Axe Compute Inc. ($AGPU)

Good morning, everyone, and thank you for joining us today for Axe Compute's Fiscal Year 2025 Investor Presentation. Our common stock trades on NASDAQ under ticker symbol AGPU. On today's call, you'll hear Chris Miglino, our Chief Executive Officer; Kyle Okamoto, our President; and me, Josh Blacher, our Chief Financial Officer. Before we begin, I'd like to direct everyone to Slide 2, which contains important legal disclosures. Please take a moment to review those. With that, let me hand the call over to Chris.

How are you? I'm excited to start as the new CEO of Axe just over 30 days ago. What we've accomplished in the last 30 days is quite remarkable. I'll get to that in a little bit. In September of 2025, we launched our strategic compute reserve, our GPU capacity platform. By December, we initiated a strategic expansion into AI infrastructure and rebranded as ax compute on NASDAQ. In February of this year, I stepped into the CEO role and just this month effective March 26, we added 2 exceptional industry leaders to our board, Theodor Zoo and Thurston Dirks. Theodor, Zoo is a semiconductor pioneer with over 85 patents in the United States, the founder of IoTelligent Technology and former President CEO of CELESTIAL Semiconductor. Thurston Dirks is 1 of Europe's most accomplished technology executives with nearly 2 decades of Board experience and former CEO roles at Telefonica Dutch land, Lufthansa, Eurowings, ePlus Groups and Dushlandglass pacer. Together, Zu and Dirk's bring more than 4 decades of combined experience spanning semiconductor innovation, large-scale telecommunication car information, international M&A and enterprise digitization. I'd also like to announce today that Kyle Arcimoto is starting as the President of the company, and we're very excited to have Kyle join us. We're now entering the execution phase of the company, which we will be which will demonstrate how our vision of being a leader in the compute market will take shape. I also briefly want to address our drug Discovery Services legacy segment. Strategic often are under evaluation, and we'll share more of that process as it develops. Now let me expand on our vision. Our mission is simple. We're bringing choice to the market. GPU compute has become 1 of the most precious commodities in the global economy and yet access to it has been gated slow expensive and concentrated in the hands of a few hyperscalers with wait lists that stretch 36 to 52 weeks. We're changing that. AX compute gives enterprises 3 things that they can't get from the incumbents. First, we have choice, any GPU, any location, any configuration match to your specific workload, you're not locked into a single provider's hardware catalog. Second is speed we can develop in as fast as 24 to 48 hours, subject availability, not weeks, not months, but days. That means you can innovate, experiment and pivot in real time. Third, distribution with 200-plus global locations across our network, we can meet enterprises where they operate and where their customers are, respecting data sovereignty requirements along the way. This is a platform we're building and the market is just enormous. I want to direct your attention to a significant change. We've executed approximately $12 million in total contract value over the last month or so, subject to deployment and performance terms. We're entering Q2 with an expected monthly contract value of approximately $850,000 as deployments come online that translates to $7.7 million in contracted 2026 income signed at this time. And we now have more than 20 enterprise customers with over 30 active deployments live. So I just want to reiterate that, that we've signed $12 million worth of contracts for the business. And we're continuing to sign more, but we're off to a great start. I also want to speak to the quality of this income because not all income is created equal. Every contract on our books is structured with partial payments in advance customers commit to reserve capacity on a monthly basis. The structure is designed to reduce receivable risk, many contracts, including advanced payment components prior to compete delivery. These contracts are structured to support recurring enterprise income across all native companies and established enterprise verticals, covering a diverse hardware mix, including H 100s, H200, B-20 RX590 and RX590 clusters. The breadth of GPU coverage reflects the breadth of enterprise AI use cases we're serving. As you know, we've launched the compute reserve, the strategic compete reserve. The reserve provides us with tokens that can be used to purchase compute on an ongoing basis. What you see on this slide is the value of that compute reserve. As this presentation, current market cap of the company is approximately $8.6 million. The value of the strategic compute reserve is $43 million. This is based on March 27, 2026 pricing, the timing of the preparation of this presentation. The token that makes up our compute reserve is the AI infrastructure token ATH. Note that ATH token prices are naturally volatile as are many cryptos. The value of as of March 27 reflects a point in time price, and that price will move we just want you to make sure that you're aware that there's a lot of volatility there. The MNAV ratio is a multiple of the market cap. Currently, it is 0.20. And on a fully diluted basis, it is 0.79. At the current levels, the public market implies a valuation below the reserve value. before assessing any value for the operating business or the contracted income that we just discussed on the previous slide. Keep in mind, NAV is a non-GAAP metric and should be considered a substitute should not be considered a substitute for GAAP financial measures refer to AX Computes 10-K further details on any discounted valuations. On the compute reserve, we currently have 5.9 billion tokens, which is made up of $2.86 billion unlocked liquid tokens and an additional 3.12 billion tokens that are locked with a monthly unlocked cadence running through December of 2028 we're not telling investors what the discount should be. We're presenting the numbers and letting the market form its own view. In our opinion, the operating business and early commercial traction you saw on the prior slide is not reflected in the $8.6 million market cap, and we believe execution over time may help close this gap. Investors should keep in mind that AX has the ability to convert these unlocked tokens into cash through the purchase of compute that we then resell. So let me talk to you a little bit about what that business model looks like. The commercial traction you just size is real. The slide explains why it's structurally durable. Our model has 4 components, and they work together in a flywheel, not as independent features. Reserve GP capacity customers commit to reserve clusters priced per GPU per hour with monthly payments in cash in advance and usually with prepayments on long-term contracts. That eliminates pricing volatility for the customer and receivable risk for us. We get paid before computers provided. Disputed provider network 200-plus global locations and many providers, customers get location choice and data sovereignty, your AI infrastructure isn't constrained by where we've built, we meet you where you operate. OpEx models for customers, AX can finance and build dedicated infrastructure, including B-300 clusters. So customers get the capacity they need without carrying CapEx on their balance sheet. In an environment where every CFO is scrutinizing capital commitments, that is a meaningful commercial advantage strategic compute reserve. As we utilize the capital from these stakes to purchase more ATH, our strategic compute reserve, we may receive up to a 20% token-based incentive depending on network mechanics. The model is designed to grow the reserve over time. The reserve expands capacity, expanded capacity enables the next deal. So customers pass in cash. We used the cash to purchase ATH tokens, and we use that then to buy compute. We may receive additional ATH tokens currently targeted at up to 20% and depending on network mechanics that can be used to buy more compute in the future. So I just want to make sure that everybody is clear, the inside our business, the amount of money that you saw on the earlier slide that we have in the strategic compute reserve can be turned into cash through selling that utilizing those tokens to purchase compute and then reselling that compute to our customers, thus generating cash for the company. So now I'd like to turn now I'd like to turn the call over to Josh Blacher, our CFO, who will give an overview of the financials for 2025. Josh?

Thank you, Chris. Before I walk through the financials, I want to address something important upfront. Our fiscal year 2025 income statement includes 2 large noncash items that significantly affect the reported net BOSS figure and are not reflective of our underlying operational performance. I'll explain both clearly and then walk through the underlying results. First, we recorded $152.5 million in losses on digital assets at fair value. Under ASC 350-60, which we adopted effective January 1, 2025, and digital assets are carried at fair value at each reporting date with that mark-to-market flowing through the income statement the ATH tokens we acquired through December 31, 2025, at a cost basis of $102.7 million declined in market value to $24.4 million at December 31, and 2025, a $78.3 million unrealized decline on those holdings, which is noncash. The tokens remain on our balance sheet. Second, we recorded $52.7 million in losses on derivative instruments. This relates to the Crypto pipe transaction executed on September 29, 2025 because the value exchange involved ATH token rather than a fixed dollar amount, U.S. GAAP required us to treat the contract as a derivative liability at fair value through earnings evaluating was performed to measure the change in fair value of the instrument between execution on September 29, 2025, and on settlement on October 7, 2025. When the deal closed on October 7, the derivative liability was derecognized and the ACH tokens were recorded as assets while the prefunded warrants as equity. The $52.7 million reflects the fair value movement between September 29 and the settlement on October 7, again, entirely noncash. Please keep in mind that Together, these 2 items account for $205.2 million of our $233.1 million net loss. Full year 2025 income statement highlights. Revenue from continued operations for the fiscal year ended December 31, 2025, was $125,000 compared to $85,000 in fiscal year 2024 this revenue is entirely attributable to our legacy Drug Discovery Services business segment. Our compute services segment did not generate revenue in 2025 as our go-to-market pipeline was still being developed during the company's strategic transition we expect compute services to be a meaningful contributor beginning in fiscal year 2026. The Total operating costs and expenses for 2025 were $28.6 million compared to $10.4 million in fiscal year 2024 and on a per share basis, the net loss was 13.37% based on a weighted average diluted share count of approximately 17.4 million shares the weighted average share count reflects a significant share and prefunded warrant issuances associated with our pipe transaction during the fourth quarter of 2025. Balance sheet highlights. Turning to the balance sheet. We ended fiscal year 2025 with $10.8 million in cash and cash equivalents compared to $0.6 million at December 31, 2024, and this is driven by the proceeds of our pipe financing transaction in October 2025. Our balance sheet now includes 2 new asset categories that did not exist a year ago. First, digital assets of $24.4 million, representing our $2.8 billion ATH tokens on hand. Second, the digital asset receivable of $15.5 million, representing our contractual right to receive additional HTH tokens in the future pursuant to time-based vesting conditions also referred to as locked ATH tokens. Those tokens are expected to unlock on a predictable schedule through December 2028. And Total assets as of December 31, 2025, were $52.9 million compared to $5.0 million for the prior year. Stockholders' equity was $47.7 million at year-end compared to a deficit of $0.2 million at December 31, 2024, and the restoration of equity to positive territory and are regaining of compliance with Nasdaq stockholders' equity requirement in December 2025 were significant milestones for the company. This concludes my financial review. I'd now like to turn the call over to Kyle Okamoto, our President, to discuss the company's out strategy in a near-term pipeline. Kyle.

All right. Thanks, Josh. Appreciate that. fired up this morning. Good morning, everyone. I'm absolutely thrilled to be here to walk you through how our business actually works and why the market dynamics are so favorable from what we're building. So let's start with the mark. I would say this in brief, we are in an AI supermarket, right, super cycle of AI. In 2026 alone, Gartner projects AI spending will reach $2.5 trillion with the team. AI data center spend is expected to grow at over 31% annually by 2030. And in 2026 alone, global AI infrastructure CapEx exceeds $1 trillion, up from just $290 billion 2 years ago. That's massive growth and a cumulative projected spend of $6.7 trillion by 2030, sourced by a bunch of reputable research organizations, and I think everybody overall in this space that spends day to day, we'll think that those numbers are just far too low. The structural shift, though, is what matters most for aXcompute. AI workloads grow from 25% to 70% of all global data center activity by 2030. That's tripling the share of workloads within data centers in just 4 years. That's not just more AI. That's a fundamental shift and recomposition of what compute infrastructure is really for. I would say 4 things are driving this paradigm shift. One, hyperscaler CapEx is running at $600 billion combined in during 2026. 7 gigawatts of new capacity are coming online through 2030 three, inferencing is now overtaking training as the dominant AI workload by 2027. And meaning distributed latency-sensitive compute that cannot just run in 3 U.S. regions has to be where people are, where applications live or agents live, et cetera. And lastly, enterprises are actively moving towards bare metal performance as they get smarter and more capable of using GPU-based compute and away from hyperscaler abstraction. We AXCompute our position directly in the path of this shift. Let me walk you through how we actually serve customers. When you put a compute side by side against the public Neocloud segment, the contrast is pretty strike. And frankly, the investment case becomes very clear. Our distribution model, as Chris mentioned, spans over 200-plus locations over 100 different configurations. That's choice. Our competitors are largely operating in a single country or maybe a few data centers in limited regions, usually with a single configuration step. So AXCompute has anchored around providing choice in a highly centralized and consolidated market. We also target deployment time lines this size of 24 to 48 hours, or sometimes faster, of course, subject to availability. This is versus weeks and months, like Chris mentioned, for the rest of the market, right, the overall market. We're primarily leveraging an OpEx model, and that allows for a very predictable economic scaling, but ultimately less drag. We don't have to spend 18 months building out a data center. We work with partners that are building out data centers. Our peers require heavy CapEx infrastructure investments to keep up with sales. We can meet that demand now across a very wide note. We're designed from the start to deliver 100% bare metal performance. in Tier 3 and Tier 4 data centers backed by enterprise-grade SLAs designed really to support those high-performance workloads. Our peers obviously offer a mixed bag of virtualized environmental solutions. But ultimately, we're built from the ground up for this high-performance enterprise-grade workload. And then look at valuation. Our basic market cap, as Chris mentioned and Josh mentioned, $8.6 million. Neocloud peers range from hundreds of millions to more than $38 billion a couple of days ago. Our peers are trading at 7 to 15x revenue PDAs. We're trading at roughly 1x expected income from signed contracts, 1 versus 7% to 15% against the $52.9 million audited asset base. The bottom line here as the slide says, the market is not yet pricing aXcomputes and neocloud with real income and real assets. We believe that changes as we continue to simply execute. So let me walk you through the pipeline and some deal economics. One of the questions we get from investors is how predictable are income streams. Let me walk you through an example of that to highlight the more detailed mechanics of our business. For example, a 10-node H-100 cluster on the 12-month commitment generates around $1.3 million in ARO. Gross margin on a deal like that runs around 8%, maybe a little higher customers prepay in advance 10% of the total contract value at sign. That shows a commitment to stick around for the full 12 months and put money on the table. And they also pay the first month's rent in advance. -- which means we have cash in hand before we tie up compute resources. Each settlement then may generate additional ATH token incentives currently targeted at that up to 20% level by Christmas, which then compounds the strategic compute reserve so that our COGS each month for the next month that we book that compute get 20% cheaper, and that compounds over and over. So income is priced at a per hour per GPU rate on a reserve capacity basis. So clients get a consistent bill each month with no hidden fees. We've built acts around what's wrong with the centralized cloud world today. And 1 of those key tenets is you get a surprise bill at the end of the month and you don't know where your spending is, you don't know what you're paying for egress or hidden storage charges or API calls or service tickets or things like that. So no hidden fees. Contracts are typically 6 to 12 months or longer and always was monthly advanced payment. The structure is really designed to reduce receivables exposure but also to give customers predictability in their underlying economics and align with their monetization model. And of course, we also have ancillary upsell services like shared storage, things like VAST or WECA, CPU nodes when customers have larger clusters for orchestration or things like managed Kubernetes that is simply out of the box on a software and API basis. These things all can bolt on to a deployment ultimately driving up margins and stickiness with clients. What this creates is really a business where contracts are structured to include upfront payment components prior to deployment, contracts are long term. Contracts are recurring, and every deal simultaneously grows both our P&L and our token reserves. And so that's that compounding model. That's that flywheel that Chris was talking about. It's not a linear model. And so let me shift to how customers actually get started with X and how they can grow over time. So we basically serve customers through 2 main approaches, and there's a pathway between them that's frictionless and intentional. Tier 1 is what we call start fast. It's designed for speed. It's designed for experimentation and ultimately, it provides massive choice. Clients can get running in less than a day, 24 to 48 hours on average across any GPU type, any configuration, all 200-plus locations with flexible term lengths, full parametal performance from day 1. This is the lowest friction on ramp in the market. They can see all of our selections, what's available, where is it, what the network and the configs and all the little details that they need you care about most and in most cases, they can easily try that out before they buy. Tier 2 is what we call X Forge, and it is for enterprises who want to scale very strategically with a larger cluster. And here, we have either our partners or AX ourselves, can cover the CapEx build costs. So customers can still stay on that OpEx model even at significant scale. We're talking thousands of high-end B-300 GPUs, for example. We use NVIDIA reference architecture design. We can offer service level commitments if we deploy late to stand behind our commitments to deliver that cluster on time and at the performance levels that, that client expects. And we also allow clients to upgrade during the committed term as technology evolves to the latest and greatest sheep use. The whole point is to structure a long-term partnership, not a vendor transaction with Rigid lock-in allowing customers to grow and to excel and to exceed over time. Customers can also move from Tier 1 to Tier 2 as their needs to grow, right, with no disruption, no penalty, no lock in. That's the whole intention of this model. So in short, we want to grow with our customers, not trap them. I hope that was helpful. And with that, I'll hand it back to Chris to close the sale.

Thank you, Kyle. We're really excited to have you join the company. and we look forward to you bring in your expertise in this area to the AX compute family. I want to cover a couple of things in closing it's a this is a massive market. It is growing faster than anything I've ever seen in my entire life, reminds me of when the Internet started except the demand is persistent. And I think that we are just at the forefront of companies getting into this space. I think that a lot of people have embraced AI, but I still think that there is a world of companies out there that are just starting to get into it. I think that we're well positioned to take care of those people as they come online throughout the world and do it quickly. We have in the last 30 days, signed $12 million worth of contracts. That's a lot of money for a 30-day run, and we're just getting started there. That's that brings us around $835,000 on a monthly contract value, and that's over 20 different customers. And I could tell you that there's a lot more that's in the pipeline that we're seeing out there. So we're bullish on what we're seeing here. The idea that we can continue to use the tokens that we have, to go out into the market and turn that into compute, which then turns into cash, I think, is something that investors really need to understand because this is this provides us with the ability to not have to go to market to raise additional capital in addition to the cash that we have on hand, we also have the strategic compute reserve that can also turn into cash by turning that into compute. So as you become more familiar with our business, I think you'll start to understand the true value of that compute reserve and how we can gain additional tokens from participating in buying more of the ATH tokens over time so I think if anybody is interested in setting up a one-on-one call with us, we would welcome having the conversation you can reach out at [email protected]. And I'd like to thank the investors that have come along with us on this journey. And if you do have any questions, please let us know, hopefully, today's presentation gave you a little bit more insight into where we are and to where we're going. Thank you for joining us today.

Goodbye.