Image source: The Motley Fool.

DATE

Thursday, Aug. 13, 2026, at 5 p.m. ET

CALL PARTICIPANTS

• Chief Executive Officer-Dinakar Munagala
• Chief Financial Officer-Harminder Sehmi
• Chief Revenue Officer-Stephen Patak

TAKEAWAYS

• Revenue -- $12 million, driven by momentum with the NeoTensr account and increased adoption of the integrated solution stack.
• H1 Revenue -- $14.7 million, compared with $3 million in the same period last year, representing a 390% increase.
• Fiscal 2026 Revenue Guidance -- $40 million to $43 million, reflecting the exclusion of opportunities pending binding terms and the impact of higher memory pricing.
• Gross Margin -- 8%, compared with 58% in the first quarter, due to a revenue mix weighted toward lower-margin third-party server hardware.
• H2 Gross Margin Guidance -- 17% to 19%, reflecting a planned shift toward higher-margin branded hardware and AI Services.
• Adjusted EBITDA Loss -- $20.9 million, compared with a loss of $13.9 million in the prior quarter, primarily due to a significant receivable provision.
• Fiscal 2026 Adjusted EBITDA Loss Guidance -- $62 million to $65 million, driven by lower revenue expectations and the impact of the Starshine provision.
• Cash and Cash Equivalents -- $36.8 million as of June 30, 2026, supported by $32.8 million in net proceeds from an equity offering.
• Server Agreement -- $70 million for 2,000 servers at current memory prices, with approximately $20 million expected to be recognized as revenue in the second half of 2026.
• Contracted Backlog -- $50 million expected to be carried into 2027, based on binding, noncancellable purchase orders.
• Starshine Receivable Provision -- $7.1 million, representing a full reserve of the outstanding balance as the company reevaluates the commercial relationship.
• Operating Expenses -- $31.5 million, an increase of 32% sequentially, reflecting the Starshine provision and higher investments in new chip development.
• Research and Development Expense -- $10.5 million, which included $3.7 million in stock-based compensation and costs for third-party intellectual property.
• Customer Payments -- $9.4 million received during the second quarter.
• Forward Buying Investment -- $8 million to $9 million invested in components and inventory to manage supply chain costs over the next six to nine months.
• Related Party Settlement Charge -- $2.8 million, representing a one-time noncash charge incurred during the quarter.
• European Purchase Order -- Thousands of units, marking the company's first major order and shipment into the European market.
• Next-Generation Chip Investment -- $1 million additional investment during the quarter as development continues for the staged hardware program.
• Stock-Based Compensation Guidance -- Approximately $34.7 million for the full fiscal year.

Need a quote from a Motley Fool analyst? Email [email protected]

RISKS

• Sehmi stated, "With respect to Starshine specifically, we made the decision not to engage further until Starshine pays its outstanding balance," noting that there is meaningful uncertainty regarding whether the relationship will progress.
• Munagala noted that "memory pricing has risen materially this year and we expect that to persist," identifying supply chain cost inflation as a primary headwind to the revenue outlook.
• Sehmi warned that regional uncertainty has pushed a smart city opportunity in the Middle East into an "extended field trial," delaying the expected conversion to recognized revenue.

SUMMARY

Blaize (BZAI -2.94%) management reported a significant reduction in full-year revenue expectations as the company transitions toward more conservative forecasting based on binding agreements. The strategic focus has shifted toward the economics of AI inference workloads, specifically targeting physical AI in autonomous systems and national-scale hybrid AI platforms. To address rising supply chain costs and extended customer deployment timelines, the company is implementing cost-reduction measures while prioritizing the development of its next-generation hardware architecture. The company stated that it is seeking nondilutive debt financing and advance payments from customers to manage working capital demands.

• Munagala stated that applications and AI services represent sustainable business models compared to GPU rentals.
• CRO Stephen Patak stated, "the customer needs sustained real-time performance inside a fixed envelope rather than peak benchmark numbers," when describing the company's architectural advantages.
• The company is in active discussions regarding a 150-megawatt national government data center program that would involve professional services and AI architecture advising.
• Management noted that a second sovereign program with a similar partner model is also under discussion with a different government.
• The next-generation AI inference product currently under development will incorporate confidential computing capabilities specifically designed for sovereign customers.
• Blaize expects the first revenue contribution from its AI Services software suite to occur within the 2026 fiscal year.

INDUSTRY GLOSSARY

• GSP: Graph Streaming Processor; Blaize's proprietary AI computing accelerator architecture.
• SDK: Software Development Kit; tools provided to developers to build and deploy applications on specific hardware.
• AI Services: A software suite designed for AI inference orchestration and model optimization across hardware.
• Inference: The process of running data through a trained machine learning model to generate predictions or outputs.
• LPDDR: Low-Power Double Data Rate; a type of synchronous dynamic random-access memory used in energy-efficient systems.
• NeoCloud: A cloud architecture designed for deploying AI services and applications.
• NeoTensr: A major customer account for Blaize that utilizes the company's full AI solution stack.

Full Conference Call Transcript

Operator: Good afternoon, everyone, and thank you for joining Blaize's Second Quarter 2026 Conference Call. Before management begins the prepared remarks, we would like to remind everyone that earlier today, Blaize Holdings issued a press release announcing its Second Quarter 2026 Results. Earnings materials are available on the Investor Relations section of the Blaize 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 forward-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 No. 1 on 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 Dinakar Munagala, Chief Executive Officer of Blaize Holdings.

Dinakar Munagala: Thank you, and good afternoon. With me today are Harminder Sehmi, our Chief Financial Officer, and Stephen Patak, 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 our 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. What that number does not show you is what we have already secured. We hold a signed agreement covering 2,000 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. Second, other opportunities are still in progress and expected to close later than we forecast. And 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 we 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. It changes shape into constant tuning and specialization.

These sites run many models on many kinds of chips. And they're built on purpose not to depend on 1 vendor. Both are hybrid, and the operators have learned something important. Renting 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 1. First, revenue. We earn revenue in 2 ways. 1 is our silicon and SDK 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 sits AI Services, our 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%, reflecting a mix weighted to third-party hardware.

Our branded hardware and AI Services is what we expect will shift that mix. And 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. Serving 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, 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 inference 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. Harminder?

Harminder Sehmi: Thank you, Dinakar, 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 that, and how we're managing the balance sheet through this transition. Dinakar 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 3 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 the methodology matters as much as the figure.

As mentioned, our revised guidance is between $40 million and $43 million for the full year 2026. 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 30, 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 NeoTensr 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%, compared with 58% 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 Blaize. The first quarter gross margin benefited from a mix of higher-margin Blaize 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 add-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, 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 outlook. Stephen?

Stephen Patak: Thank you, Harminder. First, let me start by adding 1 additional comment on the quarter. Our Q2 revenue was driven by our momentum with NeoTensr, 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. We are 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 ramp later than we had forecasted. Those deployments are moving now and the pace is picking up. 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 is working. We 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 2 primary revenue engines for Blaize moving forward.

First is aerial 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 Blaize, does better than a general-purpose part.

We are being pulled into these opportunities across every market where we have presence. Once we qualify our solutions, opportunities arise for recurring revenue. Our second revenue engine is a hybrid AI platform, which we brought to market 2 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.

In 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.

As an example, there's a national government program building out a data center in the range of 150 megawatts. Our role is in active discussion, and while there are many details to work out, the intent on both sides is to move toward binding terms. We have a 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 2 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.

Operator: [Operator Instructions] Our first question comes from the line of Kevin Cassidy of Rosenblatt Securities.

Kevin Cassidy: I guess, just to 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 slowing them down that maybe there's not as good return on investment? Maybe just a little more detail around why are there so many delays.

Dinakar Munagala: 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 Blaize. So that's the delay. I'll let Stephen jump in on that as well.

Stephen Patak: Sure. For 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 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.

Dinakar Munagala: Right. And just to -- the underlying demand is definitely there, and that's where the whole backlog comes in that we've announced based on contracts.

Kevin Cassidy: And then, opening up Europe, that's great news. And just wonder if you could compare the opportunities that you have in Europe? How do they compare to the U.S. or Southeast Asia that you've been winning? Are they faster time to market or are they -- I guess, or is it just more shots on goal having another territory opened?

Stephen Patak: Yes. No, they're -- this is Stephen again. Look, they're very similar and in line with where our revenue engines are -- across all 3 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. And naturally, you mentioned Asia Pacific, 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, 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.

Operator: Our next question comes from the line of Gil Luria of D.A. Davidson.

Gil Luria: Based on the -- you gave something on Starshine and NeoTensr. Based on the other business that you have and the backlog that you're looking at right now, can you give us an early sense for what the picture looks like in terms of revenue and revenue growth for 2027?

Harminder Sehmi: Sure. The $50 million that we're talking about is essentially a commitment from NeoTensr, 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 2.5x to 3x where we are guiding this year to end at.

Operator: Our next question comes from the line of Craig Ellis of B. Riley Securities.

Craig Ellis: The first question is really a clarification inside of the updated calendar year '26 guide. So our midpoint is $41.5 million, and that would imply after the revenues in the first half of the year about $26.5 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?

Harminder Sehmi: Hey Craig, so I would see, I'm expecting Q3 to be similar to Q2, and Q4 will be a little bit heavier than that. Any upside that some of the things that Stephen 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.

Stephen Patak: Maybe I can answer -- 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.

Craig Ellis: 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 4 national-scale hybrid AI platforms, can you just help us understand 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 4 different opportunities?

Stephen Patak: 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 4 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.

Dinakar Munagala: Yes, and just to calibrate, I think the trends that we're witnessing are important, everything 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.

Operator: [Operator Instructions] Our next question comes from the line of Richard Shannon of Craig-Hallum Capital Group.

Richard Shannon: I guess the first 1 is on you talking about an elongated sales cycle here. I guess 1 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?

Stephen Patak: 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.

Richard Shannon: 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 engagements to reaccelerate 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?

Harminder Sehmi: Hey, Richard. So we're doing a couple of things. Number 1, 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 $8 million and $9 million 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 6 to 9 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, 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.

Richard Shannon: Okay. That makes sense. And 1 last question for me, and I'll jump out of line here. You mentioned these 4 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 press release and 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.

Stephen Patak: 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, 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 1 we've mentioned in the past, 1 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.

Operator: 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 Dinakar for closing remarks.

Dinakar Munagala: Thank you all for joining us. Let me close with the 3 things we are driving from here. First, conversion. Turning what we are pursuing into committed orders so 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 3 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.

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