Digi Power X Inc. (DGXX) Earnings Call Transcript

Hi, everyone, and thank you for joining us today to go over Digi Power X's second quarter 2026 results. I'll start with the highlights for Q2, covering the three-month period that closed on June 30, 2026. We recorded $6.6 million in revenue, a figure that mirrors the intentional phasing out of our legacy business lines as we moved toward AI compute and colocation. On the GPU side, we recognized $1.1 million, which represents our first bare-metal GPU rental, coming from an initial batch of B200 and B300 GPUs set up at the Columbiana, Alabama site; that deployment kicked off around the third week of May and continued until the close of June, giving us roughly five weeks of revenue. On an adjusted EBITDA basis, we came in positive, posting $3.3 million, which is a $3.2 million improvement over Q2 2025. After factoring in depreciation along with all noncash-related items, the quarter ends with a net loss of $14.4 million. Now looking at the balance sheet and liquidity position as of June 30, 2026: cash and cash equivalents came to $142.4 million, compared with $1.7 million. Working capital reached $131 million, marking a $130 million year-over-year gain versus June 30, 2025. Basically, twelve months ago our working capital was zero. Digital assets holdings totaled $14.3 million. We spent $127.5 million on net fixed assets and equipment deposits for the Columbiana data center buildout — that's up 447% from a year earlier, and it represents capitalized investment at the facility in Columbiana, Alabama. We carry no long-term debt. Assets totaled $279 million, compared with $37 million the prior year. Turning now to the balance sheet and liquidity as of today: our cash and cash equivalents sit at about $150 million right now. So far this year, roughly $110 million in capex has gone toward GPU gear and data center construction tied to our Cerebras contract at the Alabama site. Let's shift to operational highlights and the outlook. For the AI campus located in Columbiana, Alabama, the target remains having Phase 1 available for service by December 2026, with Phase 2 following at the close of Q1 2027. Across its segments — GPU-as-a-Service and bare-metal rental, Energy Sales, and Legacy Colocation — Digi Power X anticipates Q3 growth of roughly more than 100% versus Q2. Q2 represented a major inflection point for the company, since it recorded its first-ever AI revenues, together with positive adjusted EBITDA and considerable year-over-year expansion. The strength of the company continues to come from our balance sheet. Right now we're sitting on roughly $150 million in cash and cash equivalents, we don't have any long-term debt on the books, and so far this year we've already put about $110 million of CapEx to work building out our infrastructure over in Columbiana, Alabama. For the Alabama data center, we're in advanced talks on debt financing at the moment, which moves us further along the growth path and helps reduce dilution. As everyone knows, I've been a major shareholder since the private days back in 2016, when I put in around $8 million of my own money and haven't sold a single share; dilution hits me first and foremost, so I'm quite sensitive to it. That puts me on the same side as the vast majority of shareholders. Dilution is something I work hard to keep to a minimum. Go back a year, and we had virtually no money on hand — only $1.7 million — and almost no assets, $37 million. There was effectively no working capital. In order to bring in cash and thereby meet the requirements for debt financing, we were forced to depend on particular instruments, including ATM. At this point I'm ready to take questions. We've more or less completed our equipment purchases for the first 15 megawatts. All of the long-term equipment items are now on order, and delivery is actually starting this month. As a matter of fact, we're running a few weeks ahead of schedule. That makes us quite confident Phase 1 will be ready by December. Phase 2 is essentially handled as well; we locked in — and as was said before, within our total assets you noticed a large share of deposits for equipment; every long-term equipment item is secured for delivery in November and December; and we're confident we're positioned well to finish Phase 2 by March 2027 as well. That's a really good, interesting question. Power and permits aren't a problem for us, because 10 megawatts isn't much within our power infrastructure. The power and the allocation are indeed available to us. GPUs demand a tremendous amount of CapEx. We began purchasing GPUs, and our operation is running at 100% uptime with great success — a genuinely strong result for us. As everyone knows, we placed more orders for GPUs—specifically the Vera Rubin—and those should show up early in the first quarter of next year, bringing in additional revenue from GPU bare metal. Recently there's been encouraging news, which you've likely seen: NVIDIA, together with a number of banks—BlackRock, Blackstone, Goldman Sachs, KKR—created a fund and, in doing so, effectively set a residual value for GPUs. With that in place, securing debt financing or asset-based financing against GPUs becomes much easier, which allows us to move faster on our GPU bare-metal initiative. So we feel confident that we can get financing for GPU expansion separately from the data center; that's excellent. A lot of shareholders ask me about this all the time. I want you folks who own shares to understand something: we've been running a Bitcoin mining operation that goes back a long way — one location has been active since 2016, and another since 2021. At that time, we possessed complete permits allowing us to run at those two locations under moratorium rules. This isn't some brand-new piece of legislation; it's actually a law that has existed for the last four or five years. And here in New York, we hold grandfathered status. We cannot expand the power capacity we already have. As I understand it, we're using 60 megawatts at North Tonawanda and 18 megawatts at Buffalo. That means we're allowed to carry on drawing that same amount of power. Right now, we're not allowed to add more power. But the goal is straightforward: secure one more colocation agreement worth 40 to 50 megawatts, along with GPU-as-a-Service or bare-metal for an extra 8 to 10 megawatts. That's our target for 2027. Okay, that site has plenty of capacity to handle what we're going after, which is great. The plant is partly owned by Ajay Gupta, a member of my board, and a letter of intent has been signed. This is a massive piece of infrastructure. With 1.3 gigawatts, the scale is enormous, and I have zero interest in getting dragged into fights over the generation side of that facility. What I’m after is access to the land so we can connect to the electrical infrastructure and draw utility power as high as 1.3 gigawatts. So I’m putting in significant effort alongside my fellow board member to chart a expansion course that begins at 100 megawatts and scales to 1.3 gigawatts, since that infrastructure already exists, and we can do this without taking on possible liabilities tied to the power plant. That explains why progress has been somewhat slow. Regardless, this would likely be something that plays out as a growth initiative between the close of 2027 and 2030. We hold 2 parcels in that area, totaling roughly 40 acres. Over the past 12 months, we bought an adjoining piece of property so that we can properly plan and construct a suitable data center on the site. Right now we’re engaged with Duke Energy and working through zoning to secure every required load study and permit for a large-scale data center.

We also think that asset can be put to work around 2029 and 2030, adding somewhere between 150 and 200 megawatts of power. So from now through 2030, '31, we're trying to map out our growth across all of these sites; and if we simply manage to convert or shift just half of these assets into colocation and GPU bare-metal, that will turn into an enormous, enormous, enormous company. Right. Now, the reason U.S. Data Center holds 48% today instead of 55% is that we began raising dollars separately so we wouldn't have to pull the cash we need entirely off Digi Power's balance sheet to support both current data centers and the growth of future ones. We can't get sidetracked financing what is essentially an equipment business — which is what U.S. Data Center is for: designing, manufacturing modular systems, prefabricated mini data centers that get assembled at various sites; so we didn't want to divert or combine any cash from the Digi Power pool. That's why we started raising money separately, and we brought in as much as $125 million at a pre-revenue valuation. We intend to raise a bit more at a higher valuation; and the plan for '27 is to deliver modular systems, essentially like a piece of equipment, to different sites. So its business purpose is entirely different from Digi Power, which is focused on power and data center processing plus GPU bare-metal; so that's among the biggest concerns for every shareholder, myself included, as the largest shareholder, all right? At times the ATM is a necessary evil. And 100%, I can promise you that with $1 million last year, $1.7 million last year of cash, there was no way I could launch, borrow for, or halt any kind of AI business — impossible; so I had to be — I had to place the company in a position where we could qualify to draw lenders with a strong balance sheet. Naturally, all of this must be done in moderation. And I believe we raised the bulk of our ATM capital early in Q2 at an average of — I won't promise the average, but considerably above where our stock trades today. I think our most recent ATM draw came at $7.25 or $7.50 per share. So we didn't draw any dollars under that, if I remember correctly. So at this point we've built up enough cash to, first, self-fund most of the Alabama data center; so from a financial standpoint we aren't exposed to execution risk. And I think securing the execution of that first data center, which is a genuine catalyst for us, mattered a great deal. After all, how could I talk to a lender without cash sitting on my balance sheet? Lenders want to see cash, a strong balance sheet, especially this year. Alright, since our balance sheet is in solid shape — and I'll get into the talks we're having today about debt financing for the company — we ultimately brought in Goldman Sachs, a highly respected institution with a track record of syndicating substantial financing packages for major corporations, right? And we feel really good about securing debt financing not just for this one project, but also getting cash returned, because as was mentioned before, we've already spent more than $110 million on CapEx. So cash is coming back to us. Our strong balance sheet stays intact. We're finding an excellent partner — a financial one — not just for this year, but for next year and the growth years ahead. Every card is on our side. A bit of trust from shareholders — the same trust I put into my own company when I invested privately 10 years ago. And within the next 3 months to 6 months, we'll be sitting in an incredible position. You need to recognize that over 12 months, we transformed that company's figures enormously. We moved from $1.7 million to $150 million in cash as of today, plus $110 million in CapEx. That's an enormous difference. We went from $37 million in assets to $279 million in assets. Here you have a company holding genuine assets, with what I consider a very, very low market cap relative to those assets and relative to the cash. So we're extremely near to carrying out our plan. We execute. That's what we're here for. Execution is where our attention is; we possess every element needed for it. And in the next quarter, we ought to have a fantastic update for all of you. I believe I've just covered it. Those were my closing remarks. I must stay focused, and the whole team must focus on the work. I wanted to mention as well that we're assembling a very, very sharp team in Silicon Valley alongside our CTO, Jag; we've got highly talented engineers developing the software layer for GPU-as-a-Service. We secured an office opening soon, in fact quite near Cerebras, NVIDIA, Supermicro. And we intend to deliver more across all verticals at every different layer of this business; plus the team we're assembling comes from places like — Jag, for instance, from Oracle, where he spent around 11 years as a lead on 2-gigawatt data centers. So I'd say we're in good shape. We simply need to keep pushing focus. We are — the path exists, right? Everything comes down to execution. And we're tying up the loose ends on the financial side to reduce dilution. That's the aim for — it's my aim. It's every company's aim, though at times you must reach a point where debt financing becomes available to you. We've reached that point now. Thank you, everyone. Thank you.