Home / Transcripts / Digi Power X Inc. (DGXX) · May 15, 2026

Digi Power X Inc. (DGXX) Earnings Call Transcript & Summary

May 15, 2026

NASDAQ US Information Technology Software earnings 24 min

What were the key takeaways from Digi Power X Inc.'s May 15, 2026 earnings call?

In Q1 2026, Digi Power X Inc. (DGXX:US) reported a revenue of $6.8 million, reflecting a strategic transition away from cryptocurrency mining towards AI computing infrastructure. The company achieved a positive adjusted EBITDA of $1.1 million, a significant improvement from a negative $1.3 million in Q1 2025. Management emphasized a strong balance sheet with $125 million in cash and no long-term debt, positioning the company for future growth and expansion in AI services, with a target of reaching 9-figure annual revenues.

What topics did Digi Power X Inc. cover?

What were Digi Power X Inc.'s May 15, 2026 results?

Digi Power X's transition to AI infrastructure and strong financial position present a compelling investment thesis. The company's ability to generate AI revenues and secure significant contracts positions it well for future growth. Investors should monitor the execution of expansion plans and the competitive landscape in the AI sector as potential catalysts and risks.

Earnings Call Speaker Segments

Michel Amar executive
#1

Hello, everybody, and thanks for being with us today. Our results for the first quarter of 2026 highlight a year of transformation, one in which we reinforced our balance sheet, began winding down our crypto mining operations, and established ourselves as a capital-light infrastructure platform for AI computing at scale, with a clear route to annual revenues in the nine-figure range. Financial highlights for the first quarter of 2026. The three months that ended March 31, 2026. Adjusted EBITDA came in positive at $1.1 million, versus negative $1.3 million a year ago, in Q1 2025. Revenue reached $6.8 million, which reflects the deliberate winding down of legacy operations as we shift toward AI compute and colocation revenues. Balance sheet and liquidity as of March 31, 2026. Cash and cash equivalents totaled $71.4 million. Working capital stood at $67.2 million, compared with negative $0.8 million as of March 31, 2025. Digital asset holdings were $13.6 million, an increase of 208% year-over-year. Net fixed assets came to $26 million, up 29% year-over-year, reflecting capitalized investment at our Columbiana, Alabama facility. We carry no long-term debt. Balance sheet and liquidity as of May 15, 2026. Roughly just under $25 million in cash and cash equivalents, about $15 million in digital assets, and approximately $45 million in year-to-date capital expenditure directed toward GPU equipment and data center build-out, mainly at the Columbiana, Alabama facility. Operational highlights. Columbiana, Alabama AI Campus. The company aims to have phase 1 ready for service in December of 2026, with completion in Q1, 2027. NeoCloudz GPU as a service. NeoCloudz, our GPU cloud business, will book its first revenue in May 2026 from its initial fleet of NVIDIA B200, B300 GPUs deployed at the Alabama facility. Closing message. Q1 2026 represents an inflection point for DigiPower X. Adjusted EBITDA went positive even as we intentionally ramped down legacy revenue to clear space for a far larger AI compute business. Our balance sheet has never been stronger. As of today, we hold approximately $125 million in cash and $15 million in digital assets, with no long-term debt. We have already deployed roughly $45 million of CapEx year-to-date into GPUs and infrastructure at Columbiana. That gives us the firepower to execute phase one and the operational platform that comes after. Thanks for being with us today; we will now take questions. Good morning, everyone. As mentioned earlier, we actually finished with $73 million in cash on March 31st, with adjusted EBITDA positive at $1.1 million, versus negative a year ago. Adjusted EBITDA is essentially the net loss with all the non-cash items deducted. It demonstrates disciplined capital management and gives us sufficient cash to carry out our phase 1 Columbiana commissioning. It also lets us use our balance sheet to secure better debt financing down the road. As of March 31st, 2026, we continue to hold a zero debt capital structure. As of May 15, we still have that zero debt. We're talking with a number of lenders as a way to reduce dilution; our balance sheets are the strongest they've ever been in the company's history, with cash and cash equivalents of $125 million plus roughly $15 million in unencumbered digital assets; that adds up to essentially $140 million in cash equivalents, along with digital assets we can use as leverage to secure comfortable debt financing for some upcoming data center builds, so we avoid tapping equity dilution. That was the best move we ever made. Shifting from Bitcoin mining into AI infrastructure was the single most consequential strategic choice in the company's history; in 2025, 2 realities came together. First, Bitcoin mining economics had grown more and more compressed and cyclical, while demand for AI compute infrastructure was moving into a generational growth phase powered by frontier model training and inference scaling. Above all, we've devoted years to building a portfolio of power-rich sites in Alabama, Niagara Falls, North Carolina, Buffalo, and those sites were uniquely suited to hyperscale AI workloads. Accumulating these assets, which the company fully owns and which are extremely, extremely valuable today, took us roughly 10 years, in a twofold way. 1, speed to market. There's no need to wait on an interconnection with the utility; we, we own the substation. We have utility interconnections; we also produce power via our combined cycle gas power plant that we bought in 2022. That puts us in an extremely, extremely strong position when it comes to speed to market; that's why one of the reasons we managed to land a major contract from a major frontier AI company and deliver in what was basically a record time period by the end of this year. We aren't a traditional colo operator; we aren't a hyperscale cloud; we believe in the structural advantage for the AI demand environment as it stands now. A few points to underpin that. We own and control our power; our total power footprint comes to about 393 MW across 4 sites. We have 2 business streams. 1 is NeoCloudz GPU as a service, which, by the way, went live as of today, and that greatly de-risks our ability to execute. We are in fact live and earning AI revenues as of today, through a contract we signed and announced a few weeks ago. This is our first NVIDIA B200, B300 GPU as a service offering, which we intend to scale up. Then there's the colocation strategy, which is essentially a modified lease structure where we sign a 10-year deal for $1.1 billion that locks in stability and income without the enormous CapEx of GPUs. Our business model today is to build out both businesses — colocation, because we do have power and sizable power, and GPU as a service, because we are vertical and we try to optimize every megawatt of power we own. Being vertical gives us a much bigger revenue stream. The capital-raising process over the past 6 months was painful, and it diluted the company. At this point, with essentially less than $25 million in cash and zero debt, we're able to use that balance sheet, and we can now steer clear of dilution down the road or reduce it by funding the expansion of our upcoming data centers via debt financing. A solid balance sheet gives you access to, you know, more favorable financing terms; we're contemplating a 70/30 LTV loan to cash. Rather than moving from 100% self-financing, we will shift to 70/30% loan to cash financing. We've already signed a term sheet with a lender so we can grow the business through smart debt financing. As of today, we've brought roughly 210 MW online connected to the grid that we can convert into AI revenues. We've got another 180 MW on the way, where we expect to receive our load study back by the end of 2028. That will bring us to a total of approximately 393 MW of secure capacity across the sites we own. We also have an LOI announced a few months back with a massive power plant in West Virginia totaling 1.3 GW, and we're now in a stronger position to explore it and reach terms, if possible, to scale our business exponentially for 2028, 2029, 2030. The year-to-date has been an execution achievement.

Michel Amar executive
#2

We signed a bare metal GPU rental deal with SubQuadratic, an emerging new lab, and we got it done in a very short window and on schedule. Our contract, you know, was dated May 15, RFS, and we hit the deadline. We think that was amazing teamwork. We joined forces with the NVIDIA folks and the Supermicro folks plus our crew in Alabama, and as of today we completed our first GPU bare metal rental. That was a big achievement, number one. Number two, we built up a very solid balance sheet financially that we can draw on going forward, and right now we have no worries about running short of cash. We're able to grow. Number three, we inked a huge agreement with one of the top chip manufacturers in the world, worth $1.1 billion and expandable to $2.5 billion; that gives us stable, foreseeable revenue over the coming 10 years. Plus a great deal of credibility. At this point we're talking with numerous leading players about expansion across the next 3 years. Today. We began today; we wrapped up last night, the first GPU bare metal GPUs for our customer. You know, we've been grinding very hard over the past few weeks. We got the GPUs from NVIDIA in the first days of May. We ran tests, commissioned them, and got the GPUs over to the customer. As of today we're kicking off our AI revenues. In 2026, we anticipate growing our GPU bare metal by another few megawatts. The first contract we just delivered today came in a bit under 1 MW, and by year's end we expect to deliver an extra 6 MW. We'll see the revenues possibly in the last quarter of this year or the first quarter of next year; the initial phase of this colocation contract is set for delivery December 2026. That's roughly 15 MW; we'll wind up bringing the second phase in-house, Q1 2027, adding another 25 MG for 40 MW total, which works out to a running $8 million-$9 million a month run rate for colocation. On the GPU side, it'll come to about $7 million-$8 million a month, first quarter of 2027. Power isn't a problem for us. For most companies, getting access to power is one of the biggest bottlenecks. We've got access to power. Our plan is to convert that power into AI revenues. For 2027, what we're aiming at is having 90 MW of colocation up and running, plus roughly 10 MW to 12 MW of GPU as a service. Together, that gets us to a total run rate of $300 million per year. Then in 2028, the plan is to bring on another 50 MW of colocation along with 20 MW of GPU bare metal, which would put us at a run rate of $450 million to $500 million annually. Come 2029, we'd add 100 MW more of colocation and another 50 MW of GPU bare metal, taking us to a run rate of $800 million to $1 billion each year. That's the objective. We can get there provided we secure the financial strength and the debt financing instruments, and on the financial side we're executing on that right now. Our stock is very liquid, which makes it quite attractive to various institutions. We're seeing a lot of interest from firms wanting to partner or lenders wanting to partner with us and back our growth. That's what we have planned over the next 36 months.

Operator operator
#3

Appreciate it. That wraps up the Q&A portion for today. Our thanks go out to each of you for joining us and for the interest you've shown in this conference. At this point, feel free to drop off the lines or exit the webcast, and we hope the remainder of your day treats you well.

For developers and AI pipelines

Programmatic access to Digi Power X Inc. earnings transcripts and 133,000+ 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.