Axe Compute Inc. (AGPU) Earnings Call Transcript & Summary
May 18, 2026
What were the key takeaways from Axe Compute Inc.'s May 18, 2026 earnings call?
In Q1 2026, Axe Compute Inc. reported a revenue of $35,000, a significant decline from $110,000 in Q1 2025, primarily due to reduced sales in its legacy Drug Discovery Services. The company incurred a net loss of $7.7 million, largely attributed to a $4.3 million noncash loss on digital assets. Management highlighted a landmark $260 million contract expected to generate $21 million in quarterly revenue, signaling a strong growth trajectory and a robust pipeline valued at over $4.3 billion. The outlook remains optimistic as management anticipates closing around $1 billion in transactions this year, suggesting potential for substantial future revenue growth.
What topics did Axe Compute Inc. cover?
- Landmark Contract: Axe Compute signed a historic 36-month agreement worth $260 million, which is expected to generate $21 million in quarterly revenue once deployed. CEO Chris Miglino stated, 'The deals we're making now will make the company for years to come.'
- Revenue Decline: Total revenue for Q1 2026 was $35,000, down from $110,000 in Q1 2025, reflecting challenges in the legacy Drug Discovery Services segment. CFO Jeremy Yaukey-Witter noted, 'Our Compute Services segment contributed minimal revenue in Q1 2026.'
- Future Revenue Pipeline: Management reported a pipeline of over $4.3 billion, indicating strong demand for AI compute services. CEO Miglino expressed confidence, stating, 'I believe that we close around $1 billion in transactions this year.'
- Digital Asset Losses: The company recorded a $4.3 million noncash mark-to-market loss on digital assets, impacting the net loss for the quarter. CFO Yaukey-Witter explained, 'Much of it is noncash in nature and driven by accounting rules.'
- Operational Transition: The transition to the Compute Services segment is ongoing, with a focus on building asset value. Yaukey-Witter mentioned, 'Our financial statements reflect the transitional state of our business.'
What were Axe Compute Inc.'s May 18, 2026 results?
- Total Revenue: $35,000 (vs $110,000 in Q1 2025, -68% YoY)
- Net Loss: $7.7 million (driven by noncash digital asset losses)
- Quarterly Revenue from Landmark Contract: $21 million (expected from $260 million contract)
- Pipeline Value: $4.3 billion (reflecting strong demand for AI compute services)
- Operating Loss: $3.4 million (excluding noncash items)
- Cash and Cash Equivalents: $6.9 million (down from $10.8 million at December 31, 2025)
Axe Compute is positioned for growth with a promising pipeline and a significant contract that could bolster future revenues. However, the current financials reflect transitional challenges that investors should monitor closely. Key catalysts include the successful deployment of the landmark contract and the potential closure of additional deals in the pipeline.
Earnings Call Speaker Segments
We appreciate you being here today. This is the Q1 2026 investor presentation for Axe Compute. Before we get started, please turn your attention to slide two, where our important legal disclosures and forward-looking statements can be found. Kindly take a moment to go through them. Be aware that during today's discussion, forward-looking statements will be made that are subject to the risks and uncertainties outlined in our filings. We would urge all of you to review our 10-K filed on March 31st, 2026, along with our 10-Q filed on May 15t, 2026, so you have the complete picture. Having said that, I will now turn things over to Chris Miglino, our CEO, to kick us off.
Thanks, Erin, and good morning, everyone — appreciate you being here. I’m Chris Miglino, and I serve as Axe Compute’s CEO. Joining me are Josh Blacher, who is stepping away from the Chief Financial Officer role, Jeremy Yaukey-Witter, who is taking over as CFO, and Kyle Okamoto, our President. We’re here to walk through results for Q1 2026, and we’re grateful you’ve made time to learn about the business. Many of you probably aren’t familiar with us yet, so I’ll focus on three points today: the work we do, the reason it succeeds, and the direction we’re headed. Companies have been told they simply have to live with the constraints on the computing power and data centers they can get — determined by delivery timelines for those sites, their pricing, and their locations. Regarding GPUs, the guidance amounts to: get in line and wait. The queue runs 30, 36 to 52 weeks long, and whatever geography the data center hands you is what you take. Whatever architecture that provider supports is what you get. The expectation from those suppliers is that off-takers — the firms leasing that hardware — will adapt their own operations to fit the offerings available. Those represent the compromises businesses have had no choice but to make regarding AI compute. We see things differently. We work alongside businesses so they can build according to their own requirements. Three elements make that possible. First, choice. Whatever GPU type, wherever it’s located, whatever configuration is needed — aligned with the workload that company requires. We’re seeing customers ask for every kind of GPU, not only the newest ones available in the market. There’s a wide spectrum of customers seeking hardware that’s been around for years. For the enterprise, we take care of sourcing, matching, and logistics, while our customers build around their own operations. Second, we have global reach. Our supply group is talking with data centers worldwide, and that allows us to offer inventory that can be ready very quickly, while certain other clients might prefer custom build-outs at those different facilities globally. We understand the process for obtaining the power needed — power that is on hand so off-takers can get done what they require. The schedule shown at the bottom demonstrates how fast we have executed our strategy. At the end of September, we rolled out the strategic compute reserve, which enables Axe to keep tokens for buying compute. Here’s the easiest way to understand it: in the same way consumers and teams rely on tokens to engage with OpenAI and Claude, we rely on our tokens to obtain bare metal compute. This turns into money for the company. That's why investors should view our compute reserve as though it were additional cash. Even though GAAP accounting doesn't formally classify it that way, whenever we need to, we can turn those tokens into cash, and we can do it pretty fast. After that, we rebranded and got listed as AGPU on Nasdaq in December. Then in February, I joined the company as CEO. Even though this is my first time taking on a CEO role, I had already been heavily involved in assembling this entire deal with the legacy public company, helping with the negotiations over its terms with that legacy company, and assisting with the raise connected to the initial formation of the compute reserve. I understand it well, and I had gotten myself fully current on the business. February saw us complete $12 million of compute transactions, and after that, in April, we closed a landmark $260 million deal. We've put together a fantastic team, one that sells to off-takers — those who lease the compute — plus a team selling to the data centers. Right now, I couldn't be more pleased with the team we've built. With this team, everything is going great. Let me share a bit about the big transaction we completed, since I realize we put out a press release announcing it, yet we've never discussed it. First off, this represents the largest deal the company has ever signed — and keep in mind, we didn't even begin working on it until February of this year. It gives a sense of where we think this business is headed. The agreement spans 36 months and includes 2,304 NVIDIA B300s, backed by 4.8 MW of dedicated N+1 redundant power housed in a U.S. Tier III data center. We're aiming to deploy in Q3 of 2026, and work on that is already happening right now. The way it works is the customer commits to paying monthly in advance, regardless of how much they end up using. Naturally, the data center itself provides the power and redundancy. We announced and signed this deal at the end of April. Construction is currently progressing. Equipment orders are in. Power capacity has been assigned. Hardware installation is underway. This month, we expect the initial payment, and that payment is quite large. After the cluster becomes operational, we'll record revenue from this deal at approximately $21 million each quarter. That figure is locked in. To put it another way, once installation finishes, $21 million will come in every three-month period. Should launch happen during the second month of the upcoming quarter, then we'd book two-thirds of that amount—but going forward, it's $21 million per quarter across the following 36 months. In our view, this approach is what will generate value for Axe, along with our cash flow and equity value. Allow me to walk you through how that functions. Let me explain the reason enterprises pick Axe as their partner. First, the customer tells us how large their workload is and which regions they need it in — that information comes to us from them. From there, Axe handles three things. We line up the appropriate GPUs, the appropriate data center site, plus everything that supports it: networking, power, racks, the NVIDIA reference architecture, and more. Then we assist in structuring the financing for the build-out. Sometimes SPVs get set up for this, while in other deals we take a full stake in both the equity and the debt sides. Payments come from customers monthly and in advance; an initial prepayment of 15%-30% is typical, and whatever remains gets billed each month upfront. When all is said and done, the hardware purchased with our equity is owned by Axe. On our balance sheet, that equipment shows up as a hard asset. Once a customer's contract period concludes, those very machines can be redeployed for the following customer. That capital can be utilized by us to fund the equipment. Think of it in the same way institutional investors view data center real estate buildings. Those get amortized over 30 years by them; when it comes to GPU clusters, we follow the same approach over three, and sometimes even five. Every deal we close contributes asset value to the balance sheet, and not just cash flow. Even though money is coming in from these deals, it's going toward reducing the loans tied to the hardware we put in place for the cluster buildouts. As we grow, you could think of us as a data center operator without any buildings of its own, but with servers sitting all over the world; and what we've seen is that appetite for NVIDIA processors that came out three years back is still going strong. In fact, some of the older processors are fetching higher rental rates right now than they did two years ago. Once the next three years of amortization are done, those processors should be worth a lot, based on what we're projecting. With that out of the way, I'm going to hand things over to Josh Blacher, who has been our CFO and is now leaving that role. Josh has been really important to where this company has gotten. I'd like to say thank you to Josh, who has been serving as a fractional CFO here for a good while. We've now gotten to the point as a company where having a full-time, dedicated CFO makes sense for us. We're going to miss Josh, and we appreciate everything he's done, but we're also very excited to have Jeremy joining us as the new CFO. Josh.
Thanks, Chris. It’s been a privilege to work for this company as Chief Financial Officer starting in September 2023. As we shared earlier, this marks the final earnings call I’ll take part in as I step away from the position. I’m glad to bring in the person taking my place, Jeremy Yaukey-Witter. Jeremy, congrats on stepping into this new job. This company is in great hands. Jeremy will take you through the numbers for our first quarter, which closed on March 31st, 2026. Jeremy.
Thank you. Thank you, Josh. I want to begin by expressing my gratitude to Josh for everything he has contributed to this company. The groundwork he helped put in place is something we value, and I'm eager to continue building upon it. For all of you joining us today, I'm happy to be here with you. My time with this company began 3 years ago, and before that I held the role of controller here. KPMG is where my professional foundation was formed, and there I delivered audit and attestation work for both public and private businesses spanning a range of sectors, among them technology and energy. I have been present through every phase of our strategic transition, and I'm genuinely enthusiastic about where this company is going. Before I walk through the figures for our first quarter, which ended March 31st, 2026, I'd like to give context around what you're seeing in our filings, since our financial statements for that first quarter capture the transitional condition of our business at that point in time, with 2 distinct segments sitting at very different phases; they also include certain non-cash items that I want to be certain are plainly understood. Turning to our income statement, the net loss we reported for Q1, 2026 came to $7.7 million. Let me tackle that number head-on. A big chunk of it isn't cash-related at all—it comes from accounting standards that require us to revalue our digital asset positions every quarter. Specifically, we recorded a $4.3 million loss on digital assets this quarter. U.S. GAAP requires our Aethir token holdings to be reported at fair value on each reporting date, with any movement running through the income statement. During the first quarter, the Aethir token's price fell, producing a $4.3 million non-cash mark-to-market loss on the tokens we hold. Excluding the fair value adjustment tied to digital assets, we posted an underlying operating loss of roughly $3.4 million. That figure mirrors the cost structure we're carrying as the business continues its shift toward concentrating on our compute services segment. Revenue for Q1 2026 totaled $35,000, down from $110,000 in Q1 2025. This decline compared with the prior year stems from lower sales in our legacy drug discovery services business, which is still reported in continuing operations while we weigh strategic alternatives. Revenue from our compute services segment was minimal in Q1 2026, coming from only a small number of compute contracts that began at the close of March. Consistent with U.S. GAAP, we recognized revenue on those compute contracts on a ratable basis across the service period. Customer prepayments we've received before recognizing the related revenue—what we report as contract liabilities—went from $144,000 on December 31, 2025, up to $786,000 on March 31, 2026. Roughly $650,000 of that amount came from compute services, and it's contracted revenue we'll recognize as those services are delivered in later periods. Operating costs and expenses overall came to $3.5 million in Q1 2026, versus $2.4 million in Q1 2025. The Q1 2026 figure was mostly attributable to general and administrative expenses of $2.9 million, which is about $1.1 million higher than Q1 2025. What drove that was mainly a one-time severance charge for our former CEO following his departure in February, together with the board appointing Chris as CEO, plus other expenses tied to personnel. Looking at it per share, net loss came to $0.36 per share, calculated on a weighted average share count of roughly 21.2 million shares. Under U.S. GAAP rules, that weighted average figure counted the 14.7 million pre-funded warrants still outstanding as of March 31, 2026, which is lower than the 16.8 million pre-funded warrants initially issued through our October 2025 private placements. Let's move on now to cash flows and the balance sheet. Cash and cash equivalents for the company fell by about $3.9 million over the quarter, going from $10.8 million on December 31, 2025, to $6.9 million on March 31st, 2026. That drop mirrors what we spent running the business during the quarter. Operating activities used $3.7 million of cash in Q1 2026, versus roughly $1 million in Q1 2025. The rise mostly comes from more cash going into working capital along with higher cash operating expenses. The working capital cash use largely involved paying down outstanding accounts payable and accrued expenses. Higher cash operating expenses mainly came from payments for extra professional services tied to the company adopting its treasury strategy in late 2025. Cash used in investing and financing activities was not significant in the first quarter, as the company concentrated on transitioning its operations. Fair value of our digital asset holdings stood at $20.2 million as of March 31st, 2026. That's against $24.4 million on December 31st, 2025. That decline is driven primarily by the mark-to-market adjustment referenced earlier, and it ties straight to how the Aethir token's market price moved lower over Q1. As of March 31st, 2026, the fair value of our digital asset receivable stood at $15.4 million — that figure reflects our contractual entitlement to receive further Aethir tokens later on, subject to time-based vesting conditions (these are what we call locked Aethir tokens), and those locked tokens follow a predictable vesting schedule running through December 2028. Breaking down that total, $9.4 million fell into the current category, which means those tokens should vest and be claimed during the coming 12 months, whereas $5.9 million was placed in the non-current category. I also want to call attention once more to two items on the balance sheet: accounts receivable and contract liability. Both of them, as you can see, went up by more than $600,000 from December 31st to March 31st. That growth comes from invoices we issued for monthly prepayments that our compute customers owe and that cannot be cancelled or refunded. Those invoices demonstrate the momentum we built late in the first quarter as we began to get customer contracts signed. As of March 31st, 2026, total assets stood at $45.2 million, versus $52.9 million at December 31st, 2025. We ended with total liabilities of $5 million and total stockholders' equity of $40.3 million, which is lower than the $47.7 million recorded at the close of the previous year; the primary reason for that shift was the $7.7 million net loss we posted for the quarter. I'll now hand things over to our president, Kyle Okamoto, so he can walk us through what we're doing on the business side.
Appreciate that, Jeremy. It's been 6 weeks in this role, and I can honestly say my enthusiasm for this market and what lies ahead for this company has only grown. There are 3 topics I want to touch on today. First, the way our business model has changed over time to serve our customers. Second, a deeper look at how the Axe Build delivery model works. And third, our pipeline — the figures there are strong, to put it mildly, and I believe they paint a very persuasive picture of where this business is headed. Okay. Starting with how customers grow alongside Axe. There are really 2 tracks when it comes to working with customers. With the access available now compute, we'll stand up a dedicated cluster. So, path one and path two. The majority of customers come in through Axe Compute's immediate access inventory; it's quick. They take advantage of our existing compute network. Getting started can take only 24 to 48 hours. These clients turn to Axe Compute because their need is urgent, and what they require isn't available from traditional hyperscalers or even other neo clouds. The triggers are very consistent. It isn't only wait lists of 36 to 52 weeks. Beyond that, this is inventory you simply won't find anywhere else, along with the sites those clients need to meet their own customer standards — things like keeping data within national borders and getting the inference speed they're after — and of course, clear pricing that anybody buying in this market ought to get, plus the standard of service they've got coming to them in the end. As for Track 2, that's the build program Chris walked through, tied to the $260 million landmark deal; the clients here are after sizable, dedicated infrastructure capacity in a set geographic area. What they want is one partner to handle the entire build-out, covering both the engineering and the financing. Clients turn to us because of what we know and our capacity to find the space and power, arrange financing, secure chips and other hardware, and manage everything from construction through ongoing operations. Typically these arrangements last either three or five years, with keeping costs down as a key goal and, of course, enterprise-grade SLA demands throughout. Across both of these paths, the things customers care about are identical, right? They're looking for options. Whatever they need, they want it delivered on their timeline and in their preferred manner. What they require is know-how, assistance, and above all worldwide coverage, since today's world is extremely flat—you must be present wherever your customers are, you have to enable data sovereignty, and in the end your performance requirements need to be exceptionally demanding. The fourth point is essentially our service. Our mission is to guarantee that regardless of how clients choose to interact, they never need to look elsewhere. If they stay on track one or move over to track two, the experience, the support, and the relationship stay seamless. So now, let me take you through what delivering a large cluster—such as the $260 million deal—actually involves for Axe Compute. A question that often comes up is this: for a build this big, what makes a company go with Axe Compute? The answer really comes down to us being more than a vendor — we're a full infrastructure partner. So let me take you through those 6 steps and explain what that means in practice. When it comes to architecture, everything we design is shaped by the customer's biggest priorities. That could mean selecting the GPUs, the CPUs, the onboard storage, high-speed shared storage, the various network fabric choices, private and public networking, power and cooling, how racks are laid out, ancillary services, and take Kubernetes or Slurm as an example. Every bit of it serves one goal: building a true AI factory for our clients. Our clients communicate their requirements to us, and then we determine the best way to accomplish the task. Financing and procurement make up step two. The build gets funded by Axe; we obtain the hardware, whether that's NVIDIA B300s, Grace Blackwell GB300s, or anything else the job calls for. We take care of aligning the lead times across different components, and we do it all transparently so customers have the information they're looking for at every stage. There's no need for the customer to supply their own CapEx—we're offering this through an OpEx structure. Naturally, we lock in data center space and power in the appropriate location, ensuring we hold right of first refusal for customer expansions, and ensuring we deliver N+1 redundancy to meet enterprise-grade SLAs or better. Everything lines up with what’s needed for data sovereignty and performance, honestly. Naturally, we have to build it, deliver it, and support it, and that work is underway right now. We handle installation, testing, turnover, response-time commitments, resolution-time commitments, SLAs, ticketing-system integrations, and so on. Naturally, by relying on NVIDIA reference architecture, we ensure the client is deeply engaged in that effort right from the start. Axe Compute also stands out because we provide an upgrade path — during the contract term, customers can shift to the newest GPU architecture. We have no interest in trapping them on an older GPU generation and preventing a move to the most advanced option, so they can stay aligned with market innovation and growth. And naturally, Axe Compute is behind that expansion. Ultimately, the hardware ends up belonging to Axe Compute, correct? Once an Axe Build wraps up, those machines sit on our books as a tangible asset. It's ours—we have the ability to move it elsewhere, and we can offer GPUs as a service to various customers. Each transaction grows that asset pool, and that final idea is genuinely what makes this model appealing past simply generating positive cash flows, right? It can be repeated, and it accumulates. Now let me get into the pipeline. Before I bring up the pipeline—oh, there it is. Let me establish the market backdrop here, all right? Since those figures are important. The marketplace we compete in, based on figures from firms such as McKinsey and Gartner, has reportedly already passed the $1 trillion mark for AI compute as of 2026. Yes, that's trillion with a T. Through 2034, it's expanding at a compound annual growth rate above 30%. McKinsey puts anticipated total data center capital expenditures at more than $6.7 trillion across the period stretching to 2030. There's no question the demand is there, and nothing suggests it's easing off. Actually, the opposite is true. So after just a handful of months in operation — with this business, as Chris mentioned, truly kicking off earlier in the year — here's what that demand looks like in our pipeline. And naturally, following the close of our first big Axe Build deal, which happened during my initial three weeks with the company, that certainly gives the pipeline a significant boost, correct? Let me be clear about something: a pipeline is just a pipeline, correct? What we're talking about are qualified prospects, not executed contracts. Some of them won't convert; we're sharing this because we see it as important context for grasping the opportunity ahead of us, not as a projection or pledge regarding any particular revenue figure, correct? This represents a sample of, I believe, 45 prospects at this more advanced pipeline stage; together they account for more than 36,000 GPUs. Looking at the breakdown of those GPUs, about 72% — close to 26,000 units — consist of Blackwell orders. That covers the Blackwell family: Grace Blackwell, B200s, B300s, GB200s, GB300s, and similar models. What this makes clear is that customers want the strongest computing power on the market right now for their most advanced applications and breakthroughs, which ultimately reshape and touch every aspect of the globe. Taken together, these qualified deals represent a total contract value exceeding $4.3 billion. The most telling detail is what the mix of commitment durations reveals: the bulk of these agreements cluster around 36-month and 60-month terms. Our customers understand that securing AI infrastructure is now mission-critical for their day-to-day operations. They want certainty around compute availability, so they're willing to commit in advance rather than rely on short-lived, spot-style capacity that lasts only hours or days. What we're seeing is enterprises treating this as a long-term infrastructure decision, and they're looking for a true partner in that effort. The bottom line is this: if we finalize just three or four such agreements within the year, that translates into hundreds of millions of dollars in incremental contracted revenue, building on the $260 million we've already booked. We're thrilled about that, and it's absolutely one of our top priorities. With that, I'll hand it over to Chris for some closing comments.
Thank you, Kyle. Kyle and his crew have been performing incredibly well. We are laying the groundwork for the company's success over the long haul. We find ourselves in the ideal spot at the ideal moment — more than anything else I've witnessed during my entire professional life. The agreements we are putting together right now are going to define this business for many years ahead. In my view, the deals that get finalized during this year are going to deliver revenue that is significant across the coming 3 to 5 years. We see no indication of that easing up whatsoever. Our strategy involves growing very quickly. We aren't the data center itself, but rather the enterprise that constructs and possesses everything located within the data center. I realize we've gone over this many times, yet that initial deal we completed unlocked a flood of possibilities for us; beyond the $21 million each quarter that this specific transaction delivers, it has generated a pipeline exceeding $4.3 billion, in case you didn't catch that when Kyle mentioned it before. I've stated openly that I expect us to finalize roughly $1 billion worth of transactions during this year. Looking at the pipeline figures the team has, I have great confidence that this outcome will materialize. Off-takers — or those who lease the equipment under our model — want to collaborate with us. The crucial element is accumulating asset value. That's what counts most for investors evaluating long-term worth. Both of our tracks contribute significant revenue-generating assets to our balance sheet; through Axe Build, each cluster we bring online adds AI infrastructure we own to our balance sheets. These are assets that depreciate across the contract period and get redeployed to subsequent customers. We aren't merely producing cash flow — we're constructing an enterprise with a solid base for what lies ahead. Having said that, I'd like to express gratitude for your time, your attention, and your interest in Axe Compute. Clearly, the figures shown in this initial quarter don't truly represent the underlying business, and you'll begin noticing that take effect in Q2 and Q3 of this year, once the framework we've established genuinely begins showing up in the income statement, cash flow, and balance sheet. Thank you for joining us today, and we value the time you've given.
For developers and AI pipelines
Programmatic access to Axe Compute Inc. earnings transcripts and 132,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.