Digi Power X CEO Michel Amar: $1.1B Cerebras Deal & AI Growth | Frank Curzio Full Interview
35 minAudio
Digi Power X (NASDAQ: DGXX) Chairman and CEO Michel Amar joins Frank Curzio at the 2026 Curzio One Wealth Forum for an in-depth discussion on the company’s AI infrastructure expansion, financial strategy, Alabama data center development, and outlook for 2027. In this full interview, Michel discusses Digi Power X’s transformation from Bitcoin mining into a vertically integrated AI infrastructure company, highlighting its $1.1 billion, 10-year colocation agreement with Cerebras Systems, ongoing construction of its 40 MW Alabama AI campus, and the growing role of NeoCloudz in GPU-as-a-Service and AI inference. The conversation also addresses important questions from investors, including the company’s balance sheet, its announced $200 million debt financing commitment syndicated by Goldman Sachs, shareholder dilution, equity compensation plans, and management’s approach to long-term growth. Key topics discussed: • Progress at the 40 MW Alabama AI data center campus and anticipated development milestones • The $1.1 billion Cerebras agreement and potential expansion opportunities • Digi Power X’s financial position, capital investments, and financing strategy • Revenue-generating GPU-as-a-Service operations through NeoCloudz • NVIDIA Blackwell GPUs, AI inference, and the NeoCloudz Token Factory • Power infrastructure advantages across Alabama, New York, and North Carolina • Shareholder concerns, stock dilution, RSUs, and employee incentive plans • Management’s revenue expectations, execution priorities, and 2027 growth outlook Michel also shares his perspective on why access to connected power infrastructure, disciplined capital allocation, and speed of execution are becoming increasingly important in the AI data center industry. Watch the full conversation for an update directly from Digi Power X’s leadership. Recorded October 26, 2026, at the Curzio One Wealth Forum. CHAPTERS 00:00 - Introduction with Frank Curzio00:45 - Digi Power X’s transformation and financial progress02:58 - Balance sheet and $200M financing commitment04:30 - Power infrastructure and development advantages06:00 - Cerebras agreement and AI customer demand07:30 - NeoCloudz and revenue-generating GPU services11:03 - Alabama campus economics and 2027 outlook13:40 - Shareholder concerns and executive compensation16:47 - AI inference, NVIDIA GPUs, and compute economics19:36 - Power availability and the NeoCloudz Token Factory22:10 - Alabama operations and community engagement26:34 - Construction progress and upcoming milestones29:28 - Capital allocation, financing, and dilution31:46 - RSUs, stock options, and attracting AI talent34:20 - Closing remarks ABOUT DIGI POWER X Digi Power X is an AI infrastructure company operating a vertically integrated portfolio of power assets and data center capacity. Through its infrastructure development and NeoCloudz GPU-as-a-Service platform, the company is focused on delivering scalable computing solutions for the growing AI economy. Website: https://www.digipowerx.comOfficial Press Releases: https://www.digipowerx.com/press-releases NASDAQ: DGXX | Cboe Canada: DGX Special thanks to Frank Curzio and the Curzio Research team for hosting the discussion at the Curzio One Wealth Forum. Disclaimer: This interview contains management opinions, estimates, projections, and forward-looking statements that involve risks and uncertainties. Anticipated financing, construction timelines, revenues, and other future developments are not guaranteed. Viewers should refer to Digi Power X’s latest official press releases and regulatory filings for additional information. This content is for informational purposes only and does not constitute investment advice. #DigiPowerX #DGXX #MichelAmar #FrankCurzio #AIInfrastructure #DataCenters #NeoCloudz Get full access to Mining’s Substack at miningdiscovery.substack.com/subscribe
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_01I really believe that Michel is one of the most misunderstood CEOs because I've met him at his house even before I recommended the stock in Miami. And it was fantastic. He cares so much about his shareholders. And to see what he's done with this company, I think everything that he said he was going to do since I first met him, you have done. And I want to start out by saying, you know, how do you feel about the industry? What's going on? We just had a guest that we just talked to, which is Alex from Vivo. But you're seeing a whole industry coming down, CapEx concerns.
SPEAKER_01I don't know if people know that you have your funding in place from Goldman Sachs. So, you know, a lot going on where I feel like it shouldn't really pertain to the two companies that we're talking about. You know, the state of the industry, how are you dealing with it?
SPEAKER_00Good morning, everyone. So I will start with a little bit of background. And I will separate speculative versus execution. And I am myself a shareholder. I invested in 2016 $8 million of my own money into the, at the time, Bitcoin mining. Started to build Tier 1 farms. Then 2020, COVID, we got public there a couple weeks before COVID.
SPEAKER_002021, we uplisted to NASDAQ. We acquired a power plant, Gas Combine Cycle, which we've been running successfully for the last almost five years now. And we are still running legacy Bitcoin mining beyond the meter. So we do understand how to generate power. We are getting capacity dollars from that power plant. And when I, in retrospective, when I look at last year, Last year, January,
SPEAKER_00last year, 2025, we had $1.7 million in cash. Our net assets on the books were about $37 million. We were in the Bitcoin mining, a dying business. And this year, 2026, a year after, we have about $135 million cash. We capped this year $155 million so far towards Alabama data center,
SPEAKER_00we have at least, and I would say we have much more than that, $280 million of net assets, which is much higher than our market cap of today, by the way. And that doesn't include the gas power plant that, in my opinion, is worth $70-$75 million. It doesn't include all the assets. We own all the lands, all our sites, Buffalo, North Carolina, Alabama. So I think we are in a very great position financially.
SPEAKER_00We still have zero debt. We got approved for a debt financing from a major fund for $200 million, which we should close, hopefully, before Thanksgiving. That $200 million is going to help us reinforce our balance sheet, because basically we are self-funding our data center. In terms of execution, we signed a billion-one contract with Cerebras
SPEAKER_00and OpenAI in May. And the data center, and you are welcome, all of you, to come to Alabama, is basically almost done. Phase one will be done in seven weeks, eight weeks. And phase two, we started already, will be done in March, 2027. So it took us basically from scratch to execution from the contract in May.
SPEAKER_00It took us about seven months. To execute, which makes us one of the best executor in the U.S. So a lot of people talk about real estate and a lot of friends call me and say, look, I have a site, power site, and a lot of them do not know what they are talking about. But I do because we are in that business for 10 years. We generate power.
SPEAKER_00We have utility contract for the last 10 years. We understand speculative from reality. So a lot of people talk about lands that they claim they own with tons of gigawatts, but maybe they are lucky it's a three to five year story. Very little companies really own a connected land. So in Alabama, We already have the substation connected,
SPEAKER_00so it's much easier to be fast, correct? In North Tanarwanda, we have a power plant by generating power. We have 60 megawatt generation and 60 megawatt hydropower from the utility. So all of this can be, of course, developed and prepped. Obviously, we have a very strong financial discipline. And as you know, we are right with what's happening in the market. You go from big companies that have huge debts, not to mention people like Corweave,
SPEAKER_00$35 billion debt, down to companies that have zero debt, us. We are much smaller, but after the bubble burst, like it's doing now, Who's going to be the best position to survive and to thrive? This is us. Because we have a contract. That's not a potential contract. It's a signed contract. Billion one to two and a half billion dollars. This tenant, Cerebras and OpenAI,
SPEAKER_00every week are asking me to take all my power available.
SPEAKER_01Is that right?
SPEAKER_00100%. I could sell every megawatt that I own to Cerebras Opener. We choose not to because we have on our lap today five or six ready contracts to sign with other customers. Some are IG customers. And upon the advice of our bank, Goldman Sachs,
SPEAKER_00they recommended us to try to diversify our customer base. So we don't have an issue. There's a huge demand in AI today. And the AI market today, AI shifted from training to inference. Every query prompt that you use on ChatGPT or Cloud or Mule, you know, Any query, any robot, any agent runs on tokens.
SPEAKER_00We run GP as a service as we speak since May very successfully.
SPEAKER_01How many companies are doing that making money off it already? Because I think that's a big part of the story too.
SPEAKER_00So it's very important because, you know, we just don't want to drive a car. We want to know what's inside the car, the engine, the wheels. We want to understand every slack of AI. And we are. We are capable today, and we proved it. And many companies came to our site to look at our modular system and our bare-metal GPU rental. We are running now, as we speak, since May,
SPEAKER_00100% uptime The B300, B200 from NVDA. And it's bringing us income, about $10 million a year, million a month basically now. Cash flow. All of it was capexed and self-funded, by the way. And we proved the concept. We are capable of running vertically. Nebius, N-Scale, Cerebras, OpenAI saw our operation in Alabama and are very,
SPEAKER_00very impressed with our execution level. We are very close and we announced a partnership with Supermacro, where Supermacro is partnering with us to develop this modular pod system that are very fast. We are working on a deployment for 2027. Of course, we are very, very careful how we present it to the public because of what's happening in the last few weeks.
SPEAKER_00In the last few weeks, we basically lost 50 or 60% of our market cap. We're not the only one. Misery likes company, thank God. But last year, January, our market cap was $50 million. Today, it's about $225 million. After, we went down 70%. So, you know, I'm a very simple guy. I look at my assets. I look at my debts.
SPEAKER_00What are my assets? I spent this year $155 million. I invested $155 million that are bringing me money and cash flow for 27, okay, so far. I have $135 million cash. That's $285 million. I own all the assets. The power plant is worth 75 million. That's 360. The land in Alabama we own, 100%, 70 megawatt. It's worth 40, 50 million.
SPEAKER_00So we're at $400 million already. We have North Carolina. We have Buffalo site. Both of them are owned by us. North Carolina will be 150 megawatt power site. That'll be worth 50, 60 million. So if you take all the addition of our assets, Not including the good faith, the fact that we have developed proprietary software under the company, we are at over $400 million of assets.
SPEAKER_00Our market cap is $225 million. So I think it will be an adjustment soon, in my opinion. Now let's talk about 2027. So 2027, December we start to collect Rent as a co-location deal with Cerebras. March will be collecting full rent. Run rate starting March, EBITDA,
SPEAKER_00about $6 million a month. All of this is already way advanced in execution. It's not a project, it's not speculative. It's there.
SPEAKER_01It's not what you're going to do. It's what's happening. Okay.
SPEAKER_00And we can extend to two and a half billion dollars with this Cerebras contract. I got very close to the CEO of Cerebras, Andrew, who wants to push us to do more. And he wants to push his Cerebras via the CS4, CS5, CS6. We are designing now pods for the CS5 and CS6. Incredible product for inference. Very fast. We have a very good relationship with NVIDIA.
SPEAKER_00We developed and designed our pods with NVIDIA and Supermacro. We are in a great, great position in my opinion. So now we are with the entire market getting hurt. I never sold a share of my company. So it doesn't hurt me personally because I look at long term and I believe that The race is on and your time to execution,
SPEAKER_00to revenue is very, very important. That ratio is very important. We cannot talk about 2035, 2040. We have to talk about what happened in the next 36 months. In the next 36 months, we should be on the run rate of $40 - $50 million a month revenues and on the run rate of 20, 25% EBITDA on these revenues.
SPEAKER_00So I think we are an incredible company. I'm very proud of the team, considering that we have a very, very small team. And when I read the sometimes comments on X. I
SPEAKER_01want to get to that. I want to get to that. Because there's a lot going on in X. And sometimes when I talk to CEOs, especially when I work with them, It could be crazy, right? So you actually had a proxy where it was an option plan, sparked heavy debate because people thought you were increasing that because you wanted to pay yourself more, when in reality, I think, and this is a good story to tell, that that needs to be told when you have smaller companies. When you have smaller companies and you're in the right sector, Google and all the hyperscalers and Meta and Microsoft have the ability
SPEAKER_01to pay good employees anything related to AI. They'll pay them millions of dollars. They'll poach everyone that they can, which makes it very difficult for the smaller companies to operate. So what makes it beneficial to smaller companies? Like the engineers in the room and working with smaller companies is you get the benefit where if you get options, you participate where you'll make much, much more of that if the company does well. So these people are great engineers and they're only going to go to companies that they believe are great. So talk about that, because I think you got a little heat on X, and people are a little pissed off about that.
SPEAKER_00So I want to clarify two subjects. Subject number one, last year, compensations, that I see on X that I made about $19 million. That's how I wish I made it. 86% is paper vested in three years to motivate us based on A market cap going up and so on. And I should be motivated because we are doing a lot of work.
SPEAKER_00Myself, Alec, we hire the CTO that comes from Oracle that's pretty smart. And we are trying to build a team now in Sunnyvale, in Silicon Valley. But we are doing a lot of work. And I compare sometimes with other companies of my size. And they have like 50, 60 people. We don't. We are a very small team because we want to execute. We don't want to have these inefficient,
SPEAKER_00lasting meetings that lead to nowhere and no decision. We are very, very compact. We make a lot of good decisions. As I said, I'm not a seller. I never sold a stock. I just bought stock, actually, Alec as well, the president. I just bought 40,000 stocks just to show good faith. I would buy more now, but I'm not allowed because the Q3 is coming, earnings. But in order to attract talents,
SPEAKER_00today it's very difficult in Silicon Valley. So we acquired, by the way, a beautiful office next to, it's an office lab, next to Cerebras, NVIDIA, Google, great location, amazing space. We closed on that office. It's about 18,000 square feet on parking. We're going to do a lab there. We're going to run some GPUs and test GPUs.
SPEAKER_00We closed on it a couple of weeks ago. And of course, we want to build up a team. So as AI has shifted from training to inference, all these You know, new agents that you're using, hopefully, GrockBot and Muse and whatever, they need a lot of tokens, in and at. Pre-fill and decode. They need a lot of tokens, and these tokens are processed by our data center, okay?
SPEAKER_00And the chips that we put in it, the GPUs, and the value of the token In my opinion, is at least double the value of bare metal rental. So you see the price increase in bare metal rentals. It went from six months ago, a B200, B300 would be like $350 to $375 an hour. Today, they are renting at $6, $7 an hour. Huge return. You're talking about instead of 36-month return ROI, now it's 18-month return.
SPEAKER_00So now we place the new Vera Rubin, which we will get end of Q1, incredible return on investment. You're talking about less than two years. So it's a business we like as long as we resolve the financing. And I believe we resolved the financing. How? I'm asking 30% deposit on the total contract value,
SPEAKER_00and customers are accepting. So take Vera Rubin, for instance, five years at $8 an hour, which is a good price. We could sell it for $10 an hour, $12 an hour. At $8 an hour, a 3.6 megawatt pad that we build with our sub-US data center, by the way, A 3.6 megawatt brings about $400 million in five years.
SPEAKER_00Only for 3.6 megawatt. Now imagine that we resolve the financing with a customer deposit, 30% of $400 million, it's $120 220 million. I take this $120 million, I buy the chips, I buy the parts. I mean, business costs me nothing. No capex issues. And I'm getting a return in 24 to 30 months. And then it's 90% margins.
SPEAKER_00And I don't see that business slowing down. Not from the perspective of GPUs. I think the constraint is not the chips anymore. The constraint is smart power. So what do I call smart power? Versus speculative power. If you take all the speculative power out, which are Greenfield or Brownfield, but there's no connection, there's no utility contract, there's no generation,
SPEAKER_00they're talking about turbines that takes three years to get. If you take real power like we have, we have real power connected to the grid or connected to a gas plant, This power is very, very valuable if you run it all stack as GPU bare metal or if you sell tokens. So I'm proud to say that this week in Berlin, Alec and JAG and Naresh and our engineers, they launched our Neoclade-Z token factory so we can basically now sell tokens
SPEAKER_00to everyone. Tokens is about double the price Of bare metal. Very huge margins. So the run rate in token factory slash bare metal slash GPU as a service, we're going to increase it as we resolved the capex and financing. And that's going to give us a huge revenue per megawatt.
SPEAKER_00So power being scarce, and we don't understand that, that's why when I hear people talking about gigawatts here and there, gigawatts, gigawatts, where is it? It's very hard because first, as you're all aware, politically, anything that's above 100 megawatts, communities, governors, PS, everybody's slowing it down for years. They don't want to see a one gigawatt center anymore.
SPEAKER_00So we are under the radar at below 100 megawatt. You know, they are not paying as much attention that above 100 megawatt. So when people talk about gigawatts, practically, I don't see it. Okay, it's a dream. It's not going to happen. The politicians are going to stop it. Okay, that's a fact. And if you see what's happening in the midterm elections, you realize that the big subject is data center. That's a big subject.
SPEAKER_00People feel AI, they misunderstand AI. I try to explain that our AI data center, we don't use water, it's all in loop. Liquid cooled, but all in loop. It's very, very low sound. And I have a video I posted. You can barely hear when you're outside the modular system or the data hall. You cannot hear the sound. So no sound, basically.
SPEAKER_00No water. We take care of the environment. We are very, very, very happy with the community there in Alabama. We participate in every event. We are very close to the mayor and the community. We hired a lot of people. And that's what matters to the communities. We hired a lot of people. We have about 250 people working seven days a week now in Alabama.
SPEAKER_00A lot of these people are local. We are creating a boom, economic boom, in that city in Alabama. To the point that they love us. So we are trying to do a very good job of being loved where we go. Alabama is a great state. I think, hopefully, Tommy Tuberville is going to be elected, and he is actually a good friend of mine. And hopefully, you know, we're going to do more centers in Alabama.
SPEAKER_00We are working on centers in West Virginia. They are very pro-data centers. But the common denominator is We are not responsible for what Cloud or OpenAI does. When we do a center, we are basically a compute cloud service center. We are not deciding how OpenAI is going to develop their apps. So they are the ones that you have to fear, not us.
SPEAKER_00I tried to explain that and share that information with the communities, but I think we're doing a very good job with our communities. The toughest part is New York, but I think we are prevailing. We are prepping New York now as we speak, and I think New York is going to be a great center. Basically, my accounting with the community is like, if I'm going to spend $100 I'm going to give the community $3.
SPEAKER_00So if we spend $300 million in a data center, I'll give them $9 million, 3%. And $9 million, when you have 5,000, 6,000 people in the community, it's impactful. I impact them so they forgive me, or they forgive the perception that data centers are bad. Now, coming from a Bitcoin miner, For me, it's like a breeze, because being a Bitcoin miner was really evil and dark at the time. So now we feel much better.
SPEAKER_00So I believe we're in a very good position. We have signed contract, ten years plus seven years for two and a half billion dollars. We are going to sign, it's signed already. And I would say that Cerebras is very, very happy with us. Among all the contracts they've signed, we are their number one executor. We are the rare co-locator
SPEAKER_00that was capable of developing a center from scratch may design to execution in seven months. The good news, we used the designer that actually cost us a lot of money, but we used the designer that Elon Musk used for his center that he did in 100 days. Same design firm. Cost us about $4 million just for the design. But they came through in 60 days.
SPEAKER_00In 60 days they designed an entire liquid cooling 40 megawatt IT load center. We also, and I don't think I mentioned it publicly, but it is public, we had there about 20 acres and we bought next door the land 33 acres. So now we have 50 acres which give us enough room to operate and to do construction.
SPEAKER_00We have about 50 trailers there, walking trailers, to support the construction. We work seven days a week. 250 people. We work shifts sometimes 24-7, 24 hours. Night shifts where they pour the cement, it dries, and then we put all the chillers, by the way, for phase one, phase two, installed. All of them. All of the equipment bought for both phases.
SPEAKER_00So we're in pretty good shape. How do I see the company growing? Execution. Next 36 months, we want to be a $500 million a year or more revenues with a 25% EBITDA. That should be very doable even though we are slowed down by the crunch on debt financing.
SPEAKER_00Debt financing seems to be a little bit scarce these days. I think it will reopen after the bubble is dealt with. And I think after the election you'll see a rally. That's my opinion.
SPEAKER_01So people don't know a lot, I know this from knowing you for a while now, is The innovative part of you, where you were one of the first Bitcoin mining in 2016, right, I think? You were first to adopt this strategy where, you know, the irons and everybody else are getting in, where you're generating money. You're actually generating money off of AI. You're not saying you're going to do it. You understood when we talked even last year of the funding part and the CapEx part where you created a USDC and understood that.
SPEAKER_01I feel like investors didn't understand that. We came and told that story and helped that part out as well. You've been ahead of the curve every single time. I guess, what's next for you going forward? Because you've been ahead of the curve so far. You see the risks coming. I guess, how do you better tell that story for yourself? And also, what are the biggest risks that you see coming? Because it seems like you're ahead a lot of them before everybody else.
SPEAKER_00So, in retrospective, in May, when we announced the contract and we raised a quick $175 million, diluting at $8, $7 a share, That was the right move, correct? That was the right move and the only move for us to be able to self-fund and to be in a cash position where we are safe. We are not relying on debt. Debt is to make our balance sheet strong so we can move forward and leverage it for new deals.
SPEAKER_00So that was the right move. You know, there are cycles. That was a good cycle. I'm very happy we did it. Sometimes people are saying, oh, you're using the ATM. We haven't used the ATM since June 13. I haven't sold one stock. So I protect the shareholders because I protect myself. I am a shareholder. I never sold back. So, of course, I'm trying to mitigate dilution as much as I can.
SPEAKER_00And that's why we are doing debt financing. I hate debts, but I think that To be able to have access to debt financing is important. And specifically in this environment, the fact that we are eligible for debt financing shows you that we are in a very strong financial position. Because they are very hard on delivering debt. Actually, a lot of my peers I see did convertibles.
SPEAKER_00They couldn't get debt. I'm not doing a convertible. I'm doing a 100% debt. And I could have done that debt financing a few months ago with BlackRock, HPS. Their rate and terms were too high, so I refused. I was approved by BlackRock and I refused. And I moved to syndicate with Goldman Sachs and we got two points less, which is enormous because on the $200 million,
SPEAKER_00two points is $4 million a year of interest that I saved. So that's important. I just wanted to go back to one of the questions that I did not answer you.
SPEAKER_01The RSUs and— If you can, we have two and a half minutes, three minutes left.
SPEAKER_00RSUs and option plans, okay? We did a temporary proxy, and we are doing a final proxy this week. And there were blanks on the RSUs and options. And you have, by law, to put a ceiling— So the lawyers put a ceiling, and people thought the ceiling was the plan. It's not. For the last five, six years, six years, the plan has been 10% RSUs, 10% options, rolling.
SPEAKER_00It's basically the same, except because it's full for the six years, I want to add 10% a year on the 10%. 10% plus 10% is 11%. 1% of the 10% to attract talent. My only way to get good engineers from Oracle, from Microsoft, is to tell them that here you have a huge upside on the stock that you're not going to get with Meta or Google. So I'm competing with guys that go and pay five, ten million dollar bonus, cash, to these guys.
SPEAKER_00And I want to build that team because we have very good ambition. In our business, I want to be a very strong software company as well. So we are different. And to attract them, I have to give these guys RSUs, options, so I get talents. And not go and start to dilute cash and buy people $10 million or $5 million. That's ridiculous. It's working. I motivated my CTO.
SPEAKER_00I motivated capital engineers. I lost one good engineer. You know why? I was paying him half a million dollars a year, which is not low salary. And he went for a job for three million dollars a year. Okay? An engineer. I lost him. He went from half a million to three million. So now my only way to keep and attract talents is to say, listen, we're going to be a multi-billion dollar market cap company. The opportunity is here because we are at the low.
SPEAKER_00I'll get you RSUs. You'll have 5, 10, 20x that attract them. They believe it. And then, together, we can all enjoy a huge upside. So that's where we stand. We're in a great position. Maybe next meeting I'll talk about US data center, which has very, very big potential. We have a lot of action. We are raising dollars separately because it's a different business.
SPEAKER_00And maybe next time we'll do an interview on that company.
SPEAKER_01Definitely the podcast. So listen, I really appreciate how busy you are. Thank you so much for coming. Yeah, thank you so much. Thanks.
What sets U.S. Gold Corp. (NASDAQ: USAU) apart in the gold and copper sector? Located in southeast Wyoming, the CK Gold Project has advanced through feasibility studies and permitting, positioning it among a limited...
Digi Power X (NASDAQ: DGXX) is making significant progress at its 40 MW AI infrastructure project in Alabama, advancing construction to support Cerebras and the growing demand for high-performance AI computing.In this...
Monumental Energy: New Energy. New Zealand. Monumental Energy is focused on helping meet New Zealand’s growing need for reliable and affordable energy. The company is pursuing multiple applications within the country’s...
DigiPower X CEO Michel Amar takes you inside the Company’s AI Infrastructure Campus in Columbiana, Alabama - purpose-built for the next generation of high-performance computing and artificial intelligence.The deployment...
Arras Minerals (TSXV: ARK | OTCQB: ARRKF) is accelerating exploration at the Elemes Project in Kazakhstan, with four drill rigs advancing an expanded 40,000-metre program across the Berezski Trend. In a new Resource...