revenue timing on the Nokia MOU. But I guess, in general, if you could just talk a little bit more about that broader opportunity set and if there's similar infrastructure wins that could come up that are, I guess, along the lines of this deal?

Yes. We do have similar opportunities that we are working on in other continents. And as they materialize, we'll be sure to update you. There is, of course, Asia, quite a bit of momentum we are witnessing. Africa is another place that we have seen some initial traction. Of course, U.S. as well, there is massive infrastructure happening, and they do want hybridized AI to serve business outcomes, right? Less to do with what's under the hood, but more about, hey, can you solve my business outcome in a certain CapEx and OpEx spend? These are the kind of questions that our team gets asked.

And our solutions are a perfect fit also because we are seeing that the model -- average model size is dramatically shrinking, right? These models rival the larger models, and they still achieve the same business outcome. And this is a perfect fit for our graph streaming architecture. In combination with GPUs, we're able to deliver to this outcome. So we're seeing such kind of momentum. This is across the board, right, wherever our sales teams are present.

And I would now like to turn the call back over to Dinakar Munagala for any closing remarks.

Thank you, operator. Before we close, let me briefly recap. We delivered strong revenue growth and expanded our global footprint, driven by key partnerships, including Nokia and cloud service providers in Asia Pacific as well as our work with state government initiatives in India. We're preparing to launch our Blaize AI services platform in Q2, positioning us to capture the next phase of AI monetization while improving our revenue mix and margin profile.

At the same time, we're seeing a clear shift towards smaller task-specific models that deliver strong performance with far greater efficiency. This aligns directly with our graph streaming processor architecture and strengthens our position as AI infrastructure build-out continues to scale. I also wanted to acknowledge the situation in the Middle East. Our priority remains the safety of our employees, partners and customers in the region, and we are committed to maintaining continuity and stability in our operations. Thank you to our analysts, investors as well as our customers and partners for your continued support. We look forward to updating you the next quarter.

Thank you. This concludes the conference. Thank you for your participation. You may now disconnect.

Hello, and welcome to the Blaize Third Quarter 2025 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. It is now my pleasure to introduce Vernice Pozynski, Investor Relations.

Before we begin the prepared remarks, we would like to remind you that earlier today, Blaize issued a press release announcing its third quarter 2025 results. Earnings materials are available on the Investor Relations section of Blaize's website. Today's earnings call and press release reflect management's views as of today only and will include statements related to our competitive position, anticipated industry trends, our business and strategic priorities, our financial outlook and our revenue guidance for the fourth quarter of 2025 and full year 2025, all of which constitute forward-looking statements under the federal securities laws.

Actual results may differ materially from those contained in or implied by these forward-looking statements due to risks and uncertainties associated with our business. For a discussion of material risks and other important factors that could impact our actual results, please refer to the company's SEC filings and today's press release, both of which can be found on our Investor Relations website. Any forward-looking statements that we make on this call are based on assumptions as of today, and other than as may be required by law, we undertake no obligation to update these statements as a result of new information or future events.

Information discussed on this call concerning Blaize's industry, competitive position and the markets in which it operates is based on information from independent industry and research organizations, other third-party sources and management's estimates. These estimates are derived from publicly available information released by independent industry analysts and other third-party sources as well as data from Blaize's internal research. These estimates are based on reasonable assumptions and computations made upon reviewing such data and Blaize's experience in and knowledge of such industry and markets. By definition, assumptions are subject to uncertainty and risks, which could cause results to differ materially from those expressed in the estimates.

During this call, we will discuss certain non-GAAP financial measures. These non-GAAP financial measures should be considered as a supplement to and not a substitute for measures prepared in accordance with GAAP. For a reconciliation of non-GAAP financial measures discussed during this call to the most directly comparable GAAP measures, please refer to today's press release.

Good afternoon, everyone, and thank you for joining us today. Q3 was a breakout quarter for Blaize, defined by strong execution, commercial traction and expanded global visibility. We delivered a solid quarter with revenue of $11.9 million, up 499% from Q2. We further expect the Q4 revenue to double from here.

To support this acceleration, following the close of Q3, we secured a $30 million investment from Polar Asset Management Partners to scale commercialization and the next-generation chip development as we scale into 2026. Together, these results mark a step forward from validation to scale, demonstrating growing customer adoption and investor confidence in Blaize's strategy and solutions.

We strengthened our ecosystem through two new key partnerships. First, we announced a collaboration with Technology Control Company, or TCC. Second, we formalized a partnership with Reach Digital, the digital transformation arm of Reach Group, a subsidiary of International Holding Company or IHC, one of the largest investment holding with market capitalization of $240 billion.

Blaize's presence on the global stage continues to expand. We participated in the world's most influential innovation forums, the Milken Institute Asia Summit 2025, the GITEX Global 2025 in the Middle East and the Web Summit 2025 in Europe, each reinforcing Blaize's growing role in shaping the future of efficient and deployable AI. Together, these achievements reflect a company executing with discipline and scale and demonstrate validation of the Blaize hybrid AI platform through active deployments across key industries and geographies.

They also validate the next chapter in AI's evolution, a new paradigm we call Practical AI. This marks a turning point for the industry from large models to practical outcomes and from dependence on the cloud to sovereign AI infrastructure that organizations can now own and control. We call this next phase Practical AI, AI that is useful, deployable and sustainable at scale.

First, Practical AI is business-driven and outcome focused. It solves problems that improve safety, productivity and efficiency, helping customers optimize costs and create value across sectors such as smart infrastructure, defense and industrial automation. Enterprises are prioritizing energy-efficient, cost scalable inference while governments are investing in sovereign AI infrastructure that they can own and operate end-to-end.

Second, it is hybrid by design. It combines heterogeneous compute, our graph streaming processor alongside GPUs and CPUs, giving customers flexibility to choose the right fit hardware for each deployment, balancing performance, cost and efficiency from cloud to edge.

Third, it is efficient. Practical AI delivers a clear total cost of ownership advantage, achieving better performance per watt while reducing energy consumption and maintaining responsiveness. Efficiency defines the economics of AI at scale, enabling sustainable and sovereign deployments that work in the real world.

Together, these principles, business-driven, hybrid and efficient define what Practical AI means to our customers and partners. Let me highlight a few programs that illustrate our progress. First, Starshine Hybrid AI infrastructure, a $120 million collaboration with initial shipments in Q3 2025 and continuing through 2026. The partnership will focus on building AI infrastructure for smart city development, industrial automation and public services across Asia.

Second, TCC, the Saudi Arabia sovereign AI infrastructure announced in September. This partnership positions Blaize as a technology enabler of Saudi Arabia's Vision 2030. TCC is working with us to build hybrid AI infrastructure. Together, we are developing energy-efficient AI systems to accelerate adoption across the Kingdom's public safety and infrastructure sectors.

Third, Yotta Smart Infrastructure. Our AI-powered public safety rollout across India continues to advance. We are fulfilling Yotta's purchase order and expect initial deliveries completed this year. Fourth, new partnership with Reach Group announced recently at GITEX Global 2025 strengthens Blaize's position in the Middle East through collaboration on Practical AI solutions and regional infrastructure initiatives.

Beyond these programs, we continue to expand our engagement worldwide through workshops with data center providers, sovereign operators and system integrators while advancing proof-of-concept work in next-generation smart radar, facial recognition and vision AI. Together, these initiatives strengthen Blaize's position as a practical sovereign-ready AI platform partner, helping governments and enterprises deploy AI securely, efficiently and at scale.

On the technology front, Q3 was about execution. We continued the commercial rollout of the Blaize AI platform, integrating hardware, software and orchestration into one unified stack that simplifies deployment and accelerates time to value. The platform's orchestration layer gives customers flexibility for model packaging, deployment and optimization across diverse environments.

In hybrid AI infrastructure, Blaize's GSPs and GPUs work together, complementing each other to boost inference performance and power efficiency. At rack scale, this combined architecture delivers up to 2.4x higher performance per rack and up to 3x better power efficiency, enabling greater performance per watt and lower total cost in real-world deployments.

At GITEX Global 2025, we showcased these capabilities in live demonstrations from city safety analytics and incident detection to autonomous mobility, including ruggedized systems that operate reliably in environments up to 70 degrees Celsius. We are also continuing development of our next-generation chip, working closely with ecosystem partners to extend our leadership in low-power programmable AI.

Next, we have strengthened our capital position. Earlier this week, we announced a $30 million private placement investment from Polar Asset Management Partners, reinforcing confidence in Blaize's long-term strategy and market opportunity. This new funding provides flexibility to advance commercialization to fulfill customer programs and accelerate next-generation platform. It positions us to accelerate the future silicon development, expand ecosystem partnerships and continue executing with financial discipline.

Looking ahead, we expect continued growth momentum in Q4 and into 2026. Our priorities are clear: scale deployments, expand revenue through integrated AI solutions and advance development of our next-generation GSP architecture. In 2026, our focus turns to global expansion of Practical AI, delivering solutions that are efficient, scalable and sovereign capable. We will deepen partnerships to drive adoption across key sectors such as urban AI infrastructure build-out, defense and retail.

Our strategy centers on hybrid AI deployments that combine the strengths of Blaize's GSPs and GPUs across heterogeneous environments, enabling secure, energy-efficient and programmable AI infrastructure. Blaize is helping leading this shift towards real-world sustainable and sovereign AI that bridges innovation with impact and turns technology into tangible progress for industries and societies. We reported revenue of $11.9 million for the third quarter, reflecting strong execution and continued growth. With that, I'll turn it over to Harminder to walk through the financial highlights and our outlook for the remainder of the year.

Thank you, Dinakar, and good afternoon, everyone. I'd like to start with a few highlights. We reported $11.9 million of revenue, beating the upper end of our guidance by $400,000. We beat our Q3 adjusted EBITDA loss guidance by $2 million, coming in at $11.1 million. This reflects better-than-expected execution and stronger operating discipline across the business. And we closed a $30 million financing with Polar Asset Management Partners.

I will now move on to reviewing our financial performance for the third quarter of 2025 in more detail and provide guidance for the fourth quarter. The results that I'm sharing today demonstrate our shift from customer validation to growth at scale. This quarter, we delivered our strongest quarter yet with revenue of $11.9 million, which was a sequential increase of 499%. I'm pleased to report that revenue surpassed the upper end of our prior guidance range by $400,000.

Approximately $10.4 million of the third quarter revenue was driven by the initial shipments of servers under the Starshine contract into the Asia Pacific region, which we expect to collect in full before the end of the year. Gross margin was 15% this quarter compared to 59% in the second quarter of 2025. As I noted in my remarks last quarter, as expected, initial gross margins related to the Starshine contract would be impacted by the higher component of third-party hardware in the system.

Going forward, we're working with Starshine software teams to replace most GPUs in these servers with Blaize GSP cards. This is expected to result in lower average selling prices for customers and improved margins for Blaize in the quarters ahead. We continue to fulfill the Europe purchase order and anticipate that the initial approximately $6 million of revenue contribution to complete this year.

Let's now turn to our third quarter operating expenses, which I will discuss on a non-GAAP basis to exclude stock-based compensation charges. Research and development costs of $6 million were down slightly from $6.4 million in the second quarter and represented a year-over-year increase of 7%. Sales and general and admin costs totaled $8.5 million, largely flat versus the prior quarter and an increase of $3.6 million year-over-year. Blaize remains disciplined on costs as the business grows.

Our third quarter adjusted EBITDA loss was $11.1 million, down $1.8 million sequentially and marginally up from Q3 of last year. This reflects better-than-expected execution and stronger operating discipline across the business. Reported net loss in the third quarter of $26.3 million was lower than the $29.6 million net loss for the second quarter. Both include significant noncash adjustments related to stock-based compensation and fair value charges. The reconciliation between GAAP net loss and adjusted EBITDA is included in our earnings press release.

We are very excited about our November 10 announcement of a $30 million private placement financing by Polar. This investment positions Blaize to continue its trajectory of delivering results from contracts in hand, converting pipeline opportunities into new business and advancing its chip road map. We welcome Polar as a long-term anchor investor in Blaize. We have also taken advantage of recent strong trading volumes to exercise our right to sell common stock to B. Riley under the committed equity facility signed in July this year.

These initiatives have resulted in a significantly improved cash balance of over $60 million today. And combined with expected inflows from current customer contracts, we believe we're strongly positioned to fund our operations well into the second half of next year. I will now share outlook for Q4 2024. Total revenue for the fourth quarter is expected to be between $21.1 million and $23.1 million, almost doubling our third quarter performance. We anticipate adjusted EBITDA loss to be in the range of $15.6 million to $18.6 million, reflecting the variable nature of next-gen chip costs. The share-based charge and weighted average shares outstanding estimates are provided in our earnings press release.

Looking ahead, our pipeline opportunities based on the current generation of silicon remain robust. Approximately $160 million from the Yotta and Starshine deals are expected to support our revenue projections over the next 6 quarters or so. Our partnership with the Kingdom of Saudi Arabia's technology control company is progressing well. We anticipate initial revenues from delivering ruggedized AI boxes capable of operating in harsh high-temperature environments and professional services to begin in 2026.

We expect our recently announced partnership with Reach Digital to significantly enhance our profile as a provider of Practical AI solutions. The Blaize Hybrid AI platform is resonating well in the market, and we look forward to providing further updates on customer progress. Let me highlight upcoming events for the financial community. Blaize will be at the Craig-Hallum Alpha Select Conference in New York on November 18 and at the Wells Fargo Annual TMT Summit in California on November 19. We look forward to seeing you at these events. Thank you. And with that, we'll now open the line for questions.

[Operator Instructions] Our first question comes from the line of Alexander Platt with D.A. Davidson.

Our first question comes from the line of Kevin Cassidy with Rosenblatt Securities.

Congratulations on the good results and the really strong revenue growth. Tied in with the Starshine project, how many more quarters do you think it will be if you [Technical Difficulty] the third-party hardware?

Kevin, this is Harminder. So your voice was a little cracky, but let me just repeat what I understood you to say -- to ask, which is how many more quarters before we start to shift to a GSP-heavy server. Was that right?

That's right. Just when can gross margins start to expand again?

Yes. So we expect in the early part of -- in the latter part of the second half of next year. So work has been going on with both software teams to create this orchestration layer that allows workloads to seamlessly go across both the GSP and the GPU. So as we start to shift -- as we start -- as that work completes, the servers that will start to shift perhaps in the second quarter onwards will start -- will be -- will have more GSP components in them.

Okay. Great. And I'm always interested in your next-generation silicon. Can you give us any hints on what you're targeting with that? And what will be some of the improvements?

Sure. I can take this. So as you know, we're in the key markets that we're in, majority of the customers, we are actually capturing at the business outcome level. And our software is quite coming to life as well, the whole platformization, orchestration layer. Now what this is helping us is taking all of this customer demand and feedback into the definition of the next-generation chip. Certainly, it will be addressing the existing video image visual workloads, but also we're expanding it outside of this into other areas to help us capture a wider set of workloads. So it is a TAM expansion for us when the silicon comes in. But the good part right now, because we're platformizing and our software is playing a role, it already helps us understand what the customers' needs are. And by adding this next-generation silicon, it further improves our margins. But including things like language models, et cetera, are part of it. Also there are newer kinds of AI that are emerging by being programmable and having the flexibility of our architecture, we're able to address all of that.

And our next question comes from the line of Craig Ellis with B. Riley Securities.

Congratulations on the momentum in the business, guys. I wanted to start, Dinakar, just following up on comments regarding TCC and Reach. Is it possible for you to help us scope the size of those two deals? And if not in their entirety, help us understand how material they might be to 2026?

Sure. Happy to. So the interesting thing is both of these are in the Middle East. So let me address the size of the opportunity, not getting into each opportunity size, but the market is rapidly growing there. There is a lot of thrust for AI solutions and particularly Practical AI solutions that can help their cities become safer, the defense entities and so on. So fortunate to have been working with TCC specifically in the Saudi region. And this is all part of the Vision 2030, where they have a pretty big -- we're collaborating on AI solutions there. And Reach is in UAE. So we have solid partners in both these countries. And each of these, as we solidify the contracts, we will start announcing them to the market. I don't know if Harminder wants to add.

No, no, you covered. I just want to make one very important point. I mean the proof points about Blaize technology and particularly where we are able to exist in very harsh thermal conditions, we were the only solution that was able to still continue to do high-performance compute at the edge up to 75-degree centigrade temperatures at a total -- at a cost of -- you can, of course, put a system together and have lots of extra cooling. And these are the kind of things that are now resonating really well with customers.

Exactly. The temperature-grade testing was done, and we passed in flying colors. And as Harminder mentioned, right, and they picked the hottest month of the year, which is September. So we were literally on the rooftop testing these. So we -- all of those results are helping us get into the commercialization phase, which is the next...

That sounds like it's very compelling proof points for your partners, and it should translate well into what your sales can do in other areas with that deployed. So a follow-up question, given the momentum that you have with each of those partners, should we expect there to be meaningful revenue recognition next year? Or will we be in a planning and deployment phase that would precede sales and activity? Just trying to understand when these start to really tip towards revenue-generating partnerships for you?

So Craig, as I mentioned in my prepared remarks, I certainly expect that the TCC relationship will start to contribute towards 2026. The exact timing of that, of course, will depend on how the deployment -- the solutions are deployed. Reach Digital is a relatively recent engagement. However, what we're starting to see is that as certain solutions are deployed with one customer, I think the pace at which some of those other customers in the same industry or same vertical and how quickly they adopt should accelerate.

That's very helpful. And then just building on that point and going back to Starshine, where we are into development. From early development, what are your customers learning about the advantages of the system? And how is that impacting the pace at which they're choosing to move forward?

So certainly, hybrid AI is very popular, which is how we complement GPUs with Blaize GSP for the best business outcome and better cost and better operational expenses. So this is certainly resonating. Plus Blaize is programmable. And therefore, the workloads that can seamlessly move across GPU and GSP is another advantage. So these are the key learnings, and they're applying it to real problems such as smart infrastructure, agriculture and so on, right? So this relationship is growing well.

May I just add one more comment on that, which is the affordability side. If you have a server, which is full of GPUs, the -- I mean, it's the reason why margins are low, but also the customer affordability is impacted. As we start to replace those with Blaize GSPs, the selling price of that server comes down significantly. And our margins, of course, increase because we don't have to necessarily pass all of that benefit on. And that's where we'll see probably an acceleration of adoption of solutions in the back end of 2026.

It's a very good point. And ROI is one of the things that I've always found quite resonant with the solution that you're providing. Lastly for me, I think the last time we spoke in a forum like this, we were talking about a pipeline that would have been quantified at about $725 million. Is that still the right way to look at the pipeline? And any color on where there might be candidates for conversion as we look across the fourth quarter and into early 2026.

So the pipeline still remains robust. It's -- as you know, it's a living beast. The ones that we expect to convert, we've already talked about. Obviously, we continue to ship on our Starshine contract. We continue to work with -- you'll have seen an announcement that enhances the relationship with Yotta into the Middle East. Whether that hits 2026 and so on will depend on how fast we work together. But we feel very strong that -- confident that the pipeline remains strong, and it's based on currently shipping product and the more deployments that we start to make, then as I said, the conversion should accelerate, and we should add more customers that are not in the pipeline today from those verticals into that pipeline. So we'll talk about that more when we do our annual results next year.

And to your point, the whole ROI is the key metric of what's driving this acceleration. As we engage with one customer and they're seeing the ROI clearly, there's a land and expand within the customer, but also these same solutions are relevant across the entire geography, and we're getting that momentum as well.

Our next question comes from the line of Richard Shannon with Craig-Hallum.

I think my first two questions are going to be interrelated. And first one, Harminder, I'd love for you to just repeat one of your last comments in your prepared remarks. I think you mentioned something around $160 million from, I believe, Yotta and Starshine of revenue over the next 6 quarters. Can you verify that I got that right? And I assume that next 6 quarters includes fourth quarter that we're in now to the first quarter of '27. Is that accurate?

That's accurate. It's basically the $120 million and the $56 million that we announced a few months ago, and we're starting to deliver on those. And yes, it includes Q4 of this year. It starts from Q4 of this year.

Okay. And my second question is just following up on the prior commentary you've had on your calendar '26 revenues. If I got my notes right here, that talked about at least $130 million for next year. Is that a number you're reiterating? Or would you change that anyway?

So we're not changing that at the moment, Richard. But we feel confident that the minimum is $130 million, yes.

Okay. Perfect. Let's hear. Maybe a question on OpEx. You mentioned your guidance here for an increase in EBITDA loss here. You mentioned that -- related to the next-gen chip development here. Maybe give us a sense of the degree to which these elevated expenses will continue into next year.

So I'll start on the numbers, and then maybe, Dinakar, you can add a little bit more on -- so as a fabless company, we benefit from actually the core of the GSP, that design, which is around which we have all of the IP that customers are seeing the benefits from. That core design remains constant across our road map. That's number one. So the internal costs of getting additional features, making maybe a chip bigger, et cetera, are disproportionately low. They don't expand linearly. We're then left with is the external costs of getting third-party IP of just having a partner put that IP into our chip. And then, of course, the largest expense is the foundry itself. Those costs, we don't disclose how much those are. They're just commercially sensitive, but are typically paid over a 20- to 24-month period, generally back-end loaded. And we write these costs off through our P&L. So it's the reason why in Q4, for example, I've got a slightly wider range on my adjusted EBITDA just because certain costs will -- particularly IP and so on, we have to pay before we can actually start the work. The NRE is generally spread over time. So it will have an impact on 2026, which we have accounted for. We've been very fortunate in the past of having strong relationships with partners that allow us some favorable payment terms, and we'll continue to pursue those.

Okay. Great. One last question for me, I'll jump out the line. Dinakar, you've announced a number of partnerships and contracts in the last number of months here. All of them are kind of based in Asia, Southeast Asia, which is interesting and noteworthy here. So I'd love to get a sense from you to the degree to which this is a core focus for you? What's kind of your advantage and what's driven your success there so far? And then to what degree do we -- should we expect you to announce Western world or even U.S.-based partnerships and customers in the near future?

Sure. So Asia and Middle East are areas where there's a lot of new smart city developments, et cetera, particularly in the Middle East as well as Asia has a lot of camera infrastructure that we're trying to upgrade. So naturally, there's a good amount of business that's happening there. But our pipeline does pan U.S. as well as North and South America as well as Europe. In fact, and we will -- as we are able to announce, we will, but we're in smart retail kind of use cases in the Americas and also smart restaurants where they're looking at using video analytics for better margins and so on. So there are other Practical AI use cases that we are part of our entire platform, hardware plus software that is undergoing POCs. And as we solidify and we start booking revenue, we'll be sure to announce deals in the U.S. as well.

And our next question comes from the line of Gil Luria with D.A. Davidson.

Glad I was able to get through. It sounds like you're on track for this year, on track for next year. You're building up the book of business mostly through relationships. So I wanted to ask in terms of the conversion of the pipeline, is this strategy going to continue to be mostly focused on the partners that you're accumulating? Or is there more of a thought to also having more direct sales as you have opportunities with bigger customers?

Thank you, Gil. I can start and Harminder can add. Most of the large customers that we're engaged with, there is -- as they deploy, there is repeat business there. There is an expand of scope within those customers itself, and these are pretty large customers. So that opportunity exists. And more importantly, the -- or equally important, the use case that we are perfecting with this one particular customer is relevant across the geography, like what we do in Saudi Arabia is relevant for, let's say, UAE, Qatar and others. So there is an expansion within the geography as well, especially the example that Harminder gave, we are able to withstand harsh temperatures, outdoor settings and deliver smart infrastructure use cases, and there's massive construction that's happening there. So there is both within the customer and across. And some of these engagements, which are at various stages of POCs, et cetera, are with large direct customers. And while we can't name them today, as these get solidified, we will. So the answer is we have both channel as well as direct.

So you mentioned it now two or three times, so I have to ask how hot was it on the roof in Saudi?

It went to almost -- I think north of 80 degrees centigrades, I think, is what we -- and yes, it was very hot.

Well, that's some commitment on your part.

And our next question comes from the line of Scott Searle with ROTH.

Maybe just to quickly follow up on the qualified pipeline. I don't think you gave a number, but I'm wondering if you could just provide directionally, has it continued to increase and the diversity of that pipeline, has it continued to expand? I would imagine, given some of the announcements that you made, it's getting a little bit more diverse. And also as part of that, looking at some of those qualified opportunities, are these more GSP-heavy deployments out of the gate, so we would expect as you convert and deploy that these should have higher gross margins at the start of the contracts? And then I had a follow-up.

Okay. So the pipeline -- the way that we look at the pipeline, there is a gross pipeline, which is -- it's a significant pipeline. It's what -- when we first qualify opportunities, we say, okay, what's the likely outcome over the next 2 to 3 years from this customer or from these groups of customers? And then when we -- what changes for us is depending which customer is working faster or slower through the POC process. It allows us to put a higher percentage weighting, if you like, in terms of when things will close. So when I stand back, there is -- the $725 million number, anything that was not imminent was already out of that number. What I have not done yet is some of the new engagements that we've got. We've got some high-level indications of what these might mean for us over the next couple of years. But when we do our annual results announcement next year, we'll provide a lot more detail around conversion and so on. The second thing to say is the Starshine deal is one of the -- probably the only one in the pipeline where we are doing within the same box replacement of a GPU. And that's where the margins are low and they'll become higher. If you look at some of the other deals that we've got, we are -- our servers are coexisting with a GPU server in a data center. And so that server has -- is full of Blaize cards and it has Blaize software on it. So for example, the Yotta one, we talked about 15% of that being software revenues. So it kind of depends, but more -- Starshine happens today happens to be the only one where we are at this low margin going up to higher margin.

So the only other thing I'd add is that besides the existing smart infrastructure and defense, the industrial automation is something that's coming up. You asked us about trends in which areas. And the kind of adoption that's happening is including the software platformization, the software layer that we have. And that means higher margins as well for us in those outcomes where we participate with the customer at the business outcome level. So those are all helping us at the platform level, consuming our software plus our Blaize service.

Very helpful. And if I could, just a question on the competitive landscape. You guys have done a good job of not only from a product standpoint, but developing the ecosystem around it, which is driving that opportunity set. I'm wondering what you're seeing out there as you're going to customers. Is the competitive landscape getting a little bit more competitive or thinning? Guys have certainly been able to go out and raise capital, but I think they lack the ecosystem development around that. So I'm just kind of wondering what you're seeing out there in the trenches in terms of the competitive landscape.

That's a very good observation that software is very important because it solves the end application. And in that category, customers typically prototype on GPUs when it comes to actual deployment, CapEx and OpEx budgets are very critical, especially in the world of -- in the physical world. And therefore, complementing Blaize servers is the way to go. That's how they achieve their CapEx results and within operational margins. So -- and coming to competition, right, we -- the productized solutions that exist pretty much there's a couple of names, I think a handful of names we come across. It's -- and customers we are -- the places we're winning, it's because of our -- the combined advantage that we bring to the table in terms of TCO advantage with helping them with the CapEx and OpEx and fully productized. So in that area, we don't come across because we are at the business outcome level, that's what Blaize gets picked.

Yes. The key word there, Scott, is programmability. There are other competitive solutions that might do one or two things. And what we've realized that whilst there may be a place for that in certain parts of the market, but the kind of customers that want to deploy AI at scale want it to be a customizable programmable solution, right? So no longer are we having a conversation, how many TOPS do you have on your card? It is, can I run my real-world application at a cost that makes sense for me. And if you're, by the way, a Tier 2 cloud service provider who's making 0 money today by running all of your infrastructure on GPUs, well, with -- working with Blaize, you now have a chance -- more than a chance of providing services to customers and making money.

Thank you. I'll now hand the call back over to CEO, Dinakar Munagala for any closing remarks.

Before we close, I wanted to share a few quick highlights of this quarter. We delivered $11.9 million in revenue, beating the upper end of our guidance and marking a 499% sequential increase. We're confident that Q4 revenue will reach nearly double our Q3 performance, reflecting strong momentum heading into 2026. Excluding noncash adjustments, we beat our Q3 adjusted EBITDA guidance by $2 million, reflecting stronger execution and operational discipline across the business. We began initial shipments under the Starshine contract into the APAC region, which we expect to collect in full before end of the year. And we closed a $30 million investment with Polar to accelerate commercialization, next-generation chip development and expansion across key markets. Finally, I want to recognize our team for winning the second place at the Milestone Systems Developer Summit in Copenhagen today with our emergency first responder VLM. It's a great example of how Blaize technology is making cities safer and smarter through innovation. You can find the award-winning video demonstration on our Blaize AI YouTube channel. And thank you for our customers, partners and investors for your continued confidence. We're proud of what we've achieved this quarter and even more excited about what's ahead. Thank you.

Ladies and gentlemen, thank you for participating. This does conclude today's program, and you may now disconnect.