Fermi Inc. (FRMI) Earnings Call Transcript

Good day, everyone, and welcome to the Fermi America's Second Quarter 2026 Earnings Call. [Operator Instructions] It is now my pleasure to hand the floor over to your host, Barry Sievert. Sir, the floor is yours.

Good morning, and thank you for joining Fermi America's Second Quarter 2026 Earnings Conference Call. With me today are our Chairman of the Board, Marius Haas; our Chief Commercial Officer, Anna Bofa; our Chief Operating Officer, Jacobo Ortiz; and our Chief Financial Officer, Rob Masson. Before we begin, I'd like to remind you that today's call contains forward-looking statements within the meaning of the federal securities laws. These statements reflect management's current expectations and are subject to risks and uncertainties that could cause actual results to differ materially from those we anticipate. For a discussion of these risks, please refer to our most recent annual report on Form 10-K and our subsequent reports on Form 10-Q. Any non-GAAP measures we reference today are intended to provide a supplemental perspective on the company's ongoing operations. With that, I'll turn the call over to Marius.

Thank you, Barry, and good morning, everyone. Thank you for joining us. One quarter ago, on our first quarter call, I did something a Chairman doesn't always do. I gave you a list, a focused 90-day plan built around 5 specific commitments, and I ask you to hold us to it. I told you the new leadership team would execute and that you should measure us not on our words, but by what we deliver. Several of you questioned why I established self-imposed aggressive deadlines on the team at a time when there was external distractions and internal transitions. Yes, it was a risk. But I have a very strong philosophy that an exceptionally talented team with a clear mission with clear priorities and well-understood time lines can achieve extraordinary things. With clarity comes alignment, passion and focus. The focus eliminates the irrelevant noise. The passion was always about doing what is right for our shareholders. Today, roughly 90 days later, I'm here to report to you on our progress. And I'll say it plainly, Fermi is in a much stronger place now than it was then. To date, we've delivered on all 5 commitments, including the one that matters most. Let me take you through our results. First, a quick reconnect to the macro backdrop because it frames the problem Fermi is working to solve. Power is the binding constraint on AI compute. Not capital. Not demand. Across hyperscalers, neocloud providers and frontier model developers, the same bottleneck keeps surfacing: access to large-scale, reliable electricity on a time line that matches the pace of AI and data center development. Delays are being reported across announced projects worldwide, driven by grid interconnection queues and equipment availability. Fermi was created to relieve that constraint, and this past quarter shows you how we are doing it. Objective number one of our 90-day plan was to secure a binding customer agreement. And I will say this was, by far, the most important objective. And the team delivered. We signed a binding agreement representing up to 650 megawatts of power. We are now in the process of finalizing the backstop agreement that sits behind it. Anna will walk you through the details in a moment. I'll simply say this: the caliber of the counterparties now standing behind our advanced energy and AI ecosystem speaks directly to the value we've discussed with you since our IPO. Objective number two was to select the right CEO for the next stage of Fermi's growth. With this in mind, we've appointed Lee McIntire effective this Tuesday. Lee has served on our Board since September of 2025 and already has a detailed working knowledge of the company and the project. More than that, he has built his entire career delivering large-scale projects, and that's the most important task in front of us right now, getting power to the site. Lee has led some of the most demanding engineering and construction organizations in the world, including Bechtel, McDermott and TerraPower, and he ran large-scale project delivery inside all of them. So when it comes to getting first power to the site in 2027, I can't think of anyone more equipped than Lee. He will work alongside Anna, Jacobo, Rob and me. We'll keep running the business exactly as we ran it this quarter. Our strategy has not changed and the pace will only accelerate. The third objective was to advance strategic partnerships. Again, we delivered. During the quarter, we announced EPC partnerships with Primoris for balance of plant on Phase 1 and TSK for early works engineering on Phase 2. These are 2 of the most capable and highly respected contractors in the industry. More recently, we also entered into a strategic alliance with Hillcore, a premier and highly experienced infrastructure and power provider around the globe, which will accelerate our ability to deliver power at speed for large-scale customers. The alliance will bring an incremental 2.6 gigawatts to the Project Matador campus within the next 3 years. There is no site we know of that has equipment and the ability to deliver 4.8 gigawatts during this time period. The demand for the power delivery is already here, and we have the team and the equipment to deliver. Fourth, we've told you we'd enhance our liquidity. In July, we delivered an upsized $431 million convertible notes offering, which significantly strengthened our balance sheet. Importantly, we did it at a low coupon and with a capped-call structure built to protect our shareholders from dilution. Rob will take you through it. The fifth and final objective in our 90-day plan was to move the site closer to power. On this front, we've made substantial progress, and we are still gaining momentum. Three Siemens Energy F-class turbines arrived at the Port of Houston in July and have cleared customs. We've also progressed site development. Jacobo will give you full operational picture shortly. Now let me put these 5 milestones into context. Any one of them a binding customer agreement, a CEO, significant partnerships, a strengthened balance sheet, turbines on the ground, would have made for a notable quarter on their own. We challenged the team to stretch and they delivered, all of them in a single quarter. That's the story for this quarter. Our leadership team kept its focus on the plan and executed it with excellence. That was the institutional discipline Fermi 2.0 was built to deliver. The Board and I are extremely proud of how the team came together and without doubt, delivered an excellent quarter. With the site advancing and the commercial picture coming into focus, let me hand it to Anna.

Thanks, Marius, and good morning, everyone. I'll start with the most important thing I can tell you about this quarter. The market didn't just validate our project in concept. It signed up for it. This is only the beginning. Customers need large-scale, reliable power on accelerated time lines, and that is exactly what Fermi is built to deliver. Over recent months, we've been actively engaged with roughly a dozen prospective customers and strategic partners. These discussions have been specific and substantive around capacity planning, delivery sequencing and the commercial frameworks required to move from interest to signature. Within that broader process, TensorWave stood out as a particularly strategic first customer for Fermi. For us, this was an ecosystem entry strategy. First, speed matters. TensorWave is a highly entrepreneurial organization, operating in one of the fastest-growing parts of the AI infrastructure market. They have the ability to make decisions quickly and critically, their demand curve aligns with our power delivery curve. Second, we believe TensorWave gives Fermi a unique position inside the broader AI ecosystem. A direct relationship with the hyperscaler is incredibly valuable, and we are nurturing those too, but it's fundamentally a relationship with one customer. TensorWave sits between the semiconductor ecosystem and the ultimate users of that compute, giving Fermi exposure across multiple layers of the market. Compute is the foundation of the AI stack. And TensorWave's proximity to that layer gives us earlier visibility into where demand is forming, what infrastructure those workloads require and which customers are likely to need capacity next. Through one anchor relationship, we gain connectivity not only to TensorWave, but also to the chip ecosystem and to multiple hyperscalers, enterprises and AI companies that may ultimately consume that capacity. For us, that is strategically important. We want Fermi to become a place where the broader AI infrastructure ecosystem comes together, not simply a campus serving a single customer. Third, the multiphase nature of the relationship reinforces that opportunity. The deal has the potential to bring multiple workloads and end users onto the campus over time, giving more customers and partners direct exposure to Fermi's capabilities. The phased approach also matters from an execution standpoint. By phasing the development, we can match capital deployment more closely with delivery time lines, while reducing complexity and preserving our ability to move quickly toward first revenue. And by pairing TensorWave with a highly capable data center partner, we gain direct visibility into who can design and construct these facilities quickly and efficiently, capabilities that can be reused across future Fermi customers. Fourth, this transaction validates the unique nature of what we are building. Fermi combines large blocks of power, secured infrastructure, accelerated delivery and permitted land. TensorWave gave us the opportunity to establish commercial terms that we believe solidly reflect the value of that offering while remaining competitive for the customer. So our decision to partner with TensorWave was deliberate. We were looking for an anchor customer that could move at our speed, consume power on our delivery time line, support economics that appropriately reflect the value of this differentiated asset and importantly, expand Fermi's connectivity across the broader AI ecosystem. Let me turn to the agreement itself. We have signed a binding agreement with TensorWave, representing up to 650 megawatts of capacity if all expansion options are exercised. The agreement is structured in 3 phases. The initial phase is a 15-year commitment for 222 megawatts of gross power through a turnkey data center solution and represents approximately $6.5 billion of revenue to Fermi over the initial term. We are finalizing the backstop agreement supporting certain obligations under the lease. We will identify that counterparty once the agreement is complete, which we anticipate in the coming days. A commitment of this scale follows extensive diligence across our power strategy, permitting, secured equipment, land control, construction plan, delivery schedule and execution capabilities. So when we talk about validation, we mean something very specific: a sophisticated customer and their partners have evaluated what we are building, agreed to commercial terms we believe are competitive and reflect the value of our offering and made a long-term commitment to the campus. And importantly, we believe this first commitment strengthens the ecosystem around Fermi and creates a stronger foundation from which to scale as we continue discussions with multiple other customers. With that, I'll turn it over to Jacobo to walk through what this looks like on the ground.

Thank you, Anna, and good morning. My focus this morning is simple. Everything Marius and Anna just described is backed by real steel in the ground and by an experienced team that is ready to execute. Over the last 12 weeks, we have refined our execution plan and strengthened our EPC relationships with world-class contractors, and our team is ready to start construction as soon as project finance is in place. We remain committed to speed to power as customers sign leases, while maintaining the capital discipline that matches major expenditures to commercial progress. Our execution team has been working with TensorWave to deliver its turnkey project on time. To minimize execution risk, Fermi is partnering with a well-known data center contractor with access to critical components and deep multi-project experience executing these kind of projects reliably. We will announce this partnership shortly. First, let me take you through power delivery and then our strategic alliance with Hillcore. Fermi plans to deliver 640 megawatts of nameplate capacity by the fourth quarter of 2027, with the first 210 megawatts available July 1, 2027. Additionally, Phase 1 includes 6 Siemens SGT-800 turbines led by Primoris, plus 3 GE 6B units and 7 TM2500s led by Relevant Power Solutions, all in simple cycle along with 200 megawatts of utility power from Xcel. We continue to make progress on our '28, '29 power delivery objectives, 3 Siemens F-class units rated up to 728 megawatts in simple cycle mode, each more than 242 megawatts, arrived at the Port of Houston a few weeks ago. TSK, Spain's largest power-focused EPC firm is delivering the fast start engineering for those turbines. The F-class units are among the most widely used heavy-duty turbines in the market. They ramp up to 40 megawatts a minute, reach full speed in about 5 minutes and carry a global fleet reliability, nearly 99%. That combination of high output, fast start and reliability is exactly what AI and advanced computing required. Our project has them, and we are ready to execute. Now switching gears, because speed to power is central to our strategy, I am excited to walk you through our strategic alliance with Hillcore Energy Partners in more detail. Hillcore and its affiliates, including JV Driver, bring more than 35 years of heavy industrial and power generation experience, spanning natural gas, steam, biomass and power boiler facilities from project development and construction through commissioning and operations. The alliance between our 2 companies brings approximately 2.6 gigawatts of additional power to the site under a build, own, operate, transfer structure, doubling Project Matador's total power output to 4.8 gigawatts. Think of Hillcore as a tenant that pays rent and delivers incremental power to the Fermi private grid. This accelerates and grows our revenue stream by expanding our total power supply to meet customers' growing demand faster than a self-build strategy only. And Fermi keeps 100% of the income it generates from base rent. Under the agreement, Hillcore finances, builds, owns and operates their facility. So Fermi adds dedicated, behind-the-meter power with no upfront capital outlay. Fermi will be the anchor off-taker, but we only commit as tenant leases are signed and each block of power is triggered by contracted demand from a tenant, each power purchase agreement is matched to the term of the lease that triggers it and fuel costs are passed through to the tenant as well. That matching takes stranded capacity risk off the table and construction, operating and performance risk sits with Hillcore. Importantly, we hold an option to acquire the facility at fair market value after year 10. First power is targeted within 24 months of the notice to proceed and the Hillcore build-out importantly runs in parallel with our own, doubling our capacity to 4.8 gigawatts of total power available. We have an amazing site, a 6-gig air permit, power equipment and experienced construction team and trusted strategic partners ready to deliver Project Matador's mission. Power is the gating asset for AI and Fermi is ready to deliver what our customer needs, more power now. Fermi is where AI gets power. With that, Rob will take you through our strengthened financial position. Thank you.

Thank you, Jacobo. The headline for me since our last earnings call is a significantly stronger balance sheet and it centers on the convertible notes we issued last month. We closed an upsized offering of more than $430 million of a 5% convertible senior notes due 2031. That figure includes the full exercise of the initial purchasers' option for an additional $56 million, a clear signal of the demand for this paper. Net proceeds after the capped call were approximately $382 million. I'll walk you through the rationale in 4 parts. First, timing. We're focused on our operations and not on timing the market. We raised capital with prudence in mind to provide the operational flexibility and financial runway to focus on strategic objectives. Second, negotiating position. A stronger balance sheet lets us advance customer and partnership discussions from a position of strength, not under financing pressure. Third, structure. We deliberately designed the structure with our shareholders in mind. We believe the convertible note paired with a capped call protects our current shareholders from dilution ahead of catalysts until the stock price more than doubles. And fourth, cost of capital. A 5% coupon over a 5-year term is an attractive rate and a time line that fits our development and power delivery plan. Tied back to the 90-day plan, this checks the liquidity objective, and it is one source among several. As we move into the next phase, our message on funding the project and capital is consistent: thoughtful sourcing, disciplined deployment, matched to commercial progress. With that, I'll turn it back to Marius for closing remarks.

Thank you, Rob. Fermi 2.0 has always been about the convergence of 2 things: a tangible asset base of over $1.5 billion that we have already built into the ground; and the institutional capability now deployed to realize its full value. This quarter, you watch those 2 things come together. In a single quarter, we delivered a binding customer agreement with one of the leading names in AI compute, a new strategic alliance and premier contractors, enhanced liquidity on shareholder-friendly terms, hired the right CEO to lead the next stage of Fermi's growth, and delivered major on-the-ground project milestones that protect our speed to power advantage. But I want to end where I began, with the team, because the most important thing about this quarter is not a single milestone. It's that we set ambitious goals and made tremendous progress. That is the entire point of Fermi 2.0. This is a business that does not wobble when the headlines do. The engineers pouring the foundations, the teams managing our supply chain and permitting, the commercial team signing agreements, they showed you this quarter what institutional discipline and excellent execution looks like. We believe we have the single most valuable power and compute site in the market. And as a team, we are aligned around one overarching objective, maximizing long-term value for our shareholders. The road ahead is demanding. We have to scale this amazing business, complete the current customer agreement, add new ones that are already far along in the process and focus on flawless power delivery to the site. But we began the second half of this year with a signed anchor customer, a further strengthened leadership team that has proven it can execute, better liquidity, power moving toward the site and we begin with momentum. We are pleased with where we are, but never fully satisfied. We can always do better, and we will. We look forward to updating you on our continued progress as we execute the Fermi 2.0 vision. Operator, we're now ready to take questions.

[Operator Instructions] Your first question is coming from Paul Golding from Macquarie.

Congrats on all the progress. I wanted to ask initially just on TensorWave and the lease agreement there. How should we think about the power pass-through component of that? What is the $6.5 billion comprised of? And maybe how you're thinking of hedging power that you'd be delivering directly from your private grid perspective, given that, that is a unique aspect of your project? [Technical Difficulty]

Operator, we can hear you. Operator, are you there?

Yes, your line is live.

I think we're ready for the next question.

Your next question is coming from Nick Amicucci from Evercore ISI.

Just wanted to kind of drill in on -- you guys had outlined during -- with the announcement of the TensorWave deal, $1.5 billion already spent. Just as we think about the spending ramp towards the initial 222 megawatts as well as the incremental like the 640 megawatts. If we could just kind of frame how much spend is left to occur?

Yes, I'd be happy to take that. So if we think about cash for the quarter, we decreased by about 50% just in line with what we talked about on the general preparation. When we think about the deal we just signed, there are specific cash flows that will be associated with that. And we talked before about disciplined capital deployment. So we're going to match our outflows to real signed deals and inflows. And you heard Jacobo talk about the same thing. We have the site, as we talked about through Phase 0 complete into Phase 1, and that will match up well with the new TensorWave deal. Did that answer question? Or do you have a follow-up on that? I want to make sure I hit it all.

No, that's fair enough. I think that covers it. Great. And then as we think about too, I guess, just obviously, now the full management team seemingly in place. And as we think about kind of this somewhat of a shift towards a turnkey solution on the data center side as opposed to kind of just renting like access to the power. As we kind of think about that, I guess, what gives you guys confidence in the ability to kind of execute on that further more all-embracing kind of build-out, if you would. [Technical Difficulty]

Your next question is coming from John Hodulik from UBS. [Technical Difficulty]

Our plan is to work with a world-class data center partner to deliver that aspect of the commitment. And we have already been in talks to contract that opportunity, and we feel fully confident that we are working with the right partner to deliver that data center.

Nick, this is Marius. I'll add a little bit more to it from a context perspective, as you had mentioned, about the team. There is no doubt that what we're going to continue to do is add and scale the critical key components of the organization to ensure the flawless delivery of what we have committed to our tenants and our customers. We have deployed our executive compensation for officers. As you saw, we have deployed our long-term incentive program for our employees as well as our short-term incentive program. So we have everything in place in order to be able to scale this organization and clearly understand that we will need to scale certain functions in order to deliver the full scope of what we're talking about. Having -- doing a turnkey solution as our first offering clearly puts us in a position of putting our arms around what a total offering could look like and will give us the experience and will give us the opportunity to be able to do this on an ongoing basis. Although, as Anna indicated, that's not going to be the pervasive model going forward. Does that make sense?

Perfect. That's very clear.

Your next question is coming from John Hodulik from UBS.

Congrats on all the milestones. First, a couple of follow-ups on the TensorWave deal. Can you guys give us a total CapEx number you think required to fulfill that -- the first phase of that contract? And maybe if there's been any spending already to fulfill that deal? And also, is there a way that you guys could give us a sense for what the margins are expected on the $6.5 billion? And then lastly, just when do you expect sort of full commencement of the first sort of 200 megawatts? And then if I could follow up, maybe a question for Anna. You talked about some discussions with other tenants. Any other color you can provide in terms of how far along you might be with a second tenant? I guess following up on that last question, is sort of the neoclouds, the sort of category that we could expect going forward of potential tenants? Or just any other color on sort of what we could expect sort of down the road or timing or quality of potential tenants you may be talking to?

Yes, I'd be happy to start. We'll talk overall about the market. Generally, if you think about it in 2 phases, a typical battery CapEx project runs in the range or the power side runs in the range of $3 million to $4 million per megawatt of gross power capacity, and you can think about this one in that range. And the data center itself, typical CapEx when you're talking about a turnkey solution, you'll find in the market is about $10 million to $12 million per megawatt. And again, this project is going to be in that range. We're not disclosing overall total cost figure for the project. And we're sharing a precise number now, not to set expectations, but I think if you think of those market comps, that's what we're looking at in this project. So kind of right in the middle of what you'd expect.

Yes. And I can take the second part of your question. So in terms of the commencement, we anticipate the build and the sort of revenue commencement to start in phases. We anticipate that somewhere to be around the end of Q3, early Q4 of '27. And again, it's a phased delivery. So once the first part of the build is ready, the customer will then start paying rent all the way up until the full 220-megawatt build is complete. In terms of your other question around just color on other customers, we are absolutely engaged with additional customers and have conversations that are very far along. In terms of neoclouds, we are absolutely focused on working with customers that are ready and available to take the capacity that we have, but we are absolutely also speaking with hyperscalers. One of the things that was interesting about this deal for us is as we've been discussing opportunities with the hyperscalers, there is an interesting path on this deal where they could potentially get capacity sooner by partnering with TensorWave and the fact that they have sort of these 3 potential buildings. And so we are both having conversations where hyperscalers could work directly with us to come to the site, and they're very interested in that and potentially take the opportunity with the multi-buildings that TensorWave now has available to them that could potentially move them to site a little bit faster. But again, one of the things that is really clear is that there is tremendous demand. There is very few kind of opportunities like Fermi. There are very few kind of sites that have no permitting issues that do have capacity. And so that has really attracted all of the hyperscalers to engage with us, and we're very pleased with how those conversations are moving forward.

Your next question is coming from Stephen Gengaro from Stifel.

Congrats on busy 90 days. I think two for me and maybe a follow-up on the prior question. The expectations, I think, that we had a while back were that you do kind of $700 million to $900 million of EBITDA per gigawatt deployed on an annual basis. And I'm just curious if that's still kind of a reasonable expectation going forward? And then the other question I had was when you think about the TensorWave contract and others, can you talk about like sort of the risk parameters around it and sort of the key hurdles we should be watching as far as time to power and what you could be liable for based on timing?

I can take the -- start with the risk and then sort of discuss the first part of the question. So in terms of the risk parameters, one of the things that you do when you are engaged in the process of driving towards an agreement is you look and see how do you reduce your risks. So part of our sort of ability to bring in a data center partner is absolutely a reduction in risk on the deal construct. And that's why we did the diligence to identify who we thought was the best potential partner. And most crucially, we identified a partner who could meet the time line. So what you do is you ensure that you line up the delivery of the power, the delivery of the data center and ultimately, where you land in terms of the agreement of when those 2 pieces come together to reduce that risk. And so that's the approach that we took on the TensorWave deal to ensure that we had the ability to have line of sight into making sure that what we signed up for that we could deliver, and we're absolutely confident in our ability to do that alongside the data center partner.

And on the -- I can just talk about the EBITDA side. While we're not really in a position or going to talk about that specifically, I think you'll see from the overall contract value and the type of contract this is, it's very similar to other market deals that have been announced out there and the expectations on the profit side are the same. We have a modified net structure here. So it's a very good structure, and we're pleased with this as our first contract out of the gate.

Steve, it's Marius. Just to add to it, just a couple of things that you should be looking forward to, and this is around the corner here is obviously the announcement of the backstop with a leading IT infrastructure provider. Obviously, project financing would be step number two. That then triggers our ability to go move everything to the site and get ready to deliver on the power. So a number of those key deliverables are clearly what the team will be focused on, but those are all well in motion and around the corner.

Yes. And just -- I'll add to that further. Again, we just want to be really clear of how we reduce the risk. And of course, on a deal like this, having an investment-grade backstop is a key part of that, particularly as you go into project finance. And so we have done a lot of work to ensure that, that is lined up and ready to go. And as mentioned earlier, we hope to be able to announce that partner here in the coming days.

If I could just add one quick one. You mentioned -- you implied earlier, I thought as far as the CapEx that you're going to match outflows and inflows. Does that mean there's not a lot of sort of Fermi cash needed to build this out that the customer is going to sort of be providing some of the working capital? How do we think about that?

Yes. I think what we meant by that is we've been staging and we talked about waiting to do the next phase of development until we had a contract and we were going to execute. So Fermi marched ahead early on. We secured assets, which has given us the competitive advantage for speed to power. We developed the site in such a way that our time lines could match up well, meaning power would no longer be the timing factor in being able to stand up and deliver to the customers. That's what we're talking about. I think each project itself then has capital deployment required with it. We'll look at various funding sources to do that, the normal ones. And so we're just talking about as we commence projects, we will then put all the capital behind to build out that specific project and keep Project Matador on track.

Your next question is coming from Paul Golding from Macquarie.