Blaize Holdings, Inc. Q1 FY2025 Earnings Call

Good day and thank you for standing by. Welcome to the Blaize, Inc.'s 2025 Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speakers' presentation, there will be a question-and-answer session. Now, I'd like to introduce your host for today's program, Lana Adair, Investor Relations.

Thank you. And good afternoon, everyone. We appreciate you joining for Blaize's first quarter of 2025 earnings call. We are very pleased to be joined today by CEO, Dinakar Munagala, and CFO, Harminder Sehmi. Before we start, we'd like to remind you that the following discussion contains forward-looking statements within federal securities laws. All statements other than statements of historical fact are statements that could be deemed forward-looking, including, but not limited to, statements regarding our competitive position, anticipated industry trends, our business and strategic priorities, and our financial outlook for the second quarter of 2025 and the full fiscal year 2025. These statements are neither promises nor guarantees and undue reliance that should be placed on them. Such forward-looking statements involve risks and uncertainties available in our SEC filings that may cause actual results to differ materially from those discussed here. Additional information that could cause actual results to differ from the forward-looking statements can be found in the risk factors section of Blaize's most recent annual report on Form 10-K for the year ended December 31, 2024, as updated by our periodic reports filed subsequently to that 10-K. Any forward-looking statements on this conference call including responses to your questions are based on management's reasonable current expectations and assumptions as of today, and Blaize assumes no obligation to update or revise them, whether as a result of new information, future events, or otherwise, except as required by law. The following discussion contains references to certain non-GAAP financial measures. The company believes that these non-GAAP financial measures are useful to investors as supplemental operational measures to evaluate the company's financial performance. For a reconciliation of each of these non-GAAP financial measures to the most directly comparable GAAP metrics, please see our website. Details of our results and additional management commentary are available in our earnings release, which can be found on the Investor Relations section of our website. Finally, this call is being audio webcast on our investor relations website and an audio replay will be available on our website after the call ends. Now we'd like to turn the call over to Dinakar Munagala, Chief Executive Officer of Blaize.

Thank you for joining us for our second earnings report as a public company. Since our last update, we're now moving from pilots to real deployment. Our large pipeline includes active late-stage programs with multiple contracts advancing from field validation into final negotiations. We've begun shipping production AI chips and systems into real-world deployments across smart city and different sectors, with momentum fueled by new partnerships in Q1 and Q2, including CBIST in South Korea and Turbo Federal in the US. We've also been developing a new turnkey vertical AI solution stack designed to combine our chip, software, and validated applications to help accelerate deployment in key sectors. We look forward to announcing this offering soon. We are operating with financial discipline and a clear focus on execution, supported by a balance sheet aligned with our near-term and mid-term priorities, with feasibility to fund operations, scale customer programs, and support strategic growth through 2026. For those who are new to Blaize, we build AI processors and software that bring real-time influence to the physical world, starting with video and vision and expanding into true multimodal edge intelligence. We do this with a unique combination of energy-efficient chips and integrated software built specifically for environments where speed, power, and cost really matter. From processing satellite imagery on orbiting platforms to fusing LiDAR and video streams for perimeter security at airports and smart cities to enabling object detection and license plate recognition on roadside cameras and drones, our platform delivers real-time AI inference. For example, today most cameras can see, but they cannot understand. Video systems capture enormous amounts of data, but they rely on costly manual operation to make sense of what is happening. This is where Blaize leads. We don't just talk about AI; we deliver real-time deep inference for physical environments, enabling intelligence visually and contextually and reliably. In the first half of 2025, we moved from development to real-world deployment. In Q1, we began shipping production chips and AI accelerator cards into smart city programs. We were selected by CBIST to lead the Chungbuk digital innovation hub in South Korea, delivering edge AI infrastructure for regional smart city deployments. That momentum carried forward, and now in Q2, we're executing across geographies with active engagements in key US infrastructure initiatives, expanding into South America, Asia, and the Middle East. At ISC West, the largest physical security event in North America, we showcased AI-powered smart city solutions across critical applications. Examples include perimeter defense, school safety, campus monitoring, and public surveillance. Adding further strength to these applications, our technology is now integrated with OrionVM and VSaaS, delivering a hybrid cloud model for smart surveillance, Thrive Logic for school safety solutions, and CVEDIA for pre-trained vision AI models. This all demonstrates Blaize's ability to deliver deployable partner-validated solutions for smart infrastructure. Our diversified global pipeline includes late-stage engagements across smart infrastructure, public safety, defense, and industrial automation. Field deployments are underway, and several programs are advancing into the contract stage from field testing. Our partnership with Turbo Federal has already yielded an opportunity for 2025, focused on delivering trusted AI-powered inference and perimeter security in defense-critical environments, starting with video and expanding into situational awareness. We're finalizing a purchase order for Blaize-powered servers bundled with AI Studio, our orchestration software, bringing integrated compute and intelligence to customer hands. In parallel, our engagement with the Ministry of Defense continues to advance towards proof of concept and field qualification phases. This momentum reflects our efforts to build trusted relationships with high-value customers through targeted engagement, strategic pilots, and solution co-development across the US, Asia, and the Middle East because we are finding faster procurement cycles and government-led AI initiatives. These efforts are driving deeper engagements, strengthening our pipeline, and converting into signed contracts and revenue. This traction informed how we built our next step. Today, if you want smarter cameras or systems, it usually means stitching together parts from different vendors, which takes time, money, and technical help. Customers have made it clear that they don't just want raw components; they need ready-to-deploy solutions tailored to their use case. I'm pleased to share that I will soon be introducing a new pre-integrated vertical AI solution platform like a ready-to-deploy turnkey AI stack in a box, combining our energy-efficient chip architecture, integrated software, and proven applications from best-of-breed ecosystem partners. It's designed to streamline deployment and deliver faster results in key sectors like defense, smart infrastructure, and public safety. Think of it like what Apple did with the Macintosh. It wasn't just a computer; it was a complete experience, a comprehensive tool that made advanced technology accessible and usable. With this upcoming vertical AI platform, what used to take three to six months of custom integration could soon be field ready in half the time. And early adopters are seeing up to 60% lower total cost of ownership. As we move into Q3, we anticipate sharing formal proof points and success stories from these deployments. To support the shift towards a solutions-first approach, we're executing a dual go-to-market strategy. On one side, we worked directly with high-value customers, especially in smart infrastructure and defense, where trust, integration, and long cycle deployments matter. These include engagements with defense ministries, smart infrastructure projects in Korea, and national initiatives in India. In each case, Blaize provides the platform intelligence powering critical systems in cities, transportation, defense, and public safety. On the other side, we scale through channel partners, leveraging our turnkey vertical AI platform as the foundation. These are full-stack, vertical-ready offerings built in collaboration with best-of-breed edge AI application partners. For example, Videonetics in smart airports, Visibility in smart retail and defense, alwaysAI in mining, Thrive Logic, and CVEDIA in school safety. Each partner brings proven deployable applications. Blaize provides the compute and orchestration to run them in the field. This combined approach gives us the flexibility to serve high touch accounts while accelerating growth across broader commercial markets. Here's where we're heading. We bring together what systems see, the real video and the sensor data at the edge with what they understand, the language models running on the data. This is multimodal intelligence in action, not just vision, not just language, but the fusion of both, video, sensors, speech, and structured data to create context, to make decisions, and act in the physical world where it matters most. Think of it like this. Small, efficient models tuned for real-world tasks, like those small language models you've been reading about. They're faster, lighter, more secure, and they don't need a data center to make sense of what's right in front of them. That's the future we're building towards, and it's already starting to take shape. This combination of edge processing, scalable software, and partner-driven deployment is how we deliver value today. Because what we're building at Blaize isn't just about chips or code. It's about enabling intelligence where the world needs it most, in the field, in motion and in the moments that matter. This quarter reflects that commitment in action. From pilots to deployments, from concept to customer, we are executing with clarity and building momentum across key verticals. Thank you for your continued support. Now I'll turn it over to Harminder for the financial update.

Thank you, Dinakar. Good afternoon, everyone. In the first quarter of 2025, our revenue surpassed $1 million, slightly exceeding expectations, mainly from shipments of our PCI cards for smart city customers. Last year's revenue was $549,000, primarily from strategic consulting services for an automotive client. This quarter's cost of revenue was $327,000, compared to $306,000 last year. Earlier this year, our company completed a business combination and was listed on NASDAQ. Like other companies that have gone through similar processes, we faced one-time cash transaction costs and accounting adjustments related to the merger, which can obscure the actual performance of our operations. I will first explain these adjustments and then discuss our underlying performance, which offers a clearer view of our operations as we grow the business. In this quarter, our GAAP net loss amounted to $147.6 million after accounting for $109.6 million in other expenses and $39 million in operating expenses. This contrasts with a GAAP net loss of $16.7 million last year. Among the other expenses, $226 million was charged due to changes in the fair value of convertible notes and warrants, counterbalanced by a $116.5 million credit concerning the earn-out liability. The net effect of these two items accounted for much of the $109.6 million adjustment. Last year's other expenses were $8.4 million, mainly related to fair value changes of convertible notes and warrants, both non-cash adjustments. Our total operating expense of $39 million this quarter included one-time transaction costs of $12 million and non-cash share-based compensation of $11 million. The rise in stock-based compensation of $10.7 million this quarter was largely due to our merger completion, which triggered certain RSU vesting conditions. The underlying operating expense for the first quarter was $15.9 million, reflecting an increase of $7.6 million from the previous year. This rise was due to higher employee costs as we improved our market and customer support capabilities, investments in software and third-party IP for our next-generation chip, and expenses related to preparing our company for public market operations. Research and development costs were $7.1 million in the first quarter, up from $3.9 million the previous year, with the increase mainly due to software tools, third-party IP, and greater employee costs. Our selling, general, and administrative expenses for this quarter, excluding depreciation and amortization, were $8.3 million, an increase of $4.5 million from the prior year, driven by expanding sales teams globally, increased marketing costs linked to our NASDAQ listing, higher support function costs such as audit and legal fees, and rising insurance costs. Overall, this foundation supports our future revenue growth. Consequently, our adjusted EBITDA loss for the first quarter of 2025 was $15.4 million, compared to a loss of $7.5 million in the same quarter last year. I expect adjusted EBITDA to improve each quarter throughout 2025. As of March 31, our cash and cash equivalents totaled $45 million, down from $50.2 million on December 31, 2024. Following the business combination, we received an additional $15.3 million in equity and settled one-time transaction costs of $4.5 million. After accounting for annual insurance and software licenses of $3 million in this quarter, the underlying cash used for operations, including inventory, totalled $13.1 million. Earlier, Dinakar highlighted the growth and momentum in our revenue pipeline. Product shipments began in the first quarter, and we anticipate revenue growth to accelerate as we progress through the year and into 2026. We aim to expand our partnerships with selected ecosystem partners already announced and expect that our revenue pipeline will improve as these partners create customer opportunities for us. We continue to conduct trials and proofs of concept with customers to integrate their AI workloads onto Blaize hardware, which is a crucial step in showcasing the advantages of our programmable architecture. We foresee a faster path to production as subsequent customers in selected verticals adopt our solutions more rapidly. Recent tariff developments have led many firms to expedite inventory procurement, potentially straining the supply chain for chips and finished products, but we maintain strong relationships with our key vendors. Our current inventory, along with projected deliveries from existing orders, is expected to meet the immediate demand for the next two quarters. Together with our partners and customers, we will closely monitor any pipeline conversion impacts due to the increasing macro challenges over the next seven months of this fiscal year. We are being disciplined in managing our resources and will take steps to reduce our cash burn if uncertainty persists. For our larger revenue opportunities, we will seek advanced payment commitments from customers to minimize cash tied up in working capital. I believe these measures, combined with expected cash from revenue this year, will allow us to extend our cash runway into the first half of 2026. For the next quarter ending June 30, 2025, we project revenue to be between $1.5 million and $1.7 million. The adjusted EBITDA loss is anticipated to be between $13 million and $14 million. The stock-based compensation charge is expected to be around $10 million, with approximately 90 million shares expected to be outstanding. We are reaffirming our revenue guidance for the full fiscal year ending December 31, 2025, at a range of $19 million to $50 million.

Gil, your line is open, from D.A. Davidson.

First question is about some of these new wins and how those are a little bit more on sizing and timing of the wins at CBIST and Turbo Federal in terms of when you expect those to come in and how can you help us quantify the size of those wins please?

These are strategic engagements for us in our chosen markets. CBIST, for example, specifically around smart cities and Turbo Federal is around defense primarily. So, it's our standard products that they would use into these verticals and combining our hardware as well as software. As Harminder mentioned, we're negotiating contracts. We will start announcing them as we start signing them.

The only thing I would add is that they range in size from the low millions to tens of millions over a period of time. So, the start point is towards the end of this year for some of them, but they extend our pipeline into 2026.

To continue on that path, can you remind us what the revenue cycle is? So at what point do you recognize revenue from these different types of customers? Is it only when they turn on the new capabilities? Is it when they receive shipment? And how that relates to both these customers as well as your Gulf Ministry of Defense customer?

Generally, we sell hardware. We sell solutions with our partners which is a combination of hardware, software, and professional services. So each of those have their own revenue recognition. When we're shipping product into the customer, we recognize that revenue straight away. When it is a software license, it depends whether it's a perpetual license. As soon as that license becomes operative, then we recognize the revenue then. If it's an annual license, of course, it's spread over the period. Professional services is recognized as and when we deliver those services.

The last one from me is a little bit of a longer-term one. What's the update on the next generation of chips that you're planning and how has the very rapid development in the quality and efficiency of the AI models impacted what capabilities that next generation is going to have?

First of all, we are very excited that the kind of workloads that we're seeing and the changes that are coming about in the shrinking of AI models are a perfect fit for existing products in the market. In addition to this, of course, this is all proving to be exactly the direction we set out as we define our next-generation products. As we launch our next-generation products, we will, of course, provide an update. But all in all, the key aspects of efficient inference at the edge, having complete programmability, and supporting these shrinking multi-modal AI; these are the key facets that are the growth drivers. And we're seeing this live in action right now as we engage with customers and also driving our pipeline.

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

My first question is, in the large contract that was announced even before the start of your SPAC closing earlier this year here and you've got a bullet in your press release here about progressing through the proof of concept and field qualification stages. I know in the original announcement around this year hoping to have proof of concepts done by last month in April. So wondering if maybe you can kind of clarify what's going on there? Should we view this as some sort of delay? Because you sound like there's a deepening engagement here. So I'd love to get a little more clarity on what's going on here.

Of course, we are in constant communication with them. And we have done various proofs of concept. The exact field trial is being discussed as we speak. Everything that you've said in terms of revenue towards the end of the year is still good.

No, I think you've addressed it. So we always signal that revenue from this contract, it's a multi-year contract, it has several use cases, and we're going to start deploying towards the end of this year. Again, it's one of those classic combinations of hardware shipments; it's got software licenses; and some professional services. And we expect that to extend beyond 2026 and into early 2027.

I'll touch on the pipeline here. Maybe if there's any way that you can quantify or characterize how that's improved here since the last earnings call and frankly the beginning of the year, that'd be also helpful.

When we started off at the beginning of this year, based on all the customer engagement that we have, we signaled the pipeline of $400 million and the guidance. The pipeline is actually growing rapidly in two ways. First of all, once we go ahead and land a particular ISV and deliver to a certain use case, that ISV's pipeline also becomes ours. They want to get us into other deals. So we're actually seeing a pipeline amplification factor this way. The other part we're witnessing solid momentum is true multi-modal AI. The merging of computer vision and small language models, the combination of that creates new and interesting use cases and value to end customers. So we're seeing growth based on that as well.

One thing I'll add to that is, Richard, all of this pipeline is based on our currently shipping generation of products.

Your announcement of this partner, CBIST, in South Korea seems very interesting, and I guess I'd love to understand both the kind of the strategic development cycle leading up to that announcement, and then over what time period should we expect to see some tangible financial results from this? It sounds like, if I caught your comments correctly, towards the end of the year, which I guess certainly would make sense, but I guess I'd love to have you describe how that progressed to this point and where you expect it to go forward?

We have been engaged with them for almost a year now. So there's been work behind the scenes that went in leading up to this initial announcement, but there are active deployment cycles that we are discussing with them. And as Harminder signaled, it is revenue towards the end of the year. As we begin announcing specific projects, of course, we will set up time with you and go through specific things as well.

And our next question comes from the line of Kevin Cassidy from Rosenblatt Securities.

I saw in your press release that you're bringing your adjusted EBITDA loss down to $40 million from $55 million from $70 million to $75 million. But your stock-based compensation is going up. I wonder if you'd discussed that, maybe if you could repeat it just to understand what the moving parts are.

The $40 million to $55 million range is just reflecting our revenue range, so it depends on which side of the map we're in. So the current EBITDA range that I've guided on just reflects that revenue range. The difference between last time and this time is essentially the external costs of our next-generation chip. And whilst all the work is still continuing, we are negotiating payment terms with our key vendors right now. And we expect that some of these costs will be deferred towards the end of 2025 into 2026 onwards. So that's the big change.

New vertical AI solution platform, that seems very interesting. You've mentioned a few times that you do have software contracts, but would this accelerate maybe shipping everything with Blaize software on it? And would it turn into a recurring revenue stream?

We actually are quite excited about this. The whole vertical AI stack in a box, what it brings to the end customer is, I would say, three main aspects. The first thing is it's hardware plus software; everything, including our software as well as partner software for specific use cases, is packaged and ready to go, just like a Macintosh. You pull it out of the box, and it just works. The second aspect helps customers, especially in the edge enterprise, they're not as AI savvy. So this actually helps them from an IT spend point of view. They don't need heavy IT to deploy AI into their workflows. The third part is continuous change management and improvement. As they want to get new use cases added, it becomes seamless and easy. And to answer your question, yes, this does give us an opportunity to monetize our software as recurring revenue.

And this does conclude the question-and-answer session of today's program. I'd like to hand the program back to Dinakar Munagala for any further remarks.

Thank you, everyone. To wrap up, Q1 was about execution, and we made real progress. We've cleared the non-cash one-time accounting items tied to the merger. So what you're seeing now reflects our actual operating performance. We've moved from pilots to live deployments in smart cities, defense, and national infrastructure. We were selected by CBIST in South Korea. We're finalizing POs with Turbo Federal, and we're advancing with our anchor customers. Our commercial traction is real. Our solutions are being deployed, and our strategy is working. Looking ahead, we'll continue to update on revenue and bookings. You've been part of our journey, and we truly appreciate your support. Thanks again. We're building momentum, and we're just getting started.

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.