Iren Limited Aktie

Good day, and thank you for standing by. Welcome to IREN FY 2026 Results Call. [Operator Instructions] Please be advised that today's conference is being recorded.

I'd now like to hand the conference over to your first speaker today, Mike Power, Vice President, Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, and welcome to IREN's FY 2026 results presentation. I am Mike Power, VP of Investor Relations. And with me on the call today are Daniel Roberts, Co-Founder and Co-CEO; Anthony Lewis, CFO; and Kent Draper, Chief Commercial Officer. Before we begin, please note that this call is being webcast live with a presentation for those dialed in by phone. You can elect to ask a question through the moderator after our prepared remarks.

I would like to remind everyone that certain statements made during this call may constitute forward-looking statements. Those statements are based on current expectations and assumptions and are subject to risks and uncertainties that could cause actual results to differ materially. Please refer to Slide 2 of the accompanying presentation and our SEC filings for more information in that regard.

With that, I will turn the call over to Daniel Roberts.

Thanks, Mike, and thanks, everyone, for joining us. So Will and I started this business on a pretty simple observation. The digital world scales almost instantly the physical world does not. Power, land, data centers, these things take years to permit, finance and build. And this was the year, stopped being a thesis and became the defining constraint of the whole industry.

So if we look at the chart on the screen, across 8 models tracked by OpenRouter, weekly token usage across large language models increased nearly 17x in 8 months. Every one of those tokens runs on physical infrastructure. It is very difficult to serve demand shape like that with infrastructure on 3-year lead times and this is the start of the cycle, not the end. Cheaper compute doesn't make existing things more efficient. It makes new things economic, infrastructure enables applications, applications create demand for more infrastructure. Every build-out in history has worked this way. And that's the structural disconnect and it's only getting wider. So let me walk through how we're set up against that backdrop.

So we operate across 3 layers from the bottom up. So first of all, the data centers, the land, the power, the substations, the cooling. Arguably the hardest layer to build, and that's where the shortage begins. Then the compute, the GPUs, storage, networking that go inside the data centers. And then finally, software on top, to Managed Services and Enterprise Support. That's where Mirantis lives for us. And just today, Mirantis was named an inaugural NVIDIA-certified hypervisor. So we've now got NVIDIA validation at the software layer as well as the hardware.

Why own or free? Because each layer makes the one underneath it worth more. A grid connection is worth more with a data center on it. worth more again with GPUs inside, more again with services wrapped around the customer. Most of this market rents at least one of those layers, we own the entire stack.

So here's how the year went. Let's move on to highlights. Please, operator. Thank you. Four things to take away from this update, and I'll be quick because there's more detail coming a little bit later. So firstly, customers, new multiyear cloud contracts, including [ Cohere, Prometheus, Perplexity, Figure AI, [indiscernible] AI, Kingsfield AI ] and separately, a leading Frontier AI lab whose name we're not able to disclose just yet.

Revenue, $4 billion of ARR is now contracted for our 2026 capacity and $1 billion of that is operating today. This does not include revenue expected to ramp in 2027, such as the $700 million of ARR associated with our NVIDIA cloud contract. Delivering Horizon 1 was delivered to Microsoft this month, the first of 4 50-megawatt deployments with Horizon 2 through 4 targeted for the December quarter. And finally, funded, $6.5 billion of GPU financing now in the past 3 months, with prepayments, that's more than 100% of the associated GPU CapEx funded. And $2.8 billion of it needed no investment-grade offtake and still priced in the single digits.

So let me start with customers because everything else follows from them. Our 2026 capacity is largely sold out. So the questions we get now are all about 2027 and 2028. We're continuing to contract future capacity deliberately. Every contract opportunity gets weighed on 3 things. Firstly, who and what does this counterparty add to the platform, the strategic merit, not just the revenue. Second, what are the economics, price, prepayment, term, et cetera. And thirdly, what might it open up longer term for managed services and software. We've been saying this for a while now. Signing deals is not the bottleneck in this market, bringing GPUs online is. We also don't need an investment-grade offtake to fund GPUs anymore. So we're not chasing headline announcements, we're making long-term decisions about where we want this business to be and when we'll sign, we'll tell you. We're in late-stage discussions with a range of new customers over a significant portion of 2027 capacity and 2028 conversations are well underway to bolt-on customers, and financing.

Longer term, we want to keep building a deep, diversified base of customers across the AI ecosystem, all who have master service agreements with IREN. Every megawatt we build is scarce. Before every new cluster switches on, we want it in demand from both existing customers and new ones. So not dependent on any one customer, any one negotiation or any one point in a pricing cycle. And here's who's on the platform today.

So as I mentioned earlier, the headline is the new multiyear contract with a leading Frontier AI lab. And to be clear, this is a new contract. It's separate from [ Prometheus ], who we can now name as the unnamed AI developer from our July announcement. The most sophisticated buyers of AI infrastructure in the world keep choosing us. It also shows the strategy is working. We started concentrated because big customers with immediate demand with the fastest path to scale. As the platform has grown, we've deliberately broadened hyperscalers, enterprises, AI developers, now frontier labs across both training and inference.

But honestly, the part we care about most is the third bullet point. Existing customers keep coming back, [ Together AI and Bioworks AI ] have both renewed and expanded. New logos are proving demand, whereas renewals continue to improve ongoing delivery and ongoing execution. Once we've deployed with the customer, we grow with them across sites, GPU generations and service levels. And in terms of who we sign in and growing with us, [ Prometheus and Figure ] are building products for the physical world, robotics, real-world automation and they're contracting our compute to do it. AI is moving well beyond chat. This is the thing we keep telling people, you cannot size this market of today's usage. And that is exactly why the market keeps getting caught structurally short of compute.

Now let's move on to pricing. So pricing has moved a lot. 3-year contract pricing is up about 125% since November. Five year is up about 70. Recent 3-year contracts are pricing in excess of $20 million per megawatt of IT load paying back the compute investment in around 2 years while active discussions are now at around $25 million per megawatt. Recent customer prepayments are funding 45% to 55% of the GPU CapEx.

What's behind that? The market's tightened, no question, but it's also who we're signing, how the deals are structured and what we attach on top in Managed Services, which Mirantis keeps expanding. Shorter duration and on-demand sit above that is further upside in due course. Revenue per megawatt is shorthand, by the way, we look at the whole return, price, term, prepayment utilization, customer quality and expansion opportunity. And to be clear about what we're not doing, we're not sitting on capacity to time a spot price. We allocate capacity to build the customer base and the platform as we see fit, the pricing follows that.

Okay. So from customer to what we're building. The targets roughly 300 megawatts of IT load delivered in 2026 and another 0.5 gigawatt in 2027. That will take the platform to around 1.2 gigawatts in 2027 of gross capacity, and we're continuing to build across Texas, British Columbia, Oklahoma, South Australia and Spain. And right now today, there are more than 4,000 people mobilized across our active sites. The best example of that today is Horizon 1. So as we announced earlier, Horizon 1 was delivered to Microsoft. First of 4 50-megawatt liquid cool deployments at Childress and it achieved NVIDIA exemplar cloud status on GB300 NVL72, which matters because it proves we can integrate and operate the full platform. Hardware, networking and software, not just build the shell.

The delivery matters, but the template matters more. Every Horizon uses the same design. The same supply chain, the same site team and each phase carries the lessons of the last. Horizon 2 is in the process of working towards commissioning. 3 and 4 are in late construction. So all 3 are targeting delivery in the December quarter. And that approach is running at every site.

So quickly on 2026. At Childress, retrofit work and GPU installs are running in parallel on the air cold calls. At Mackenzie, GPUs are being racked across the first 2 buildings. At Prince George, the air cord fleet is now fully commissioned and liquid cooled installation is underway. All of that remaining capacity is targeted for the December quarter. In 2027, Sweetwater 1 is in full swing now. First building is going up. The primary substation is progressing. At Mirantis, Horizons 5 and 6 civils are now moving and underway and another 250 megawatts of vehicle conversion progressing. And it's also worth mentioning in terms of canal flats, we've now decided to convert all of that to liquid cooling for [ GB 300s ], and that will deliver more value from power and a site we already own.

But beyond that, the pipeline steps up again. 2028, Sweetwater 2 and [indiscernible] in the U.S., funding in South Australia and [indiscernible] in Spain roughly 300 megawatts and the flagship of the Nostrum portfolio of sites we closed during the quarter. All up more than 5 gigawatts announced with a multi-gigawatt development pipeline behind it, including more Texas power were a tighter interconnection process should favor real and well-capitalized projects.

And then a quick word on design because it answers the question, we're starting to get a lot. So the bear case we hear on this industry is that AI data centers get rebuilt in 10 years. We've spent this year making sure that ours don't. Sweetwater 1 is becoming the reference design for Sweetwater 2, Kiowa, [ Bundy ] and what follows after those. Common layouts, common equipment, more modularization and prefer. Each project is inherited in the last one's lessons instead of starting from zero and the design is built for successive GPU generations evolving calling, including 800-volt DC, all of which has been developed in collaboration with NVIDIA.

So we know compute change is faster than buildings and ours have been designed to adapt to that. But we're also getting more out of what we already own. Our existing sites have spare power beyond current deployment plans. That headroom can support more GPUs and more revenue without the need for any new grid capacity. And new group capacity is the scarcest input in this entire industry. So revenue that doesn't need is about the highest quality growth there is.

First up, new liquid called installs at Mackenzie, Canal Flats in Prince George in 2027. And over time, tools like NVIDIA MAX LPS which smooths GPU power draw, let us simply run more compute inside the same electrical envelope. But none of this happens without people. Our head count nearly tripled in FY '26, including hundreds of colleagues who joined through Mirantis and Nostrum, and we expect similar growth again in FY '27. 5 C-suite appointments across development, product, marketing, innovation and information security, people from NVIDIA, AWS, Oracle, Google, and other leading dots into operators. So building our organization ahead of the revenue is obviously deliberate. It costs money before it makes money, but this is where we're going.

One last piece before I pass off to Anthony is how we're funding this growth. So GPU financing first because now the model is proven at both ends of the credit spectrum. For the Microsoft contract, we raised $3.6 billion of investment-grade GPU financing at a weighted average of about 6%. With customer prepayments that funded about 96% of the associated GPU CapEx. Now here's where it gets really interesting for non-investment-grade deployments, so the rest of the AI ecosystem we just closed $2.8 billion of equipment financing. That includes $2.4 billion at a 9% fixed rate for Mackenzie led by [ Blue Hour ] and funds managed by [ Kimco ] a delayed draw term loan alongside senior secured notes, funding 90% of that GPU CapEx.

Ad prepayments of 45% to 55% in on recent deals and total funding well exceeds the cost of the underlying GPUs, which is good because that excess is now helping to support data center CapEx on those same and future deployments. And to date, talking about data center CapEx, we have deliberately kept 100% of our data centers unencumbered which is a growing asset base, we can finance when the timing is right.

So on that note, Anthony will take you how that funds the plan and the results. Thank you, Anthony.

Thanks, Dan, and good evening, everyone. Over the past 12 months, we have secured circa $19 billion in funding nearly $16 billion across customer prepayments, GPU financing and convertible notes alongside equity of approximately $3 billion. The vast majority of this funding is either in cash or yet to be drawn down, giving us significant capacity. For FY '27, we're guiding CapEx of approximately $25 billion to $30 billion. Included in that estimate is delivery of the contracted Microsoft capacity, the other deployments to deliver on 2026 ARR and GPU and data center CapEx for air [ call ] deployments scheduled across calendar year 2027. We will also support new liquid-cooled data center capacity at Childress and Sweetwater 1 for delivery in the second half of calendar year 2027 as well as earlier stage investment for 2028 and beyond.

Of course, the actual CapEx for the year will depend on a range of factors, including final costings, construction schedules, delivery time lines for GPUs and long lead items the overall contracting environment and the overall fundraising environment. Within that CapEx estimate, we expect data center and CapEx GPU requirements to be up approximately 15% to 20% for on going and new deployments with revenue increases expected to ultimately outpace those increases.

In respect to the funding plan, as noted earlier, we have a strong starting position, approximately $14 billion of existing cash and committed GPU financing and prepayments. That includes $7.6 billion of cash on the balance sheet at 30 June, of which $1.7 billion is restricted with most of that $1.7 billion set aside to fund Microsoft GPU CapEx. We're targeting roughly an additional $8 billion of GPU financing and prepayments in support of GPU CapEx requirements, noting the healthy prepayments that we are seeing in recent contracting and the growing market for GPU financing that Dan has spoken to. The balance of the requirement we expect to meet through data center financing, operating cash flows and corporate sources.

On data center financing, as Dan has spoken to, our entire data center portfolio is unencumbered today, including Horizons 1 through 4. So there is obviously a significant and growing asset base to raise asset finance against. Alongside the growing market for GPU financing, attractive customer prepayments, we feel well placed to scale our fundraising efforts to support the rapid growth of the platform.

Now turning to the quarter 4 results and outlook. The quarter's results continue to reflect the ongoing transition to AI Cloud. For the June quarter, revenue was $137.2 million, including AI Cloud revenue of $70.5 million. This was down $7.6 million compared to the prior quarter as we decommissioned mining hardware ahead of GPU installations, partially offset by AI Cloud growth.

Cost of revenue also fell $6.6 million, mainly as a result of lower electricity usage from reduced mining activities. Net loss was $684 million for the quarter, largely driven by noncash impairments of $450.4 million, mostly on account of decommissioning mining hardware, plus a $102.1 million decrease in fair value of mining hardware held for sale. Both reflecting costs of transitioning our existing sites to AI Cloud. We currently expect mining operations to be effectively decommissioned by the end of December 2026.

We expect first quarter cash SG&A to increase approximately $40 million to $50 million sequentially as we continue to invest for growth across sales and marketing R&D, development, size and cloud operations and other functions ahead of significant revenue growth over the coming periods, which brings me to ARR.

We exited Q4 at roughly $0.5 billion of ARR. It's $1 billion today following acceptance of Horizon 1 by Microsoft, and that will carry through to the end of the September quarter. And we expect more than $4 billion of ARR by the end of the December quarter, which is already under contract and includes delivery of Horizons 2 through 4. A significant amount of the December capacity is expected to come on late in the quarter, so we will see the reported revenue effect come through predominantly in the March quarter.

With that, I'll hand back to Dan.

Thanks, Anthony. So just one slide to close on. 2026 is largely sold out with the $4 billion of ARR now contracted. 2027 and beyond is deliver a runway for us. So this is capacity we're continuing to convert into a broader customer base and, of course, a richer service mix. Revenue per megawatt is rising, both on structural demand as well as site optimization, and that $4 billion comes from less than 10% of our 5 gigawatt plus portfolio of group secured grid connections.

So 8 years ago, we set out sold the gap between digital demand and physical supply. I think it's fair to say the gap is here. We own the power, we own the land, we own the data centers, we own the compute, we own the people -- the software, and we have the people to service it. So for Will and I, this is only the beginning.

But operator, let's open the line for questions, please.

[Operator Instructions] And this question comes from the line of Mike Ng from Goldman Sachs.

I just have 2 questions. First, I was just wondering if you could talk a little bit more about Mirantis and how it helps facilitate the commercialization of your compute was Mirantis used or helped to enable any of the signed or in negotiation deals today do you think you'll eventually use Mirantis to help sell some capacity into market aggregators like open router? And then I have a quick follow-up.

Yes. Happy to jump in there and thanks for the question. In terms of the Mirantis service offering, I think there's a few elements to that help with our business today. Firstly, as you identified, it opens up effectively new customer classes to us. As we've spoken about at length previously very large hyperscaler or frontier AI lab customers generally like to take compute on a bare metal basis. but smaller AI developers, labs, enterprise customers value having an orchestration layout offered over the top of bare metal compute. And so the Mirantis service offering as it relates to that orchestration layer can help open up that part of the market to us. In addition to that, it does give us the ability to provide compute in a different manner. So not only can we provide reserved managed services clusters but it also opens up the potential for us to provide on-demand compute as an example.

In addition to the orchestration layer itself, Mirantis has a long track record of servicing enterprise customers in the cloud computing space, and they bring a number of other areas of expertise, including enterprise support, monitoring and deployment capabilities that can help us monetize our platform more quickly and continue to provide very high levels of customer service. So those are a few of the areas where Mirantis adds to our existing platform.

Sorry, maybe just to add to that. I think we're now the only NEO cloud-certified hypervisor from NVIDIA as a result of today's announcement. So that kind of vertical integration that full stack is playing out lifetime.

Thank you, Daniel. And I just wanted to follow up on the CapEx outlook for next year, $25 billion to $30 billion, is that all to support the 800 megawatts that you expect to contract next year? Or is it beyond next year? And then could you maybe just talk about the financing plan beyond the $14 billion of cash, GPU prepayment and other debt financing that I think you talked about?

Sure. Just in terms of the -- and thanks for the question. Just in terms of the first question, so the $25 billion to $30 billion is obviously in the -- covering the financial year to June '27. So that covers all the CapEx requirements for the 2026 deployment to -- that contributed towards the $4 billion ARR target. It covers expected data center CapEx and GPU CapEx for sort of air cooled -- the balance of the cooled deployments expected to come over the course of 2027 calendar year. And it also covers a significant portion of the -- of CapEx required for the liquid cool deployments in the second half of 2027 calendar year. It doesn't include CapEx requirements for the GPU compute for those new liquid cooled facilities, which will be part of the following year's capital plan.

In terms of financing, obviously, we spoke to the $14 billion, so that's secured. I also spoke to $8 billion estimated additional GPU related CapEx. So that will be supporting the GPU deployments included in that $25 billion to $30 billion, obviously, we've seen a very strong fundraising prepayment environment which obviously contributes to that $8 billion number. And we've seen strong conditions, obviously, in GPU financing. Obviously, Dan spoke to the results that we have seen in the recent transactions, but there's obviously also been other transactions in the market, both for investment grade and sub-investment grade clients. We've also had the recent announcement by NVIDIA alongside 6 of the biggest capital allocators in the world announcing JVs partnerships to support further financing of AI compute as an asset class.

In addition to that, we obviously have 1 to 4 substantially built and being delivered over the course of the year. That's obviously uncovered as well as other opportunities we're looking at both asset-backed and other group level initiatives as well.

And our next question comes from Paul Golding from Macquarie.

Congrats on all the tremendous progress. I wanted to first ask about the British Columbia liquid cooling pivot. And wondering if you could unpack that decisioning process a bit. It sounds like there's excess power available and there's obviously a revenue opportunity. But I was wondering how you were thinking about the relative return there? And if that was organic or from demand that you've seen or speculative? And then I have a follow-up.

Primarily, Paul, it's driven by demand that we're seeing. And we do have a design that is very effective for the conditions that we see in British Columbia for liquid cooling. It is something that increasingly as customers are looking towards future workloads. We do see the vast majority of customers over time looking towards liquid cooled GPUs. And so we are certainly reflecting that in terms of the mix within our portfolio.

Great. And then Anthony mentioned that the compute costs that are rising are being more than covered by revenue increases. I was wondering if given the continued march upwards of compute cost, what the calculus is and your thought process and strategy around taking some of the excess liquidity from prepayments and the financing that you've done and allocating that towards data center build-out versus maybe longer-dated compute purchasing given -- and maybe on a speculative basis, given that price trajectory we've seen in compute.

Yes. I think in short, we're doing both. So as Dan mentioned, this is -- you need all the layers in the stack in order to be able to sell compute. You have to have the data center capacity. And we continue to see extremely strong demand within the industry and a lack of capacity available to satisfy that demand. And so we will continue to build out data center space. We will continue to make compute purchases over time and continue to add that software layer over the top to be able to expand our addressable market and secure better economics over time.

So we expect to see a mix of purchases on spec as well as some purchases that may be tied to customer contracts. But as Dan said, the contracts are not the driver in this industry. It's getting the compute online. And so that is what we are heavily focused on, on the execution side of the business.

Paul, I might also just add to your CapEx question around that revenue. It adds a little bit more to what Anthony said earlier. So if we step back and look at this revenue CapEx trade-off, the contracts that we're announcing now and the pricing per megawatt relates to GPUs that were ordered quite some time ago. So arguably, they're reflective of natural price increases beyond CapEx inflation on those GPUs.

Now future generations of GPUs and costs are going up, but we didn't have to pay the cost increases on the revenue per megawatt that we've announced today necessarily because they're the older generations, if that makes sense. But there's also a bit more nuance to our CapEx needs. So yes, we've guided total CapEx of $25 billion to $30 billion for FY '27. But that we deliberately don't split it in or attribute that to a specific number of megawatts of IT load coming online because the reality is that data center CapEx is running a year or 2 ahead of delivery, but that's the nature of substations and steel. So the mix in any given year is really a function of what we're delivering that year versus what we're building for the next one.

So I think we are working on how we stagger data center CapEx and how we finance data center CapEx, as we mentioned during the presentation, 100% of our data center portfolio today is unencumbered, which creates a future opportunity. And when we're receiving prepayments equivalent to 50% of the GPUs that's roughly equivalent to 100% of the data centers because GPUs are roughly 2/3 of your fully loaded cost for the data centers plus the GPUs. And if you're financing the GPUs at 90% gearing already, then you can see how you've got this funding flywheel that's emerging that requires a little equity over time to finance it. But we think going to overlay our ambitions are growing, we're building at a higher cadence to what we've done in the past, so those funding needs continue to grow. But we've got flexibility around growth based on optimizing the funding that's available to us at any point in time.

That's great color, Dan. Maybe just a quick housekeeping question on the back of that, if I can sneak this one in. Just on the 2/3 compute versus data center CapEx, is that shifting at all with the increased cost of compute? Or is the labor market being tight and other inputs on the data center construction side keeping that equation relatively stable?

No. It's staying pretty stable. I mean we're seeing inflation and costs across various components continue to dig up. So at a material level, I would assume that ratio stays relatively consistent.

We will now move to our next question. And our next question comes from Brett Knoblauch from Cantor Fitzgerald.

Congrats in the quarter, great to see kind of demand around '26 and certainly '27 and '28. I'm curious kind of following into Horizon 1 build-out. We've seen a lot of deals recently with maybe lower PUEs, are you seeing any design changes that might allow for lower [ PUDs ] at other sites that you guys have coming online or other buildings coming online in '27 and '28 or should we kind of think of that [indiscernible] with the Microsoft deal being somewhat static?

I mean PUEs are never entirely static, but they are largely driven by the nature of the ambient conditions in which your data centers are operating. We've spoken before that we use a highly efficient, closed-loop liquid cooling system. And that means we have very effective PUEs relative to the broader industry. We do see some continued improvements over time and part of that is driven by items like infers, DSX, reference architecture, where they have looked at how you can operate chips at higher temperatures without giving up any performance on the computing side, but allowing you to direct more of your overall megawatts towards the IT load and away from cooling. But those are around the edges rather than being really material drivers of decreases over time.

I think just to add, we've been deliberately conservative on the headline PUEs that we've advertised. We even had a slide dedicated into the deck to this thematic, which is we have spare power beyond our current deployment plans and some relates to optimizing this PUE over time. The reality is the average PUE for a year is substantially below the maximum PUE that you require. So there's a few little tweaks that you can make that free up quite a lot of power capacity alone. We've then got the NVIDIA DSX MAX LPS opportunity and we're aware others are using power management tools to oversubscribe megawatts as well. There's a lot of flexibility within that envelope. And to date, we've kept it simple. 200 megawatts of IT load for 300 megawatts of gross capacity, but we are also making it clear today that there is a reasonably sized opportunity in the portfolio to free up some of that spare power.

Awesome. Maybe I could just follow up on some of the pricing commentary that came across quite strong, I think, in prepared remarks and the release, kind of $20 million deals or what you've signed now, it's kind of somewhere around $25 million. Is that -- over what duration could you clarify? And is that more one-off? Or do you think like that is kind of like the ballpark of maybe the average you're seeing across all the conversations you're having with customers?

No. We are seeing that consistently across live conversations with customers at the moment. And there are a variety of things that go into it. As Dan mentioned, we look at term length prepayments, nature of the customer likely growth requirements over time. But the pricing that we're seeing is relatively consistent at the moment. It continues to show an upward trend. We're seeing very strong competitive tension for near-term megawatts so we certainly think that those numbers that we put out are indicative of where the market is currently at with upward pressure over time.

And to be clear, there's 3 to 5-year deals, not 2 year, not spot capacity. There is a substantial opportunity if you wanted to optimize near-term quarterly P&L to go and sell spot on demand. And that is something that we might entertain as part of the portfolio approach over time. But today, we are loving us building the customer base, diversifying across the AI ecosystem having longer-term contracts, getting really accretive financing back in those contracts. And over time, that will buy us a license to double more into on-demand, shorter-term contracts introduce software leveraging Mirantis and other capabilities for higher revenue profiles.

Thank you. We will now take our next question from Michael Donovan from Compass Point.

Just go back to the $20 million, $25 million in revenue per IT megawatt, how do the economics you're seeing in the U.S. compare with what you're seeing with initial conversations in Spain and Australia?

I think very consistent. At the end of the day, this is largely a global market for compute. A number of the customers that are accessing compute out of North America are based overseas. And so it does drive a genuine global market for compute. And they may arguably even be additional scarcity factor in some of those other locations where it relates to things like sovereign AI, for example. So I think that pricing that we're seeing in North America is very indicative of global conditions.

And one more follow-up, if I may. I wanted to get clarification on the Nostrum acquisition. So some understanding it brought roughly 490 megawatts. In your prepared remarks, you mentioned [indiscernible] accounting for roughly 300 megawatts for that difference is the 190 megawatts across a couple of different sites? Or how should we think about that?

Yes. So there are a number of other development sites in the Spanish portfolio where we have capacity secured. But [indiscernible] key site that we chose to focus on today is that is the nearest term build-out and the largest site within the portfolio.

[Operator Instructions] Next, we have Nicholas Giles from B. Riley Securities.

Just wanted to ask one about Texas. I know it's not a fun topic, but I was just curious if you could touch on what some of these dynamics have enabled from a commercial perspective, just given that you already have 2 large-scale energized sites there.

And by the dynamics, are you referring to Governor Abbott's directive or something else specifically in there?

Yes, Kent, exactly just a directive and just given the uncertainty around some of the kind of earlier stage sites there if that might have pushed some potential deals more towards your corner.

Yes. I think as it relates to Governor Abbott's directive, if you look at what that is targeting, it looks at things like transparency grid reliability, water usage issues in terms of impact on local communities and these are all things that we've had a specific focus on from day 1 and have always been important to us. So if you take Childress, as an example, we funded all the required grid upgrades for that project. As you know, we located a number of miles outside of the main town area there where we don't have residential neighbors and we're not impacting neighbors or the local community from an operational perspective. We use a highly efficient closed-loop water cooling system, which has very minimal ongoing water usage over time.

So I think everything that we've done in setting up our sites in our portfolio is in line with what Governor Abbott came out within his directive. So I think in that sense, we actually welcome the additional transparency within the market. And then as you mentioned, we do have 2 very large sites already energized in Texas, which I think positions us extremely well outside of just those dynamics that I mentioned as we move forward here.

Kent, that's helpful. And then maybe just as a follow-up. We've seen a lot of deals where there's kind of an initial contracted capacity, but then there's an expansion option maybe with exclusivity for some period. And I can't recall IREN having with any of those expansion options embedded. Is that something that you're considering in future deals? Or have you kind of intentionally straight away from those options?

Yes. We've intentionally straight away from that historically, but it's obviously part of the bespoke conversation with each customer. As you would guess from our thesis, which Dan outlined earlier in this call, we see very significant value in infrastructure and compute moving forward. And so we're going to give up an option to somebody over that. We need to be compensated for it so it really just comes down to that dynamic. But yes, we've generally shied away from it previously because we see more value typically than the counterparty.

Thank you. We will now take our next question from Mike Colonnese from H.C. Wainwright & Co.

Congrats on all the strong momentum here across the AI Cloud business. Just one for me. I was hoping to get more color around how you plan to allocate future uncontracted capacity here. Really the specific customer segment you're most interested in pursuing heading into 2027? And how you guys are thinking about balancing pricing with financing costs when evaluating larger versus more customer contracts. It feels like you're trying to monetize a little bit higher on the stack with the acquisition here. Just trying to get a sense of how you guys are evaluating these deals.

Yes. I think Dan touched on a lot of the elements that we look at earlier in terms of the earnings overview. I mean we look to the type of customer, the industry they're in, their planned growth rates over time. As you would have seen with our customer mix, it has been shifting towards direct end users of compute. We now have added additional names in the AI native space, a large leading frontier AI lab as well as some of these physical infrastructure-related AI developers. So we see a lot of very attractive profiles across the sector.

And for us, as Dan mentioned, we've been aiming to diversify our customer base over time. We have a range of attractive long-term contracts within the portfolio that now set us up for very attractive economics as we move forward here. And that, in turn, buys us some additional flexibility to be able to look at different monetization methods, whether that is selling on a short-term basis in terms of reserve contracts that drive additional economics, whether it's looking at true delivery of on-demand compute whether it's managed services offerings versus bare metal.

And one of the nice things with the position that we're in today is we have optionality over all of those different areas and particularly with the development of the financing market for sub-investment grade counterparties, which is something that we always expected to develop, and we're now really seeing the proof of that. It means that we can still continue to get very attractive financing for a range of different customer types.

So yes, we're certainly very excited at the prospects as we move forward here, being able to monetize the platform in different ways and get additional economics over time.

We will now take our next question from Ben Sommers from BTIG.

I wanted to ask a bit on the conversations we're having for 2027 and 2028. It makes sense that we're maybe not exploring something like the really short duration stuff now. But as you think about what you're hearing from customers in terms of the window from 3- to 5-year contracts kind of where are you seeing most customers heading from within that time range? And what's the ideal if there's any time length in your guys' perspective?

I'll jump in Ken, I think we're starting to see longer-term conversations -- or sorry, conversations around longer term than just the 3 to 5 years. But this is the balance, right? We're running it essentially as a portfolio. Anchor tenants, like Microsoft gives you the scale, the duration and cheap capital with the 6% GPU financing whereas AI developers in the broader market give you slightly higher pricing, good prepayments and a lot of weather growth is. So we want to remain flexible. At the end of the day, if the framework for assessing new customer contracts is as we have outlined on this call. It's who and what does the counterparty add what are the economics and what does it open up longer term? And yes, there's a trade-off. The investment-grade [indiscernible] gets a 6% money, non-investment grade gets 9%. So it sounds like the investment grade wins on that until you start looking at the pricing, but still you start looking at the prepayments funding around 50% of the GPU CapEx.

And I think those prepayments are probably the most exciting part for us, when they're funding half the GPU CapEx upfront, on top of the 90% financing we're getting already, these guys are sending a pretty clear signal. It's not just about contracting capacity. They're starting to finance our build-out for us. And I think that tells you a lot about demand more than any pricing chart in the presentation.

Super helpful. And then just one more, if I could. If you could just give any color on preliminary conversations around potential data center financing? I know you guys talked about potentially pursuing that down the road. So just want to ask around any preliminary conversations you've had there?

Yes, lots of preliminary conversations, and we'll let you know when we close one. But we've been busy. There's lots going on, on the customer side, lots been going on in the GPU financing, that's been the lower-hanging fruit. We've previously mentioned, we'll look at refinancing effectively horizons as they're commissioned as it makes sense to make weight till they're stabilized, to get a better financing package. We'll look to finance data centers as they're commissioned, but potentially also in advance of commissioning.

And one of the challenges we have and the opportunity associated with being vertically integrated, is you've got this staggered GPU CapEx, which starts a couple of years out from commissioning -- sorry, not GPU data center CapEx, which starts a couple of years out from commissioning because you need to order the steel, you need to order the transformers, you need to order a whole heap of long items. And yes, a lot of those payment milestones are back ended, but CapEx is incurred in the lead-up to commissioning and find efficient ways of financing that to allow us to scale into that 5 gigawatts really quickly, let alone the multiples of that sitting behind it in our pipeline. Like that's the big unlock for us. Like the customers, they're there, like I don't think anyone is disputing that anymore.

So for us, it's all about that capital flywheel and managing it efficiently to get us in a position where we can meet that market demand because we are in a really, really unique position because we started 8 years ago, we locked up all the land and power. We've aggregated the team, the expertise, the partnership and collaboration with NVIDIA to build and operate these things. And capital efficiency is a really big part of the next unlock. And GPU CapEx efficiency is now here, like objectively the data we've published today, what we've closed, that is efficient, and it will only get better. In terms of the data centers that's the next frontier, and we're pursuing it, and we're excited about it.

Will now take our next question from Joseph Vafi from Canaccord Genuity.

Congrats on all the progress. I know, Dan, you commented, obviously, power remains a major constraint. We're on this call. We're having a lot of discussions on financing. Obviously, you've got a lot of resources available at your disposal, but it's a big industry and a lot of people are doing a lot of financings out there right now as well.

Just some high-level thoughts, maybe, Anthony or Dan, on the financing environment, sustainability of the industry to continue to finance this broader build-out at this pace, if there's anything that you're worried about there, how iron may have some advantages given it's different pieces to the business. And then a quick follow-up.

Thanks for the question. I guess we've touched on a few of the sort of fundamentals at the moment. I guess we're seeing if the evolution of the market has been -- on the GPU financing, we've seen the -- it started as a private credit led product that sort of mid-teens returns and investment-grade cost of capital now is sort of in that 6% area. We've obviously seen the market open up in terms of the pools of capital participating in GPU financing, so private public markets, investment grade, subinvestment grade. That's all positive.

We've obviously, as I touched on, also had the other considerations such as the big announcement by NVIDIA with 6 of the biggest global allocators which will provide more support to the market. We've also spoken about prepayments, which are obviously a big part of the overall funding position as well. We've obviously also seen the data center financing market evolve as well to support the infrastructure to build out, both in the high-yield and the investment-grade space.

And obviously, there's obviously very strong conviction in capital circles in terms of the sustainability outlook for, I guess, this fundamental shift that we've had in technology and the investment that will -- that needs to be made and also conviction in the returns that will ultimately come. So I guess it's really that conviction, which I think will attract capital and continue to attract capital. Obviously, we need to continue to evolve our plans subject to market conditions. And obviously, market conditions can go up and down. And obviously, we need to be nimble and flexible in terms of how we adapt to that environment and adjust our plans accordingly.

We hear this question a bit, Joe, which is can the financing keep pace with the asset class. But let's just look at what happened 12 months ago, GP financing barely existed as an asset class. And in the last 3 months, we've raised $6.5 billion of it at both ends of the credit spectrum. So that's not us getting lucky with financing, that's a market for me. The markets -- they fall the same way every time. Think about real estate financing. No one asks whether an office tower with a signed tenant can get a mortgage. There's a whole capital stack that exists for construction finance, term debt, institutional money because the cash flow is contracted, the collateral is real. And that stack took years, probably a decade to build for a property, and it's building the AI infrastructure in months and quarters because the same ingredients are there, the hard assets, the contracted revenue, the institutional counterparty.

[ Blue Hour and PIMCO ], like they're not bit part players. They're the largest infrastructure lenders in the world and they've now underwritten us and we've got a dialogue at the highest levels there. But I think the analogy actually undersells it a bit because a building leased for a few decades at a few percent yield like our contracts are paying back this compute investment in around 2 years with customers prepaying half the CapEx upfront. So lenders can see their money back in really short time frames. And probably finance never had economics like that.

So I think whenever there's a new emerging market, like if the demand is real, the financing follows, it always has, whether it's for railway, for telecoms for property, for power. And in terms of whether demand is real, I think the customers are proving it. They're proving with prepayments. They're proving it with their end markets and their results and their revenue and their customer market traction. So the layer that hasn't really started and developed as much as the GPUs and the data centers. But in traditional spec, the data centers are actually the easier ones because they're more closely aligned to property in real world infrastructure, and there is cash there. We know that, and it's waiting for us to pull the trigger and we'll look more closely at that over the coming months.

Thank you. That was our last question for today. I would now like to turn the conference back to Dan for his closing comments.

Thanks, operator. Thank you, everyone, for joining. So the short version of FY '26, $4 billion of ARR contracted, $1 billion is operating today. Horizon 1 has been delivered to Microsoft. And as we've just touched on extensively a funding model that's working efficiently at both ends of the [indiscernible] spectrum. So all of this achieved on a fraction of the platform that we've got today.

And I guess, shout out to the IREN team around the world and everyone who has joined us in recent months as Mirantis, Nostrum a variety of broader players in the market, keep doing what you're doing, the plan doesn't change. We keep delivering capacity we'll keep converting into durable customer relationships. And we'll keep funding it with discipline. Thanks, everyone. We'll see you at the next result.

Participation in today's conference, this does conclude the program. You may now disconnect.

Good day, and thank you for standing by. Welcome to IREN Q3 FY '26 Results. [Operator Instructions] Please be advised that today's conference is being recorded.

I would now like to hand the conference over to your first speaker today, Mike Power, Vice President of Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, and welcome to IREN's Q3 FY 2026 Results Presentation, and thank you for your patience as we get assembled. I'm Mike Power, VP of Investor Relations. And with me on the call today are Daniel Roberts, Co-Founder and Co-CEO; Anthony Lewis, CFO; and Kent Draper, Chief Commercial Officer.

Before we begin, please note that this call is being webcast live with an accompanying presentation. For those dialed in by phone, you can elect to ask a question through the moderator after our prepared remarks. I would like to remind everyone that certain statements made during this call may constitute forward-looking statements. Those statements are based on current expectations and assumptions, and are subject to risks and uncertainties that could cause actual results to differ materially. Please refer to Slide 2 of the accompanying presentation and our SEC filings for more information.

During today's call, we will also refer to certain non-GAAP financial measures. As a reminder, a reconciliation to the most directly comparable GAAP measures is included at the end of the presentation.

So with that, I will turn the call over to Dan Roberts.

Thanks, Mike, and thank you, everyone, for joining us today. 8 years ago, when Will and I founded this business, we spent a lot of time thinking about what the digital future actually meant for the physical world. We talked about films like the Matrix and Ready Player One, not as science fiction, but as a signal. Worlds where digital adoption was total, instantaneous and infinite.

The insight we kept coming back to was this. Digital adoption curves can go from 0 to 1 overnight, but the real world doesn't scale that way. Power, infrastructure, land, data centers, these take years to permit, finance and build. The bigger the demand, the harder delivery becomes. That gap between exponential digital growth and the physical world's ability to service it, that structural disconnect is exactly what we set out to solve. That scarcity is now defining where AI infrastructure gets built and who can build it.

8 years later, that thesis is playing out exactly. And this quarter, we demonstrated what disciplined execution against it looks like at a global scale. In AI infrastructure, secured power is only valuable if it can be converted into customer-ready compute. That conversion is hard. It requires site control, grid connection work, permitting, design, procurement, construction, GPU installation, networking, commissioning, financing and customer delivery, all coming together on tight time lines.

IREN's strength is bringing those pieces together. We have experienced site teams, standardized designs and repeatable construction processes that allow us to build across multiple sites in parallel. As we scale, each phase builds on the prior phase. The template becomes more repeatable, the procurement and construction process becomes more efficient, and the site teams carry that experience forward. That is where IREN has built its moat, and why real assets and real capabilities are harder to replicate than they might appear. That execution capability is showing up in the numbers, more capacity, more revenue, stronger funding certainty, and in the partnerships we are announcing today. This was a significant quarter and a significant week. Let me run through the highlights.

On capacity, we increased secured power to 5 gigawatts, added new sites in Europe and APAC, energized Sweetwater 1 on schedule, and have Horizon 1 GPU commissioning now underway for Microsoft. On customers, all of our operational capacity is fully contracted. We are not chasing demand. We are racing to build supply fast enough to meet it. In this market, the moment compute comes online, it goes to work. That is the nature of the structural imbalance between AI infrastructure supply and demand, and it is why time to compute is the most important metric we track.

We increased ARR under contract to $3.1 billion, remain on track to hit $3.7 billion exiting calendar 2026. And this week, signed a $3.4 billion 5-year AI cloud contract with NVIDIA. The first step in a broader strategic partnership I will come to in a moment. On capital, we had $2.6 billion of cash at April 30, and we continue to progress GPU, data center and corporate level financing initiatives to support the next phase of build-out. But the headline today is the NVIDIA partnership, and it deserves a little more than a bullet point.

Let me explain what this partnership actually means. We are working with NVIDIA to support deployment of up to 5 gigawatts of NVIDIA DGX-aligned AI infrastructure across our global data center platform, alongside DGX environments and the DSX AI Factory reference architecture. The $2.1 billion NVIDIA investment is structured to reflect that. Their rights to invest only vest as NVIDIA GPU infrastructure is deployed across IREN campuses, and only fully vest upon deployment of 600,000 GPUs. NVIDIA's capital is directly tied to execution. That's not a passive financial investment. NVIDIA is a partner who wins as we deliver.

The $3.4 billion AI cloud contract announced today, supporting NVIDIA's own internal workloads is the first step in that partnership.

8 years ago, Will and I set out to build the infrastructure the digital world would need. Today, the world's leading AI infrastructure company has chosen IREN as the partner to help build it. This next slide shows exactly how we build against this.

So here is our plan. In 2026, we are targeting 480 megawatts of AI cloud capacity, 150,000 GPUs and $3.7 billion of ARR by year-end. That is the near-term plan and the clearest bridge from capacity to revenue. In 2027, we are scaling to 1,210 megawatts, with an additional 730 megawatts currently under construction across British Columbia and Texas, including Childress and the initial phase of Sweetwater 1. The construction flywheel we are running in 2026 carries directly into this next phase.

Beyond 2027, we are building against a 5 gigawatt global power portfolio. North America, our new European platform in Spain and an APAC pipeline anchored by large-scale Australian opportunities. The sequence of delivery matters because it dictates time to compute, and time to compute is what drives revenue. Each phase supports the next, that's how the platform compounds.

One more thing before we move on. This week, we welcome Mirantis into the IREN family. 650 engineers, operators and customer support professionals who have spent more than a decade running cloud infrastructure for over 1,500 enterprise customers globally. To Alex and the whole Mirantis team, welcome. I'll come back to what this means for our delivery capability later on.

But let me start with 2026, where construction and customer demand are coming together most visibly. The 2026 expansion is focused on delivering 480 megawatts of AI cloud capacity across Childress, Prince George and Mackenzie. This is where the road map translates into near-term deployments, customer handoffs and ARR conversion. We'll start with the largest and most complex 2026 work stream, the 300-megawatt Horizon 1 to 4 liquid cooled deployment at Childress, where NVIDIA GB300 NVL72 installations are now underway.

Horizon 1 is scheduled for Microsoft handoff in Q3, and Horizons 2 to 4 remain on track for delivery by the end of this year. This is a major execution milestone. It demonstrates our ability to design, build, fit out and commission large-scale next-generation liquid cooled infrastructure for a hyperscale customer on an accelerated schedule. We have around 3,000 workers on site right now. That level of activity reflects both the urgency of AI infrastructure demand, and also the depth of our execution capability on the ground.

Importantly, the model is repeatable. Horizon 1 establishes the build template. Each subsequent phase benefits from the same design, supply chain, construction sequencing and site team. That is how we drive faster deployment phase every time. Alongside the liquid cooled build, we are also converting existing air cooled capacity into AI cloud deployments across British Columbia and Childress.

In British Columbia and Childress, we are progressing 180 megawatts of air cooled AI cloud capacity by leveraging existing infrastructure. At Prince George, all air-cooled GPUs have now been delivered and are either operating, or undergoing commissioning across the 50-megawatt site. At Mackenzie, 80 megawatts of data center capacity has been prepared for GPU installations commencing in the second half of 2026. And finally, at Childress, data center retrofits are underway across an initial 50 megawatts ahead of GPU deliveries in the second half of this year.

This is a capital-efficient part of the road map. We are taking existing sites and converting them toward higher-value AI cloud workloads. And it works because we already have the operational teams, infrastructure and site control in place. Air cooled capacity can come online faster than liquid cool. In a market where time to compute is everything, that speed is a commercial advantage, and we are using it.

We're already seeing this dynamic play out commercially with capacity continuing to be contracted ahead of commissioning as customers prioritize speed to market. We now have $3.1 billion of ARR under contract, including approximately $700 million of ARR associated with a $3.4 billion 5-year contract for Blackwell GPUs to be deployed across 60 megawatts of air cooled capacity at Childress for NVIDIA. Against the full 2026 expansion, we are targeting $3.7 billion of ARR by year-end across 150,000 GPUs. The remaining uncontracted capacity represents approximately 50,000 air cooled GPUs scheduled for delivery in phases through the second half of this year. Demand for that capacity is robust. Our focus is on using our time to compute advantage to secure the right customer mix.

With the 2026 plan on track, let me turn to what comes next. The 2027 expansion where the pipeline -- the platform rather scales to 1,210 megawatts. So the 2027 plan is about demonstrating that we are -- that what we are building in 2026 is not a one-off. It is a repeatable, scalable model that should accelerate over time. Here's what that looks like in practice.

In British Columbia, Canal Flats is another example of converting existing infrastructure into AI cloud capacity. We plan to retrofit all 30 megawatts of existing air cooled capacity to support AI workloads. Capital efficient, fast to execute and consistent with the same model we are running at Prince George and Mackenzie.

In parallel, Childress continues to be the largest single contributor to the 2027 setup, with both new liquids cooled capacity and additional air cooled retrofits adding a total of 400 megawatts of gross capacity. At Childress, the 2027 plan includes 100 megawatts of additional liquid cooled IT load for Horizons 5 and 6, as well as retrofitting an additional 250 megawatts of existing air cooled capacity. Of that 250 megawatts, approximately 60 megawatts will be deployed to support the NVIDIA AI cloud contract. The combination of new liquid cooled data centers and air cooled retrofits gives us real flexibility. We can support next-generation high-density deployments while continuing to use existing infrastructure where it is the right technical and economic fit. That flexibility is part of what makes Childress such a productive campus.

In parallel, Sweetwater becomes the next major Texas campus in the 2027 plan. At Sweetwater 1, the high-voltage substation has been energized on schedule, and construction is now underway for the initial 200-megawatt IT load phase of liquid cooled data centers. Energizing the substation is an important milestone. It moves Sweetwater from development into execution, and establishes the electrical foundation for the broader site build-out.

Sweetwater 1 is being designed for next-generation chip architectures, including the NVIDIA Vera Rubin. Like Childress, we are deliberately sequencing the build so that the first phase creates the backbone for faster subsequent phases. The first 200 megawatts is not just the first 200 megawatts. It is the foundation for a much larger site. The commercial pipeline for our 2027 capacity is anchored on the same principle that is driving everything we are building.

Our vertical integration is a genuine advantage for customers because we control more of the critical path than anyone else in this market. Power, land, data center construction, the pieces that cause delays for others are the pieces we own and control. Customers want certainty that capacity will be available when promised. The phase 2027 build-out plan gives us a concrete basis for those conversations, and we are having them. We are in the process of negotiating large-scale AI cloud deployments across our 2027 capacity today.

Demand is not the constraint, however. It is highly unlikely to be the constraint. The priority is delivering capacity on schedule and converting our time to compute advantage into durable long-term customer relationships. We do expect the customer mix to evolve over time. hyperscalers, AI natives, enterprises and on-demand use cases, but we do not need to force that outcome. The platform will attract the right customers as it continues to scale.

Beyond 2027, the same execution model extends into a much larger 5-gigawatt global platform. We now have 5 gigawatts of secured power. To put that in context, that is not a pipeline number or an aspiration. That is secured power, and it represents one of the largest portfolios assembled for AI infrastructure anywhere in the world. The question now is how we build against it. The answer is a phased global platform across North America, Europe and APAC with additional development opportunities beyond that. Let me walk you through each region.

We'll start with North America, which remains the largest component of the long-term platform. In North America, the next major phase is driven by Sweetwater and Kiowa, our flagship gigawatt scale campuses in Texas and Oklahoma, where data center capacity is expected to commence ramping across 2027 and 2028. We also have multiple development projects advancing through the connection processes, including batch zero candidates in Texas, which represents some of the most strategic valuable interconnection opportunities in the country.

The North American pipeline has a natural progression of scale. Childress demonstrates the operating model today, Sweetwater expands it across an even larger campus and Kiowa provides the path to another hyperscale tier opportunity as power ramps from 2028. Every campus builds on the last. That's the compounded effect of having secured the right land and power positions early.

At the same time, we're expanding the platform into Europe through Spain. Today, we announced the acquisition of Nostrum Group and with it our entry into Europe. The transaction adds 490 megawatts of secured power in Spain, a gigawatt scale development pipeline and a team of more than 50 people across development, engineering, construction and operations. But what it really adds is a platform and the right people to build it.

I want to acknowledge Gabriel Nebreda and the Nostrum team. Gabriel spent nearly 2 decades in European energy at EDP Renewables managing gigawatts of operating assets across multiple European markets and most recently as CEO of EDP Solar. He understands European power infrastructure as well as anyone, and we are excited to have him leading IREN's European platform.

Spain is the right place to start. Supportive AI policy, abundant renewables, lower build costs and strong connectivity into broader European demand. Europe is a market where power availability and grid time lines are increasingly shaping where customers can actually deploy. And Spain gives us a credible scalable answer to that question. This is not just a power acquisition. It's the establishment of IREN's European platform.

From Europe, we move to the other side of the world and an opportunity that matches the scale of everything we've just described. Australia is obviously not a new idea for us. We have been progressing large-scale Australian projects towards secured grid access for some time. And we think the opportunity here is as significant as anywhere in our portfolio. And this is why.

Asia Pacific is home to roughly 4.8 billion people, around 60% of the world's population. That includes some of the fastest-growing AI demand markets on earth. Indonesia, Singapore, Japan, Korea, the infrastructure requirement to service that demand is enormous, and it is largely unmet. Australia is uniquely positioned to serve it, abundant renewables, a trusted jurisdiction, strong rule of law and as the submarine connectivity map shows, direct fiber links into major demand centers across the region. It is the natural anchor point for AI infrastructure service in APAC. We are already seeing hyperscalers and frontier labs make significant commitments to Australian operations, and we intend to be a major part of that story.

Beyond Australia, we continue to progress global development opportunities that extend IREN's runway further still. The platform we are building is designed to create scale into demand wherever it develops, and the pipeline gives us the flexibility to do exactly that. That is the global platform, secured power across North America, Europe and the development pipeline extending into APAC and beyond. But securing power and building data centers is only part of the equation. The other part is what happens when the compute goes live. How it is deployed, managed and supported for customers at scale? That is where I'd like to spend a moment on Mirantis.

This week, we welcomed Mirantis into the IREN family, and I want to take a moment to acknowledge that. 650 people joined IREN this week. Engineers, operators, customer support professionals, a team that has spent more than a decade building and running cloud infrastructure for over 1,500 enterprise customers globally. That track record speaks for itself. And what they bring is specific. Their k0rdent AI platform manages AI infrastructure across bare metal, virtual machines and Kubernetes environments, exactly the complexity our customers are dealing with as deployments scale. They are also a founding ISV partner of the NVIDIA AI Cloud-ready initiative, which means they are already deeply embedded in the same ecosystem we are building into.

As we scale, delivery is not just about bringing GPUs online. It is about what happens after, provisioning, monitoring, supporting customers through increasingly complex environments. Mirantis strengthens all of that. We are already seeing and they will play a central role in supporting our NVIDIA AI cloud contract. To Alex and the whole Mirantis team, a big welcome. We are super excited to have you. So what you have heard today is a company that has secured power at scale, is contracting revenue at scale and is now building delivery capability at global scale.

Anthony will now walk you through how we are funding it.

Thanks, Dan. The capital strategy is designed to support the phased build-out of capacity Dan discussed, while maintaining flexibility and capital discipline. As of April 30, we had $2.6 billion in cash and cash equivalents. We expect this, together with operating cash flows, GPU financing and additional financing initiatives to support our near-term CapEx program, which includes delivery of the Microsoft contract and deployment of aircool capacity across Mackenzie and Childress. For GPU CapEx, we are leveraging secured debt and customer prepayments. As we have noted previously, approximately 95% of Microsoft GPU-related CapEx is expected to be funded through prepayments and GPU financing. And we have work streams underway for additional GPU financing to support upcoming deployments.

On the data center side, we expect our financing approach to evolve as projects move from development to construction and contracting, and ultimately to stabilized operations. Early-stage development can be supported by balance sheet capacity and corporate level sources. As projects reach construction and customer contracting milestones, asset and project level financing can be introduced. And as assets are stabilized, refinancing and capital recycling can help support future builds. And as I note we will continue to maintain a disciplined balance of debt and equity as the platform continues to scale. I will now turn to the financial results, which continue to reflect the transition underway from Bitcoin mining to AI cloud.

Revenue was $144.8 million for the March quarter, compared to $184.7 million in the prior quarter. Within that, Bitcoin mining revenue was $111.2 million, down from $167.4 million, driven by a lower average Bitcoin price and the ongoing decommissioning of mining hardware ahead of GPU installations. This was partially offset by continued growth in AI cloud services revenue, which increased to $33.6 million, compared to $17.3 million in the prior quarter. Cost of revenues decreased by $25.9 million, primarily due to electricity costs from reduced Bitcoin mining capacity.

Net loss for the quarter was $247.8 million, impacted by noncash impairments of $140.4 million, primarily related to the decommissioning of mining hardware, as well as $23.7 million of unrealized losses related to cap calls associated with our convertible notes. As we continue to transition our remaining Bitcoin mining operations towards AI cloud, we expect to incur additional noncash impairments associated with decommissioning mining hardware. These outcomes reflect the strategic reallocation of infrastructure toward AI cloud growth, which we believe is the higher-value long-term opportunity.

Adjusted EBITDA was $59.5 million, compared to $75.3 million in the prior quarter, primarily on account of the revenue and cost of revenue items noted above. So as noted, the quarter reflects the ongoing transition from Bitcoin mining to growing AI cloud. As Dan noted earlier, we continue to target $3.7 billion in ARR by the end of calendar 2026. We expect that ramp to be back-end weighted with Microsoft revenue, and revenue from the additional 50,000 GPUs procured during the quarter expected to begin ramping in Q3 2026.

I will now turn back to Dan for closing remarks.

Thanks, Anthony. So 8 years ago, Will and I asked a simple question. What does the world need to build the right digital future? The answer was power, land, data centers and compute and the ability to bring them all together at scale faster than anyone else. Today, that thesis is playing out, and we are just getting started.

With that, we will open the call for Q&A.

[Operator Instructions] First, we have Mike Ng from Goldman Sachs.

Congratulations on all the progress. I just had two questions, if I could. First, on the 5-year NVIDIA AI cloud contract, I was just wondering if you could talk a little bit about how many GPUs are being supported by the 60 megawatts and the cost per GPU?

And then second, for Sweetwater and Oklahoma, I think you mentioned the data center capacity is coming in, in '27 and '28. I was just wondering if you could talk a little bit about like at what point do those sites become marketable, or maybe they already are? And what milestones do you typically need to hit to increase the likelihood of a tenant being willing to take that out?

Let me take that one, Dan. So with respect to your first question, we haven't disclosed the specific amount of GPUs. But as we mentioned on the call, approximately 60 megawatts of air cooled Blackwells. And we think that the contract value that we're getting and obviously, the relationship that we continue to build with NVIDIA is very beneficial coming out of that contract. Importantly, this is a managed services deployment. And so it shows our ability to be able to service different segments of the market as we move forward.

With respect to your second question, as Dan mentioned earlier, we are still seeing extremely strong levels of demand within the industry, certainly outstripping supply. And what we continue to see as we move forward is that capacity becomes increasingly scarce further out than people were expecting. So if we rewind even a number of months ago, '27, people were thinking that there was relatively decent amount of capacity available. We're already seeing that capacity available in '27 is extremely scarce, and that is continuing to push into '28 now as well.

So for us, there is certainly the ability to market those sites for '27 and '28 online dates. As Dan mentioned earlier, we're working through the type of customers that we bring into the mix and making sure that we are structuring the contracts in the right way to enable a flywheel at our end. But certainly, the demand signals are very strong.

And maybe just to add to that quickly, Kent. I think to directly answer the question, there's nothing stopping us contracting that capacity today. It just gets easier, the closer you get. So the focus is on time to compute. The demand we know is there, and all it does is make the conversations and the negotiations that we are having live time for a lot of that capacity much easier when you've got a defined construction and delivery plan, rather than trying to make things up on the fly in parallel with a full form agreement.

Next, we have Paul Golding from Macquarie.

Congrats on all the progress and the new relationships coming in-house. I wanted to ask about air cooled GPUs in general. So it sounds like with the 60-megawatt deployment at Childress for NVIDIA that will be an air cooled deployment along with the rest of the uncontracted capacity that you're deploying across British Columbia and Texas. Air cooled is going to represent a meaningful part of the strategy.

I just wanted to ask how you see efficiencies, as well as hardware performance looking so far based on the deployments that you've planned for? And how we can look at that from a financial perspective as well as we think about the model and the air cooled opportunity?

So in terms of efficiency and performance, I mean, what we're deploying across the air cooled portfolio is the latest generation of NVIDIA air cooled GPUs being Blackwells. So they perform extremely well. There is very high demand for those across all Blackwell GPU types and certainly continue to see customers finding a very good degree of performance versus cost efficiency from those units over time.

And sorry, Paul, I didn't quite understand the second part of your question in relation to how that converts into revenue over time.

That's right, Kent. So just wondering with, sort of, retrofitting and repurposing of Bitcoin mining infrastructure for these air cooled deployments, how that seems to be working out maybe from a margin perspective relative to some of the liquid-cooled deployments that you're doing around the Horizon projects, just given the simpler cooling opportunity there?

Yes. From an operational margin perspective, it is slightly more efficient than the liquid cooled deployments. But where we get the real benefit is, as Dan mentioned earlier, it's very capital efficient because we're taking existing air cooled data centers that require relatively little CapEx to retrofit them compared to brand-new build liquid cooled facilities. So that is the major difference in terms of the two. At an operating margin level, yes, air cooled is probably slightly higher but immaterial.

And if I could just sneak one more in around Europe and the Nostrum acquisition. As we think about the road map there, are you looking to use a similar form factor to what you've used either at Horizon or with air cooled facilities? Or is there a bespoke form factor you plan to leverage from that platform as you do the European rollout?

Yes. So one of the things that attracted us to the Nostrum opportunity, and we've been looking at Europe for a while is that they did have significant land holdings that came as part of that and access to a large amount of secured power. So that gives us quite a large degree of flexibility as we build out that platform over time as to the form factor that we use.

Typically, in Europe, you do tend to see slightly more condensed build-outs, but we do have the ability there to utilize our typical modular design that we use across North America, which obviously may well bring construction advantages with it. So that was one of the key elements that we saw in terms of the platform that they have and the projects they've developed.

Next, we have Brett Knoblauch from Cantor Fitzgerald.

Congrats on the, I guess, multiple acquisitions over the last week and the NVIDIA partnership and deal. I wanted to touch on Mirantis a bit because I thought that was important to the long-term story. Could you maybe just elaborate how that fits into your go-to-market motion, how it might accelerate your go-to-market motion when it comes to landing these enterprise deals, which is also what it seems like the NVIDIA partnership wants you to do as well?

Yes. Happy to take that one initially, Dan, and then you can add. So it brings with it a number of elements that we think are significantly attractive to our business, the ability to deploy quickly, the ability to service enterprise customers that may require a high level of software over and above bare metal. They also, as a large company that has very big internal engineering resources, bring very good capability on the software development side, and that can flow through to the business, not only in terms of the software stack, but also the operations of these large clusters more generally.

And further to that, again, having serviced customers for decades, they have an extremely well-built out customer support function internally. So all of those elements are things that attracted us to the Mirantis team and -- are able to add to the existing skill set and customer service support that we've already built up internally.

Awesome. And then if I could maybe just do a follow-up. Just double-clicking on the capacity ramp for '27. Am I right in thinking that of the 730 megawatts, 450 will come from the remaining Childress capacity and, I guess, the 280 would be coming from Sweetwater?

That's correct.

Next, we have Nick Giles from B. Riley Securities.

Congrats on all the developments here. I know the IREN team has a lot of experience in developing infrastructure in Australia, but maybe less under the IREN platform. So I was curious if you could walk us through some of the key differences, specifically in power procurement, maybe commercial strategy, so on and so forth?

Sure. Look, in some ways, Australia is very similar to other markets. And the operation of the electricity market in Australia managed by AEMO is very similar to what we see in Texas as ERCOT. There are markets in Australia, which resemble Texas in other ways, lots of land, good transmission line capacity, good fiber connectivity and abundant renewables, which isn't located close to other demand centers similar to what we see in West Texas. So there are a lot of parallels.

The reality is Texas is just an easier place to do business, and we've been able to accelerate faster there. But it hasn't stopped us continuing to incubate projects down in Australia, and we're getting far closer to those projects becoming a bit more of a reality. And I think the demand environment and the ability to service APAC, and the demand constraints that we're seeing and hearing in our conversations with hyperscalers, means that Australia looks like a fantastic frontier for us, and we'll look to accelerate that in parallel with North America and Europe.

Next, we have Michael Donovan from Compass Point.

Congrats on the progress. How should we think about regional customer mix as the platform expands? Are certain markets globally better suited for enterprise and sovereign AI customers versus hyperscalers? And does that change the expected contract structure or margin profile?

Look, it's going to evolve, and there's a lot of unknowns around this. But if you break it down, hyperscale contracts can mean two things. They can mean hyperscalers using capacity for their own purposes in terms of training and servicing workloads such as their own AI models. Or it can mean they're just acting as intermediaries to aggregate capacity for end customers that we're talking to directly. So obviously, in the case of the latter, whether you're dealing with a hyperscaler or going directly to the end customer, the end demand is the same.

Then you've got different types of workloads, so inference and training. Inference is a little more latency sensitive. Training, you can probably afford a bit more latency. And indicatively, we've had conversations around training models in Australia. Yes, the U.S.A. to Australia is a long geographic distance, but it's actually not that far over fiber, particularly where you're talking about training models. And given where inference sits today as well, we're all using ChatGPT or Claude, the response times are still, I guess, adjusting to the level of demand and the supply to service it.

So look, it will evolve over time, and our objective is to build out an expansive ecosystem of end customers. The partnership with NVIDIA is designed around that. The Mirantis and integration into our business is designed to help facilitate that over time in addition to all the near-term operational capabilities that it brings out. So the goal is very much to build out that diversified customer base over time across all of those markets.

Appreciate that. And a follow-up, if I may. Can you help bridge the 490 megawatts in Spain from secured power to time to first token? And what has to happen before construction begins?

So that is secured power and the sites across the portfolio there are secured as well. So from here, it's a matter of working through final design permitting, which is already well advanced at a number of those sites and then ultimately, construction of those facilities.

But one of the elements that we found very attractive was the near-term security of power. So that is power that is available on a time line that we think is going to tie in very well to general European demand, and we are already seeing a number of direct requests from existing and new customers for European capacity.

Next, we have John Todaro from Needham & Company.

This is Austin on the line for John Todaro. Maybe just a quick question on how do you intend to finance the build-out for the recently announced NVIDIA deal? It seems to be around 5 gigawatts. So just any color on that would be helpful?

I can take that. Yes. So the CapEx involved for the retrofitting of the air cooled data centers in Childress is pretty modest in the scheme of things. In terms of the CapEx for GPU, obviously, we've got a range of financing sources available to us. That obviously includes initiatives at the corporate level, but we can also look to finance GPU acquisitions in various ways in the debt capital markets through debt capital as well. So we'll be looking at all those initiatives.

In terms of the 5 gigawatts more broadly, maybe just to address that in the plan. So that's obviously a lot of capital today, but the reality is you don't need all that capital day 1. There's an S-curve of construction that takes time. It takes years to deliver this. This is the whole point around time to compute. It's not just a case of getting power and land. It's assembling multi-thousand construction teams and actually delivering it. And the funding for that just is progressive over time.

So as we've seen, as we continue to deliver, we continue to drive revenue, we can reinvest that revenue in CapEx, and it continues to unlock more and more financing sources over time. And part of the partnership with NVIDIA, we've announced they've got the ability to invest in IREN and as we commission GPUs. But equally, there's other support mechanisms being discussed to the extent that we need them.

But the reality is capital markets are open. They've been very supportive of our plan, and we anticipate that continuing at the moment that, that changes. There's a whole world of capital out there in terms of other options, whether you're creative around private markets or otherwise. And when you look at the GPU financing, which is the lion's share of that CapEx, the Microsoft contract is a great template. We financed 95% of that CapEx at an average interest rate of about 3% through prepayments and GPU financing. So the capital is out there as long as you sign good contracts and you show that you can execute and operate this capacity.

Next, we have Joseph Vafi from Canaccord Genuity.

My congratulations here as well on the great progress. Just a couple of thoughts, or some of your thoughts here just to gauge demand out there. I know you threw out a $3.1 billion contracted going to $3.7 billion contracted in ARR here exiting the year. Your confidence in that uncontracted capacity and signing contracts, how is the demand out there for that, say, extra $0.5 billion of ARR? And what kind of clients you may be looking to bring on board there? And then I'll have a quick follow-up.

I think -- sorry, go ahead, Dan.

Sorry, Kent. Look, again, we're trying to reiterate this as much as we can, and I'm very happy for someone to point it out, but there are no idle GPUs. And the prospect of there being GPUs sitting there unused, given how structurally constrained this market is a little over the near term, but in the medium term, it's not the focus.

We are having a lot of customer conversations, but all of our operational capacity is fully contracted. We're contracting substantial portions of capacity before it even arrives. And we're in discussions with a variety of customers all the way from hyperscale clients down to AI native labs all of that 2026 and 2027 capacity. So when a signature is put on a paper, it just flows naturally. Our conviction is around the demand supply, and you cannot tap into that unless you bring the capacity online. And this is a customer contract doesn't deliver revenue. Having compute online delivers revenue, and that has been the focus.

And then I was going to say many of the same things. So the one addition I would add is particularly for our air cooled capacity where we are adding substantial amounts across second half of '26 and into the early part of '27, there is very significant demand on those time lines. That is the most constrained portion of the market, and that is directly what is leading into the dynamic that Dan discussed where they just are not idle GPUs that are not being used in this market. Everything on shorter-term time lines is extremely attractive to counterparties.

Got it. That's great color, Dan. Just then on your strategy and philosophy around customers and diversification there, if you are in the catbird seat here relative to fulfilling demand from multiple parties. How are you looking at your broadening, deepening, diversifying that customer mix over time?

It's something that we're looking closely at, Joe. Like there is no set formula as to the proportional splits between different types of customers. There are benefits in having hyperscale clients in terms of financeability, contractual certainty, but there are also consequences in terms of price because you're not servicing the end customer in many of those instances. But the ability to service the end customer has been something we've focused on since day 1.

All of our early deployments have been very focused on non-hyperscale customers and getting as close to AI natives and enterprise as we can. So the Mirantis acquisition certainly helps that. I'm not going to sit here and say we're going 100% hyperscale. We're going 100% AI native end market. The reality is that blend will just emerge organically over time. And this, again, is part of the close working relationship we've got with NVIDIA.

We've spent a lot of the last fortnight in their San Jose office working through how we service all types of customers, all the way from the $1 trillion hyperscalers through to the emerging AI scale-ups where a lot of this innovation and development is taking place. And it's funny speaking to someone the other day, you don't need a sales team in this market, particularly when you've got NVIDIA, they see the whole ecosystem. The introductions, the referrals, putting us in touch with anyone that needs capacity. It's just happening so organically, so quickly live time that it will just play out a good way. But I think a combination of hyperscale, a hot combination of other is absolutely the goal.

Our last question comes from Ben Sommers from BTIG.

So I was curious a little bit on older generation GPUs. I know you've talked in the past as you've seen the useful life of older generations for like H100s extend out further than maybe people had originally thought. So kind of curious what you're seeing on the demand profile there, just what potentially type of workloads are going on for those older generation GPUs?

Yes. The comments that we made about no idle GPUs that applies to all GPUs, not just latest generations. So older generations, A100s, H100s, H200s, all fully -- effectively fully utilized across the industry. So the demand picture continues to be strong. In some instances, you're actually seeing pricing for older generation units climbing significantly, and there's a number of observable pricing points out there in the market where you can see that happening.

Yes, the type of demand may shift over time. You may have older generations being used more for inference, but also those older generations are equally suitable for certain types of training. So we just see strong demand across the board, both on the inference and the training side, and that continues to drive demand and [ elongated ] life cycles for those older generations of equipment.

Great. And then just on potential future conversations that you're having for potential contracts down the line, is there any talk of prepayment structures similar to that of Microsoft? Or just kind of curious what you're hearing in the market on that end?

Yes. It certainly plays a role in a number of those conversations, and we are still seeing prepayments being on the table in a large number of instances. Now it obviously factors in as part of the overall equation. So it's not the single factor that you're looking at. Everything has to go together with a combination of term length, prepayment, creditworthiness, price, but prepayments are certainly very much on the table in the current environment.

I see no further questions at this time. I'll now pass to Dan for closing remarks.

Thanks, operator. Thanks, everyone, for joining us today. We remain focused on execution, delivering the 2026 plan, advancing the 2027 build-out and positioning our now global platform for the opportunity beyond that. And we look forward to updating you as we deliver. Thanks, everyone.

Good day, and thank you for standing by. Welcome to IREN Q2 FY '26 Results. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference over to your first speaker today, Mike Power, Vice President, Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, and welcome to IREN Q2 FY 2026 Results Presentation. I'm Mike Power, VP of Investor Relations. And with me on the call today are Daniel Roberts, Co-Founder and Co-CEO Anthony Lewis, CFO; and Kent Draper, Chief Commercial Officer. Before we begin, please note this call is being webcast live with the presentation. For those that have dialed in via phone, you can elect to ask a question via the moderator after our presentation. I'd like to remind you that certain statements that we make during the conference call may constitute forward-looking statements and IREN cautions listeners that forward-looking information and statements are based on certain assumptions and risk factors that could cause actual results to differ materially from the expectations of the company.

Listeners shouldn't place undue reliance on forward-looking information or statements and I encourage you to refer to the disclaimer on Slide 2 of the accompanying presentation for more information. And finally, during the course of today's call, we will refer to certain non-GAAP financial measures as a reconciliation schedule showing the GAAP versus non-GAAP results in the presentation.

With that, I'll now turn over the call to Dan Roberts.

Thanks, Mike, and thank you, everyone, for joining us today. Fiscal Quarter 2 was an important quarter for IREN as we made meaningful progress as a vertically integrated AI cloud platform. Let me start with the highlights. Firstly, we secured underwriting commitments for $3.6 billion of GPU financing at an interest rate of less than 6%. Together with customer prepayments, this provides funding coverage for approximately 95% of the GPU related CapEx supporting our $9.7 billion AI contract with Microsoft. .

Importantly, this financing package provides greater clarity to also advance a broader set of customer discussions. In that regard, customer demand remains very strong and we are continuing to sign and negotiate contracts for both new and prior generation GPUs. We have multiple advanced negotiations underway for larger-scale deployments and are also seeing hyperscalers and AI enterprises increasingly focused on air-cooled GPUs given the faster deployment time lines. Operationally, execution is tracking well across the portfolio, and we expect to deliver 140,000 GPUs by the end of 2026 positioning us to deliver $3.4 billion in annualized run rate revenue.

Construction across Horizon 1 through to 4 is progressing to schedule. And in British Columbia, we continue to expand our AI cloud footprint with just under $0.5 billion of ARR now under contract for Prince George. Finally, we extended our growth runway again by securing a new 1.6 gigawatt site in Oklahoma, taking our total secured power to over 4.5 gigawatts. This reflects the strength of our internal development team in securing gigawatt scale sites in a power constrained market and supports continued conversion of capacity into customer contracts over time. So that's the quarter in summary. Those outcomes reflect the assets, capability and execution discipline we've built over time, which I'll cover next.

Over the past 7 years, we've built a strong platform grounded in real assets, power, land, data centers and just as importantly, human capital. That foundation is what gives IREN a durable competitive moat. We have secured more than 4.5 gigawatts of power, stood up 810 megawatts of operating data centers signed billions of dollars of AI customer contracts and assembled a team of over 2,000 people to execute on them. These assets and capabilities are not easy replicable. They are the results of years of hard work. As a founder-led business, we have been fully committed to building a platform with lasting value. And Will and I are deeply invested in this platform along with the rest of the management team. That mindset matters as it shapes how we allocate capital, how we partner with customers and how we think about long-term value creation.

In an industry moving at extraordinary speed, this combination of real assets, operational capability and founder led commitment is what sets IREN apart and positions us for AI cloud leadership. So the way we think about scaling the business is through what we call the 3 Cs capacity, customers and capital. The reason we focus on these 3 is simple. They reinforce each other. Capacity creates opportunity customer commitments shape the pace and scale of our investment and capital gives us the ability to execute. What's encouraging today is that we have all 3 working in parallel. First, on capacity. We have 810 megawatts of existing data centers that can be immediately leveraged for AI cloud deployments. In addition to the 3.6 gigawatts of greenfield data center sites and a 2,000-plus team to design, build and operate them end to end.

Second, on customers. As I mentioned, we are in multiple about negotiations. And at this point, demand is not the constraint for us. The focus is on choosing the right long-term partnerships that support durable platform level growth. And thirdly, on capital, we continue to diversify our sources of capital to support capacity growth and customer deployment. We have multiple financing pathways underway that allow us to scale our data center and GPU footprint in a disciplined manner, while importantly, maintaining balance sheet stream. This includes additional GPU financing, data center financing and selective corporate initiatives, which Anthony will delve into.

So when you step back, the picture for us is pretty clear. We have delivered capacity. We have strong customer demand, and we have expanding capital options, all moving together, which puts us in a position to continue scaling IREN into 1 of the world's largest AI cloud platforms.

With that, I'll now hand over to Kent to walk through updates to our capacity and our customer work streams in a bit more detail.

Thanks, Dan. Vertical integration is 1 of IREN's most important competitive advantages. We design, build and operate our own data centers, supported by in-house engineering, procurement, construction, technology and operations teams. This structure gives us direct end-to-end control of our cloud offering and the ability to manage cost, time lines and service quality with far. Many of the constraints that we see across the industry today, whether it's long lead time procurement or skilled labor are areas that we've addressed in the past. So they're manageable for us and not disrupting our execution. That's why we remain on track against our plans today. And on that note, we continue to see strong, steady progress across our site portfolio with construction milestones being delivered on schedule.

At Prince George, the data center fit-outs for air cooled NVIDIA B 200 and B 300 GPUs and and now complete and awaiting the delivery of the remaining GPUs on order. At Mackenzie and Canal Flats, our teams are actively preparing the sites for AI cloud expansion. There, we're leveraging the exact same playbook we successfully executed at Prince George. ASICs are coming out of those data centers and GPUs are going in. At Childress, construction across Horizons 1 to 4 is also progressing to schedule to meet Microsoft's GPU deployment time lines. and that Sweetwater procurement activities and civil works are now underway for the first phase of data centers to be constructed. Overall, what this demonstrates is an ability to consistently take large complex projects from planning through to execution. That delivery capability underpins everything we do and is 1 of the key reasons we have a license to engage with the largest technology companies in the world.

The other factor that strengthens our position with customers is a scale of our secured power. As Dan mentioned, we have secured a new 1.6 gigawatt data center campus in Oklahoma. Further strengthening what is already 1 of the most differentiated power portfolios in the sector. The 2,000-acre Oklahoma site is a strong addition with low latency connectivity to major network exchanges and ramp schedule commencing in 2028. As with Sweetwater, the megawatts for this new site in Oklahoma remains secured, which enables commercial discussions to progress meaningfully anchored on firm deliverable capacity. It's also worth noting that this site is a result of work that has been underway for years. It reflects the depth of our internal development capability and our team's ability to consistently source and secure gigawatt scale grid connections in a power constrained market.

Importantly, this new site does more than just add capacity. It broadens our U.S. data center pipeline beyond ERCOT into a market that's attractive to hyperscalers at a time when AI demand continues to outpace supply. As Dan mentioned, we're advancing multiple active negotiations with a range of hyperscale and non-hyperscale counterparties. And what we consistently see is that customers are focused on partners who have secured power can deliver full data center infrastructure on a defined time line and who can grow with them and scale over the long term. In other words, time to data center has become the key decision point in many of these commercial discussions. That dynamic plays directly to our strands.

Our vertically integrated model combining in-house delivery capability with secured power gives customers a high degree of certainty on execution. We're also seeing hyperscalers and leading enterprises actively pursue both liquid and air cooled GPU deployments as they work to accelerate rollout. The increased focus on air cool deployments aligns extremely well with our existing footprint of 810 megawatts of already operational air cooled data centers. Importantly, all the demand and engagement that we're seeing is translating directly into contracted revenue.

Today, we have approximately $2.3 billion of annualized revenue run rate under contract, including around $0.4 billion at Prince George which we expect to increase over the coming weeks as we finalize negotiations for the remaining capacity there. Based on capacity already contracted and the strong customer engagement for new deployments in Mackenzie and Canal Flats, we're on track to reach our targeted $3.4 billion ARR by the end of 2026. What's worth emphasizing is that demand is not the limiting factor for reaching this milestone. The market is extremely deep and engagement remains strong across hyperscalers and enterprises. Our focus is on selecting the right partners and structuring long-term relationship to create lasting value for IREN and as I alluded to earlier, we have the track record and capacity to drive customer acquisition.

The takeaway from this next slide is the amount of runway that we have ahead of us. Our $3.4 billion ARR target for the end of calendar 2026 reflects utilization of only around 10% of our 4.5 gigawatts of secured grid-connected power capacity. That means the vast majority of our portfolio remains available to support additional deployments. With demand continuing to build that secured capacity gives us the ability to keep engaging customers on new large-scale opportunities and to extend growth well beyond the 2026 target.

I'll now hand over to Anthony to give an overview of our Q2 results and discuss our strategy for financing our continued growth.

Thanks, Ken. Q2 financials reflected continued progress in the transition from Bitcoin mining to AI Cloud with capacity increasingly allocated to higher-value AI workloads and AI cloud revenues accelerating as deployments ramp. Total revenue was $184.7 million, down 23% on the prior quarter, primarily on account of lower bitcoin mining revenue, driven by a reduction in bitcoin mined. This was as a result of the AI transition, which lowered operating hashrate against the backdrop of an increasing global hashrate, together with lower average Bitcoin prices over the period. This was partially offset by growth in AI cloud revenue in line with commissioning of new GPUs at our Prince George side.

On SG&A, that decreased $37.6 million. Primarily resulting from higher accelerated stock-based amortization recognized in the prior period and associated payroll tax accruals. As expected, adjusted EBITDA declined primarily on account of the lower Bitcoin mining revenue mentioned just now, partially offset by the lower payroll tax accruals and lower power costs. EBITDA and net income were also impacted by several significant noncash and nonrecurring items this quarter. These included unrealized losses on prepaid forwards and cap calls associated with our convertible notes following significant gains in the prior period as well as a onetime debt conversion inducement expense in connection with the equitization of a portion of our 2029 and 2030 convertible notes. Together, these amounts totaling $219.4 million.

In addition, we recorded $31.8 million of mining hardware impairment associated with the transition to AI cloud versus $16 million in the prior period. These impacts were partially offset by an income tax benefit of $192.5 million, primarily reflecting the release of previously recognized deferred tax liabilities relating to the unrealized gains on financial instruments. Overall, these results reflect the ongoing transition of the business to AI cloud and we expect subsequent quarters to reflect a growing AI cloud contribution consistent with our ARR targets.

Now turning to capital and funding. At our last update, we indicated an intention to raise secured finance of at least $2.5 billion to fund the GPU-related CapEx for the Microsoft contract. As Dan has highlighted earlier, we have now secured a $3.6 billion delayed draw term loan financing package from Goldman Sachs and JPMorgan. The financing has a number of attractive features. It is delayed draw to align with our CapEx profile. It matches the profile of the underlying contract, amortizing in full over the 5-year term and will be secured against the underlying GPUs and the contracted cash flows from Microsoft, which supports a strong credit profile and an attractive interest rate expected to be less than 6%. When combined with Microsoft's $1.9 billion in prepayments, this package covers 95% of the GPU related CapEx for Horizons 1 through 4 allowing us to now focus our efforts on funding further growth across the platform.

Now turning to our capital strategy more broadly. Our capital strategy is designed to support continued rapid growth while maintaining a resilient balance sheet. We ended January with a strong cash position of $2.8 billion, and we continue to deepen our access to diverse sources of funding. Financial year-to-date we have now secured $9.2 billion from customer prepayments, convertible notes, including the $2.3 billion issued in December, GPU leasing arrangements and the dedicated GPU financing for the Microsoft contract. This diversity of capital sources allows us to scale with confidence. Looking ahead, a key priority will be to continue that work and expand our access further. This will include looking at efficient financing for our data centers, such as Horizons 1 through 4 when they come online. These will be extremely valuable long-term assets, which are currently being funded by the balance sheet and construction financing to support our broader development pipeline.

As well, we'll also look at select corporate level facilities when aligned with our cost of capital and balance sheet management priorities. Of course, as we scale financing activities, we'll remain focused on an appropriate balance between debt and equity to ensure we maintain that balance sheet resilience.

With that, I'll turn it back to Dan for concluding remarks.

Thanks, Anthony. So to recap, our strategy is straightforward, and it comes back to the 3Cs capacity customers and capital. We secure a large-scale, low-cost power and build quality data centers to create capacity. We pair that with long-term customer demand. and we support it all with disciplined, well-structured capital arrangements. That framework has guided our decisions for years, and it continues to shape how we scale today.

Over the past several quarters, we've made meaningful progress across all 3 of those dimensions. We've removed a significant amount of execution risk by locking in capital for our largest deployment to date. We've expanded our power footprint. Sorry technical issue my screen just cut off, I can't see anyone -- and we've continued to deepen relationships with some of the largest technology companies in the world. What's important to emphasize is that we're still at an early stage of monetization relative to the size of our platform. With more than 4.5 gigawatts of power and only 10% required to support the $3.4 billion in ARR, we have a clear pathway for continued growth.

With capital access now demonstrated at scale, we're able to engage customers with greater flexibility on how and when we bring new capacity to market while maintaining discipline around pricing and partner selection. That scale allows us to pace growth responsibly be selective in the customers we partner with and structure contracts and financing in ways that support durable long-term value creation. In summary, after more than 7 years of founder-led execution, IREN is now a scaled AI cloud platform with significant opportunity ahead.

With that, we'll open the call for Q&A.

[Operator Instructions] First question comes from Darren Aftahi from ROTH.

Congrats on the continued progress with Oklahoma. Two, if I may. There's a lot of noise around ERCOT. I'm sure people on the phone kind of want to know. But any change in those kind of amended rules with batch processing in terms of how that would potentially impact Sweetwater for you guys? And then I've got a follow-up.

Yes. Happy to jump in there, Darren. So the short answer is with respect to Sweetwater it is likely to be included in the batching process, and we believe that both SweetWater 1 and 2 would be included in batch 0, which would mean that the full 2 gigawatts of power is secured. So that's obviously a key important point there is that security of power in addition to that, there are other projects in our portfolio that are potentially also included in batch 0. So 1 of the advantages of, obviously, having a large internally developed portfolio is that we do have a number of projects that are going through this process. .

Excellent. I appreciate that. And then secondarily, so economics on colo have continue to creep up. I know you guys initially signed this cloud deal with Microsoft. Any kind of real-time thoughts on as you move forward with plans for Childress, Sweetwater, any other sites your views on AI cloud versus colocation?

Yes. I mean, as we've said before, we continue to be open-minded about how we allocate our megawatts and continue to observe what's happening in the market. We are observing, as you mentioned, what's happening in the colocation market but also seeing continued strength in demand on the cloud side as well. And I think when we look at the overall portfolio, Power is the scarce resource today. And so it's absolutely vital that you are maximizing the value of each of the megawatts within the portfolio. And today, we still see AI cloud as doing that in a more meaningful way than colocation. It's higher up, obviously, in the value chain than colocation, and you can capture materially better dollars per megawatt through cloud versus pure co-location.

And obviously, at this point, we've demonstrated the capability and execution to be able to stand up these large cloud customers. And as we spoke about at length during the call, also seeing the capital side come together. So all the elements are there within the portfolio to allow us to continue to take advantage of what we have on a cloud services basis.

Maybe just to add to what Ken said, it is something we continue to evaluate. But 1 of the knocks on GPU Cloud was the capital intensity of GPUs. So with the announcement today of the GPU financing, we've now secured 95% of the cost of the GPUs at an average interest rate of around 3% when you factor into the prepayment. So we essentially got the GPUs for next to nothing. So I think in terms of capital intensity, it ticks that box. To further to Ken's point, time to power is critical, but time to data centers is actually the more limiting factor. And when you've got scarcity around how many data centers you can physically bring on live, every incremental 200 megawatts can deliver either $300 million-ish through a co-location or multiples of that in the billions under a cloud contract.

So when we look at the monetization opportunity for our platform, and growth for shareholders and creating value, the cloud opportunity creates a lot more upside as we see it. And in terms of co-location versus cloud, we believe that AI is going to continue. We believe that data center demand is going to continue to compound. So recovering capital back in short order and mean out to compound those returns as distinct from holding effectively a bond exposure against the hyperscaler is the area we want to play. And if investors want bond-like exposures, they can buy colocation companies, they can buy bonds in the hyperscalers but we believe we are offering a high conviction for very well-managed risk exposure to the sector through this AI cloud business.

Now again, we're not dogmatic. Things can change quickly. We get a compelling colocation deal. We will absolutely pursue it. But right now, AI Cloud, it's very compelling for all those reasons.

Next question comes from Paul Golding from Macquarie.

And congrats on the additional site and all the progress. I just wanted to ask, as we think about the Oklahoma site and power market. I guess anything specific to call out about how that factors into the demand picture for HPC compute from a location perspective. Aside from, I know the low latency already mentioned, are there favorable power reliability dynamics or power pricing dynamics or just geographically relative to Tier 1 availability zones? And then I have a follow-up. .

Yes. So we think that site there has favorable characteristics on a number of levels. I did mention that low latency as you referred to earlier. It's a very large site, which gives us flexibility as we build out the capacity, it's located in Southwest Power Pool, which is a different market to ERCOT. So it provides us with some jurisdictional diversity. We think Southwest Power Pool is a very attractive market on the power side, a large penetration of renewables, low cost of power. We know that it's an area that is attractive to hyperscalers because there have been a number of hyperscalers that have been active in Southwest Power Pool more broadly, but also Oklahoma specifically. So overall, it exhibits all the characteristics that we would look for in terms of an attractive data center campus.

And then maybe a follow-up on the questions that have been asked around cloud versus colo or maybe more specifically about cloud. As you roll out these 2 different approaches to cloud even right with your British Columbia clusters versus the Microsoft clusters. How should we think about your software approach looking at the neo cloud space software is a topic that comes up quite a bit. I guess how should we think about your software offering for certain clusters where there's on-demand or smaller contracted deals versus, of course, the bare metal deals and how that development and the uptake from customers has progressed. .

Yes. Today, we're still seeing the bulk of our demand coming from hyperscalers, the largest enterprises, extremely advanced technology firms within the AI space, all of which are still looking for bare metal access. They want full ability to be able to take control of the GPUs, layer on their own software stack, set up the compute in exactly the way that they want to operate it and that is where the vast majority of our demand is coming. And as we referred to in the call, our ability to scale with them over time is 1 of the key elements.

And I think the largest customers are those bare metal customers whereas the software really is typically more useful for smaller enterprise customers. that are looking for an easy user interface and easy single spin up, spin down, service but that is a small proportion of the overall compute demand that we're seeing out there today. It may well grow over time, and that's something that we continue to monitor as we look to our software strategy. But we continue to think that it is likely to be 1 of the areas in this space that gets commoditized, the fastest. It is relatively simple in comparison to finding power building data centers, setting up GPU clusters at scale and likely to be an area where you're going to see third-party offerings and commoditization, that we may well be able to take advantage of.

So in short, continue to monitor that part of the market and what makes sense for us. But today, it is not a major driver for us because our demand is coming from bare metal customers .

And maybe just to give you some additional comfort around the way the world might go here, Paul, is we do have an internal software capability. I think we probably downplay it a bit, partly in response to the market seeming to overplay it, but we've got that capability. And to give you additional comfort, 1 of the contracts we are negotiated at the moment is a multibillion-dollar contract where we need to bring a software solution. So it is not holding us back. It will not hold us back. But the reality is exactly is what Ken said, you are dealing with the largest technology companies in the world to pretend that you can be better at software and jam something down their throat when that is their competitive moat and that is their expertise, it's just not on growth reality.

Next, we have Michael Ng from Goldman Sachs.

I just have to First, as a follow-up to the question earlier around the ERCOT batch study processing. It was encouraging to hear that the IREN site slightly will be in batch 0. I was just wondering if you could provide an update around the SweetWater 1 and SweetWater 2 energization dates and whether the batch process has affected your ability to negotiate and sign contracts with customers for those sites and what that progress looks like. And then I have a quick follow-up. .

Yes. Thanks for the question. In terms of the energization date for Sweetwater 1, we're still on track to energize in Q2, and that's a full bulk substations. So that's capable of the full 1.4 gigawatts of power capacity at that site. So energization very much on track. Construction is well advanced, both with the on-site substation as well as the utility substation there. As it relates to customer engagement moving forward on those sites, obviously, very early since this matching process has been announced. But if anything, we would actually expect it to be helpful to us. We've said numerous times in the past. There are a lot of megawatts that are put out there into this market that are made up and I think what this process is going to do is really uncover, which megawatts are real and which are not real. And for us, we expect that to actually lead to better discussions with our customers over time.

Great. Wonderful. And I wanted to ask about the $2.3 billion of ARR, which I guess, the Microsoft contract plus the $400 million at British Columbia. When should we expect those revenues to start being recognized and commencing in the P&L? Is it kind of more ratably through the year? Is it more in '27? Just would love to get any thoughts about that? .

Yes. So at Prince George, we've obviously had capacity operating there for a while and continue to install new capacity and we'll do over the upcoming weeks. So a decent proportion of the $0.4 billion of contracted revenue that we talked about is already operational there. As it relates to the Microsoft contract, that will come online progressively over the course of the year, commencing we expect Q2 in terms of initial revenues flowing through

Next, we have Brett Knoblauch from Cantor Fitzgerald.

Congrats on all progress throughout the quarter. I'm curious in your conversations with customers relative to maybe the first big deal that you guys signed with Microsoft, what you are seeing from a pricing environment, I think we have a lot of data points on the colo pricing may be improving out there. But I'm curious if you guys are seeing something similar when it comes to the cloud deals.

Yes. We're seeing very strong ongoing demand, as we referred to earlier. And I think that is flowing through in a number of potential areas. We're seeing demand for longer tenures, I think the customers that are out there in the market realize that this may be a long-dated supply-demand imbalance moving forward. And there's certainly an openness that we're that we're seeing to longer tenures than we have in the past. Another factor that we mentioned earlier, we are seeing an increased interest in air cooled capacity.

And that is primarily because that can feel immediate needs, especially within our portfolio because we have existing operational data centers on an air core basis that are capable of hosting GPUs in relatively short order with relatively minimal capital upgrades. We continue to see the ability to get prepayments from customers over time. So I think all of that leads us, as we said, to see a very strong demand picture and it is flowing through in some elements of the terms that we're getting under these cloud services contracts.

And I think also to add, to that. Price is 1 dimension of a commercial negotiation. There are other factors, as Kent alluded to, whether it's tenure and prepayments, but also the quality of the underlying contracts. We do manage risk very carefully. It's a founder-led business. This is our money. This is our platform. And we're not here to optimize revenue in the next 4 weeks compared to building something that's durable and long-term value. And to highlight what that means in tangible terms, look at the GPU that we did on the Microsoft contract.

So to step back, $5.8 billion of GPU costs to deliver $9.7 billion in revenue over 5 years of the $5.8 billion the nature of the contract is in the quality of the underlying contract, the quality of the credit, the tender and the prepayments allowed us to get $5.5 billion out of the $5. 8 million financed at an average cost of 3%. And like that is not specifically linked in a GPU hour price, but that is specifically tied to value creation on the platform.

Awesome. Very helpful. And then maybe just 1 more, ERCOT related question. I think you guys had using your words here that it's likely included in batch 0, whether that's A or B. I guess is there a time of when we would expect to know if it's included in batch 0. I know there's a meeting on the 12th and maybe on the 20th, but is that the time line that you guys are looking forward as well?

Yes. I think ERCOT will make announcements over time, Exact timing may change and whenever ERCOT makes announcements with respect to this, they do acknowledge it is in the works at their end, and they're actively working through it. So hard for us to put an exact date on it, but we do expect ERCOT to make public disclosures at some point in the relatively near future .

But to be clear on this guy, like crystal clear, that 2,000 megawatts is secure. Like none of this batch stuff, none of the market chatter is influencing whether or not this 2,000 megawatts is available. We've got the signed interconnection agreement. It was signed in 2023, it's been there for years. It's been built and commissioned in Q2 this year. There is no indication that 2,000 megawatts is absolutely secure. The only thing that this is likely to amount to is maybe working with utilities around load ramp. But the reality is we don't have 1,400 megawatts of Sweetwater 1 of data centers in April this year to energize. So in practical terms, it has very little if no effect on our business. The 2,000 megawatts is secure. We cannot reiterate that enough. .

Next, we have Nick Giles, B. Riley Securities.

Good to see all the progress here. I like the concept of the 3 Cs capital is one. I think this is mainly around financial capital, but there's a growing narrative around the human capital requirements to ultimately bring data center capacity online. So are you seeing any constraints in terms of skilled workers? Or can you just speak to any advantages you may have from having EPC partners in place.

Yes. I mean the fact that we've been building continuously for the last 3 years means that we've built up not only a large existing labor pool at Childers, but also those relationships and the relationships extend not just across construction contractors and labor but also across equipment, procurement and supply chains. So that is 1 of the major advantages that we have and having done this for so long and having been continuously constructing is that we are in a position where we're able to call on those relationships we're well positioned with those partners in the sense that they are looking for continued steady work. And when they look at us and see a secured power portfolio with construction that is going to extend over a multiyear period, they're extremely willing and active in terms of helping us and making sure that they're serving our needs.

And similarly, on the supply side, because we're continuously in the market and continuously procuring long-lead equipment, we get a very good read on where the constraints are in supply chains where the areas are that are tight, and that enables us to respond to those and be able to act well in advance. So that long lead items don't become a constraint for us in terms of our data center build out. So I think all of that history, the internal expertise we have is extremely important. And it's not just talk. I mean this is consistently delivered capacity against the targets that we've announced historically.

Yes. And again, just to add to exactly what Ken saying, like, this has been 7 years in the making of building a data center and technology platform, the very first data centers we built are now being used for NVIDIA GPUs for an AI cloud. We signed an MOU with Dell was at 5, almost 6 years ago to bring out diverse workloads to our British Columbia facilities in these data centers. So we've had a very long runway in terms of accumulating that human capital. And yes, there is more scarcity and more demand for human capital today, but we've been able to build that platform over a very long period of time and get the right people in the right roles. .

Next, we have Joseph Vafi from Cannacord Genuity.

Terrific progress once again. Awesome to see it. Just revisiting the ARR number for the year, you clearly -- IREN is always want to overpromise and under-deliver and throwing that number out on top of revenues that would be coming from Microsoft kind of feels like you've got a pretty good line of sight on things just wanted to drill down on that a little bit on those customer ramps. And maybe is there a potential on some of these other customer ramps to also see maybe some prepayments to help fund their own GPU buys? And then I have a quick follow-up. .

Yes. Thanks, Joe. I hope you were referring to under promising and over delivering rather than the other way around. Yes. So as I mentioned earlier, Prince George, we already have a lot of operating capacity there and expect over the upcoming weeks to continue to install equipment, allowing us to get to the $0.5 billion annualized revenue run rate at that side. Mackenzie and Canal Flats the the works and our end in order to be able to accommodate GPUs, very well advanced. We would expect capacity to ramp progressively over the year there. .

In terms of the additional 40,000 GPUs, which equates to around $1 billion of annualized revenue run rate. And then the Microsoft contract, as I referred to earlier, we expect to ramp progressively over the year. In terms of the 40,000 additional GPUs that we're expecting at Mackenzie, That, as I referred to the customer conversations before, we are still seeing customers very willing to make significant prepayments with respect to that. And there are a number of other areas of financing that we're looking at with relation to that, which Anthony, you may well want to touch on some of the options there as to how we can finance that.

Sure. We've obviously I guess, over the financial year-to-date proven access to both leasing-based sources of capital for GPU financing and obviously, the dedicated GPU financing that we recently procured for Microsoft. So there's a number of different pools of capital, which will obviously depend on the nature of the customer and the opportunity, but we feel well placed to continue to fund that growth efficiently.

And then just circling back on SweetWater, obviously, energization coming up here very quickly. And a lot of your peer companies would have likely announced at least there was colo, a tenant at that site by now Obviously, it's really big. There's a lot going on and not asking for a date on anything. But just getting in your mind maybe a little bit, Daniel, is it -- is it just getting your feet more wet in the GPU business and holding back there, waiting for better terms on colo maybe a multitude of things, just your thought process there on pulling a trigger on some of the sweetwater capacity. .

Sure. So I mean we've had an ongoing dialogue on that site for 12, 18, 24 months with various parties. And as we've tried to reiterate, it needs to be the right deal. And I think to date, our patience and conviction has been rewarded with the deals that we have been signing. If we look back to some of the structures being floated early in this kind of AI market narrative, where we are today, seems to be pretty objectively a better position. It is all about the 3 Cs and bringing those together and doing it in a way where you are maximizing the opportunity for shareholders. And there's only so much capacity that you can bring online that time to data center narrative.

So there's a real opportunity cost of signing a bad deal. And that is relative, right? -- as relatively good. It's still probably a good deal colocation, but can you get better given that you're constrained by how quickly you can build out data centers. And that's why those 3 Cs are a really good framework because for any business trying to operate in this space, you have to bring those 3 Cs together to sell reinforce each other. If you haven't got the power and the capacity and the ability to execute, you're going to struggle to be a player. If you haven't got the access to customers and their faith and belief in you as an execution machine, it's going to be tough. And if you haven't got the capital, then you kind of get continue to be on 0. So sequencing all of that having not reinforce each other is really important.

And I think that's why we're really pleased around the GPU financing result because it's kind of ticked that box. We're now on to the next one, and it also helps catalyze a lot of these other customer negotiations were having an advance into the next phase because we need capital and you can't build without capital. So the GPU financing is now done on to the next one. We've got the capacity, we've got the customers and the demand of the negotiations underway. And as Anthony said, we've got what we see is really good access to capital at the moment.

Next, we have Michael Donovan from Compass Point.

As on progress. So following up on questions around Sweetwater, assuming the batch process goes smoothly and getting the 3 Cs together, how should we think about ramping up phase for construction. Would this follow children's 50-megawatt tranches? Or do you have different plans? .

Yes. So in relation to that, yes, look, it's going to be a phased build out. It's a very large. That is a large amount of power and what we're continuously doing across all of our 4.5 gigawatts of secured portfolio is aligning customer discussions, availability of capital and our ability to build out data centers and that will influence how we actually build out. But as Dan referred to earlier, likely at the moment, the constraint is the actual pace of construction and ability to construct rather than power availability or capital availability at our end. So we'll be a phased approach, and we will continue to triangulate with the levels of customer demand that we continue to see.

Appreciate that. And then on Oklahoma, can you provide some more color on what assets are currently there? And then what long-term lead assets are needed for the site build-out and what additional permitting or studies are or needed at the Oklahoma site. I appreciate it.

So we mentioned the 200 acres of land earlier. So all of that land is secured. The land is immediately adjacent to a major utility substation, which is where we will be connecting to the transmission grid. On the power side, the full 1.6 gigawatts there is secured. So all of the key elements as it relates to a data center campus are there. Over the upcoming months, we'll continue to work on the various development items, which include master planning, more detailed local permitting et cetera. But with power availability there from 2028, we feel extremely well placed with where we're at today.

Next, we have John Torado from Needham.

Congrats on the progress. I just have one, it relates to kind of these credit backstopping that you might see from NVIDIA. Just how do you think the competitive dynamic changes on the cloud side? You guys obviously have a ton of power but if some of these neo calls are able to get more kind of an NVIDIA back stop, they could get more contracted power as well. Just I guess how are you thinking about that?

Well, I'm not necessarily sure an NVIDIA backstop helps them secure more power. I mean, power and more pointedly to Dan's point earlier, data center capacity is the constraint. Now having a backstop can allow you to finance the build-out of a project, but you still need that access to power and unlinking necessarily the backstop sort of help with that. That is an entirely separate process relating more to development. And as we've spoken about before, that is becoming increasingly challenging as you move forward now for new projects.

So 1 of the advantages that we have of having been doing this internal development of projects for years is that we got in early and we have secured what we think is an extremely differentiated portfolio on the power side, and that's something that can't be easily replicated and certainly not something that can just be bought if you somehow get access to a credit back stop. So I think credit backstops can be helpful in other context, but I don't think it's going to give people necessarily faster access to power or data center capacity.

And just to add to that, I think it would be very dangerous to assume that we haven't got the same access and conversations around all these different structures in the market, whether it's equity investments, whether it's credit back stops, whether it's offtake agreements to assume that we're not having those conversations that haven't been having those conversations. Yes, I'd be careful about that. .

That's exactly what I was getting at, Dan. That's helpful. So I guess the takeaway from us is would Core and some of these others are having with NVIDIA, we should think you guys are right there in the same boat, right? You're right there with them? .

Yes. I can't obviously comment specifically on counterparties, but generally, the sector is a very small sector. We are a player. We've got a $10 billion contract with Microsoft. I would encourage it to be a safe assumption that we are having very similar, if not exact same dialogue with all these different counterparties about different structures. .

Next, we have Mike Colonnese from H.C. Wainwright & Co.

Dan, congratulations on all the great progress you guys have made in the past couple of quarters. First 1 for me, and I'm sorry if I missed it, juggling a few calls here. Looking at the CapEx projections for the year versus the total amount of additional financing, we did to complete the full 140,000 GPU deployment. Can you just give us an update on that? I know you secured the $3.6 billion in GPU financing covers. It covers most of it. But how should we think about the progression of CapEx spend this year and the remaining amount of financing needed to get to your target?

Sure. Thanks for question.I'll take that one. So I guess, we guys, as you know, we've got the $5.8 billion CapEx on the compute for Microsoft. We've got the approximately $3 billion CapEx for the Verizon data centers, a material amount of which has been incurred or committed to today. And we've obviously raised the recent GPU financing package in addition to sources of cash on balance sheet. So I think we can -- effectively, that's all of the Microsoft related CapEx for the compute and DC spoken for.

When we think about the CapEx required for the rest of the ARR growth target to the $3.4 million. We've previously talked about the CapEx required for that expansion at PG and Mackenzie. So taken together, that's about circa using round the circa $3 billion of CapEx a material amount of that, which has been incurred to date and financed through the lease-based financing that we've announced to date. But the focus for financing activities going forward will be obviously that residual amount for the expansion across BC and opportunistically as we look at further growth across the platform.

Our last question comes from Ben Sommers from BTIG.

You made a comment earlier about strong demand for older generation chips. I was just kind of curious, maybe is there a different kind of customer mix for older generation GPUs versus newer generation GPUs. And I guess kind of like how long do you see the tail going on to continue generating revenue off kind of older generation ships? .

Yes. I think in general, you tend to see newer generation chips being used more for training. Typically, in training scenarios somebody is training a model to actually get a product out to market and speed to market is important. So generally speaking, they want the highest power chips in order to speed up their production times. What we see as the chips get older is that the use case can shift more to the inference side. Now that's not to say they're not useful for training. You can absolutely still do training on older generations. But often, they are used more and more for inference over time. And inference continues to become a larger and larger portion of the over pine. I think we'll continue to do so over time.

In terms of the second part of your question around economics and longevity of these chips, I mean, we still see very strong demand for older generations of chips. So I think you've got to think about the demand picture in aggregate and overall, it's still very clear that there is an undersupply relative to demand. And so what that means is people will take compute as it is available. And if that happens to be older generations of chips that they can get their hands on and they're absolutely willing and not only willing, but requiring that capacity. And if you look more broadly across the industry, if you think of A100, A100s, those are more than 5 years old and more than 3 years old, respectively, now those chips are still effectively 100% utilized across the industry and still earning very good rates of return against their original capital costs.

So we continue to believe that these chips will have a long economically useful lifetime in excess of the contract length that we are signing, even the Microsoft 1 at 5 years.

I see no further questions at this time. I will now turn the conference back to Dan for closing remarks.

Thanks, everyone, for joining. More than 7 years of execution has built IREN into a scaled AI platform grounded in real assets, delivery capability and disciplined capital structures with capital access now available at scale and strong customer demand, we're well positioned to bring on new capacity on terms that make sense economically over time. Importantly, having now absorbed the capital requirements associated with our Microsoft deployment, we're able to focus on converting a broader set of advanced customer negotiations into contracted revenue.

When we discuss secured power, we mean fully secured. Power is not a constraint for us. And the ERCOT process is providing greater transparency around which projects are genuinely deliverable. That clarity reinforces the scarcity of firm megawatts and helps customers focus on capacity that can be brought to market with certainty. And IREN we remain focused on execution and on converting our capacity into high-quality customer contracts, and we look forward to updating you as we continue to deliver. Thanks again for your time and continued support. Have a good day. .

Thank you for joining us today. This concludes today's conference call. Thank you for participating. You may now disconnect.

Thank you for standing by, and welcome to the IREN Q1 FY '26 Results Briefing. [Operator Instructions]

I would now like to hand the conference over to Mike Power, VP of Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, and welcome to IREN's Q1 FY '26 Results Presentation. I'm Mike Power, VP of Investor Relations. And with me on the call today are Daniel Roberts, Co-Founder and Co-CEO; Anthony Lewis, CFO; and Kent Draper, Chief Commercial Officer.

Before we begin, please note this call is being webcast live with the presentation. For those that have dialed in by phone, you can elect to ask a question via the moderator after our prepared remarks.

Before we begin, I'd like to remind you that certain statements that we make during the conference call may constitute forward-looking statements, and IREN cautions listeners that forward-looking information and statements are based on certain assumptions and risk factors that could cause actual results to differ materially from the expectations of the company. Listeners should not place undue reliance on forward-looking information or statements, and I'd encourage you to refer to the disclaimer on Slide 2 of the accompanying presentation for more information.

With that, I'll now turn over the call to Dan Roberts.

Thanks, Mike, and thank you all for joining us for IREN's Q1 2026 Earnings Call. Today, we'll provide an overview of our financial results for the first fiscal quarter ending September 30, 2025; highlighting key operational milestones and importantly, discuss how our AI cloud strategy is driving strong growth. We'll then open the call for questions at the end.

So Q1 FY '26 results. Fiscal year 2026 is off to a really good start. We delivered a fifth consecutive quarterly increase in revenues and a strong bottom line. Revenue reached $240 million and adjusted EBITDA was $92 million. Noting, of course, that net income and EBITDA importantly, reflected an unrealized financial gain on financial instruments. This performance reflects our continued -- the team's disciplined execution along with the benefits of having a resilient vertically integrated platform.

Microsoft and the cloud contract. So earlier this week, we announced a $9.7 billion AI cloud contract with Microsoft, which was a defining milestone for our business that underscores the strength and scalability of our vertically integrated AI cloud platform. The agreement not only validates our position as a trusted provider of AI cloud service, but also opens up access to a new customer segment among the global hyperscalers. Under this 5-year contract, IREN will deploy NVIDIA GB300 GPUs across 200 megawatts of data centers at our Children's campus. The agreement includes a 20% upfront prepayment, which helps support capital expenditures as they become due through 2026. The contract is expected to generate approximately $1.94 billion in annual recurring revenue.

Beyond the obvious positive financial impact, the contract carries strategic value of significance for us. It not only positions IREN as a contributor towards Microsoft's AI road map, but also demonstrates to the market our ability to serve an expanded customer base, which includes a range of model developers, AI enterprises and now one of the largest technology companies on the planet. As enterprises and other hyperscalers accelerate their AI build-out, we expect that our combination of power, AI cloud experience and execution capability will continue to position us as a partner of choice.

Looking ahead, we're executing now on a plan that will see our GPU fleet scale from 23,000 GPUs today up to 140,000 GPUs by the end of 2026. When fully deployed, this expansion is expected to support in the order of $3.4 billion in annualized run rate revenue. Importantly, this expansion leverages just 16% of our 3 gigawatts in secured power, leaving ample capacity for future expansion.

With that overview in mind, let's turn to the next section, a closer look at our AI cloud platform and how we're positioned to scale in the years ahead. So as I alluded to earlier, a key driver of IREN's competitive advantage in AI cloud services is our vertical integration. We develop our own greenfield sites, engineer our own high-voltage infrastructure, build and operate our own data centers and deploy our own GPUs. Simply put, we control the entire stack from the substation all the way down to the GPU. We believe strongly that this end-to-end integration and control is a key differentiator that positions us for significant growth.

This model of vertical integration eliminates dependence on third-party colocation providers and most importantly, removes all counterparty risk associated. This allows us to commission GPU deployments faster with full control over execution and uptime. For our customers, this translates into scalability, cost efficiency and a superior customer service with tighter control over performance, reliability and delivery milestones, driving tangible value and certainty. For those reasons, our customers, including Microsoft, view IREN as a strategic partner in delivering cutting-edge AI compute, recognizing our deep expertise in designing, building and operating a fully integrated AI cloud platform.

On that note, we're excited to announce a further expansion of our AI cloud service, targeting a total of 140,000 GPUs by the end of 2026. This next phase includes the deployment of an additional 40,000 GPUs across our Mackenzie and Canal Flats campuses, which are expected to generate in the order of $1 billion in additional ARR. When combined with the $1.9 billion expected from the Microsoft contract and $500 million from our existing 23,000 GPU deployment, this expansion provides a clear pathway to approximately $3.4 billion in total annualized run rate revenue once fully ramped.

Importantly, this incremental 40,000 GPU build-out will be executed in a highly capital-efficient manner through leveraging existing data centers. While we have not yet purchased GPUs for the deployment, we continue to see strong demand for air-cooled variants of NVIDIA's Blackwell GPUs, including both the B200 and the B300. And given their efficient deployment profile, we expect these to form the basis of this expansion.

That said, we will continue to monitor customer demand closely and pursue growth in a disciplined, measured way. This full expansion to 140,000 GPUs will only require about 460 megawatts of power, representing roughly 16% of our total secured power portfolio. This leaves substantial optionality for future growth and importantly, continued scalability across our portfolio. The key takeaway here is that we have substantial near-term growth being actively executed upon, but also have significant and additional organic growth ahead of us.

Turning now to Slide 8, which highlights the British Columbia data centers supporting our expansion to 140,000 GPUs. At Prince George, our ASIC to GPU swap-out program is progressing well. The same process will soon extend to our Mackenzie and Canal Flats campuses, where we expect to migrate ASICs to GPUs with similar efficiency and speed. Together, these sites are allowing us to fast track our growth in supporting high-performance AI workloads, scaling it into what is becoming one of the largest GPU fleets in North America.

Turning to Childress, where we are now accelerating the construction of Horizons 1 to 4 to accommodate the phased delivery of NVIDIA GB300 NVL72 systems for Microsoft. We've significantly enhanced our original design specifications to meet hyperscale requirements and also further ensure durable long-term returns from our data center assets. The facilities have been engineered to Tier 3 equivalent standards for concurrent maintainability, ensuring continuous operations even during maintenance windows.

A key feature of this next phase is the establishment of a network core architecture capable of supporting single 100-megawatt super clusters, a unique configuration that enables high-performance AI training for both current and next-generation GPUs. We're also incorporating flexible rack densities ranging from 130 to 200 kilowatts per rack, which allows us to accommodate future chip generations and the evolving power and density requirements without major structural upgrades.

While these design enhancements have resulted in incremental cost increases, they provide long-term value protection, enabling our data centers to support multiple generations and reduce recontracting risk typically associated with lower spec builds. In short, we're building Childress not just for today's GPUs and the Microsoft contract in front of us, but also for the next generations of AI compute.

Beyond the accelerated development of Horizons 1 through to 4, the remaining 450 megawatts, as you can see in the image on screen of secured power Childress provides substantial expansion potential for future horizons numbered 5 through to 10. Design work is underway to enable liquid cooled GPU deployments across the entire site, positioning us to scale seamlessly alongside customer demand.

Finally, turning to Sweetwater, our flagship data center hub in West Texas, which has been somewhat overshadowed in recent months by the activity in Childress and Canada. At full build-out, Sweetwater will support up to 2 gigawatts, 2,000 megawatts of gross capacity, all of which has been secured from the grid. As shown in the chart, this single hub rivals and in most cases, exceeds the entire scale of total data center markets today. While the recent headlines have naturally been dominated more about our AI cloud expansion at other sites, Sweetwater is a pretty exciting platform asset, giving us the capability to continue servicing the wave of AI compute demand.

Sweetwater 1 energization continues to remain on schedule with more than 100 people mobilized on site to support construction of what is becoming one of the largest high-voltage data center substations in the United States.

All exciting stuff. With that, I'll now hand over to Anthony, who will walk through our Q1 FY '26 results in more detail.

Thanks, Dan, and thanks, everyone, for your attendance today. Continued operational execution was reflected in another quarter of strong financial performance. Q1 FY '26 marked our fifth consecutive quarter of record revenues with total revenue reaching $240 million, up 20% -- 28% quarter-over-quarter and 355% year-over-year. Operating expenses increased primarily on account of higher depreciation, reflecting ongoing growth in our platform and higher SG&A. The latter primarily driven by a materially higher share price, resulting in acceleration of share-based payment expense and a higher payroll tax expense associated with employees -- $63 million were both significantly up, largely on account of unrealized gains on prepaid forward and cap call transactions entered into in connection with our convertible note financings. Adjusted EBITDA was $92 million, reflecting continued margin strength, partially offset by that higher payroll tax of $33 million accrued in the quarter on account of strong share price performance.

Turning now to our recently announced AI cloud partnership with Microsoft. As Dan mentioned, this is a very significant milestone for IREN. It not only delivers strong financial returns, but also creates a significant long-term strategic partnership for the business.

Focusing on the financials. The $9.7 billion contract is expected to deliver approximately $1.9 billion in annual revenue once the 4 phases come online with an estimated 85% project EBITDA margin. This strong margin, which reflects our vertically integrated model incorporates all direct operating expenses across both our cloud and data center operations supporting the transaction, including power, salary and wages, maintenance, insurance and other direct costs. These cash flows deliver an attractive return on the cloud investment, i.e., the $5.8 billion CapEx for the GPUs and ancillaries after deducting an appropriate internal colocation charge, ensuring that the project delivers robust cloud returns as well as an attractive return on our long-term investment in the Horizon data centers, which will deliver returns for many years into the future.

The transaction has also a number of features that allow us to undertake the transaction in a capital-efficient way. Firstly, the payments for the CapEx are aligned with the phased delivery of the GPUs across the calendar year '26 as we deliver those 4 phases. Secondly, the $1.9 billion in customer prepayments being 20% of total contract revenue, paid in advance of each tranche provides funding for circa 1/3 of the funding requirement at the outset.

Thirdly, the combination of the latest generation of GPUs and the very strong credit profile of Microsoft should allow us to raise significant additional funding secured against the GPUs and the contracted cash flows on attractive terms. While the final outcome will be subject to a range of considerations and factors, we are targeting circa $2.5 billion through such an initiative. And depending on final terms and pricing, there is meaningful upside to that, noting again the very high quality of our counterparty.

We also have a range of options available to fund the remaining $1.4 billion, including existing cash balances, operating cash flows and a mix of equity convertible notes and corporate instruments. On that note, turning more generally to CapEx and funding.

We continue to focus on deepening our access to capital markets and diversifying our sources of funding. We issued $1 million in 0 coupon convertible notes during October, which was extremely well supported. And we also secured an additional $200 million in GPU financing to support our AI cloud expansion in Prince George, bringing total GPU-related financings to $400 million to date at attractive rates.

Taking into account recent fundraising initiatives, our cash at the end of October stood at $1.8 billion. Our upcoming CapEx program, which includes the construction of the Verizon data centers for the Microsoft transaction will be met from a combination of the strong starting cash position, operating cash flows, the Microsoft prepayments, as just noted, and other financing streams that are underway. These include the GP financing facilities that we discussed as well as a range of other options under consideration from other forms of secured lending against our fleet of GPUs and data centers through to corporate level issuance, whilst maintaining an appropriate balance between debt and equity to maintain a strong balance sheet.

With that, we'll now turn the call over to Q&A.

[Operator Instructions] The first question today comes from Nick Giles from B. Riley Securities.

I want to congratulate you on this significant milestone with Microsoft. This was really great to see. I have a two-part question. Dan, you mentioned strategic value, and I was first hoping you could expand on what this deal does from a commercial perspective. And then secondly, I was hoping you could speak to the overall return profile of this deal and how you think about hurdle rates for future deals.

Sure. Thanks, Nick. I appreciate the ongoing support. So in terms of the strategic value, I think undoubtedly, proving that we can service one of the largest technology companies on the planet has a little bit of strategic value. But below that, the fact that this is our own proprietary data center design, and we've designed everything from the substation down to the nature of the GPU deployment, and that has been deemed acceptable by a $1 trillion company, I think that's got a bit of strategic value, both in terms of demonstrating to capital markets and investors that we are on the right track, but also importantly, in terms of the broader customer ecosystem and that validation. And look, we've seen that play out over the days since the announcement. In terms of hurdle rates and returns. I think it's worth Anthony, if you can to jump into this. I think it's fair to say that IRRs, hurdle rates and financial models have dominated our lives for the last 6 weeks. So there's probably a little bit we can outline in this regard.

Sure. Thanks, Dan, and thanks for the question. The -- yes, just in terms of -- yes, the returns on the transaction, obviously, as I noted in the introductory comments, we -- when we look at the cloud returns, we obviously take away what we think to be an arm's length colocation rate, so effectively charge the deal for the cost of reaching the data center capacity. After we take that into account on an unlevered basis and assuming that there are 0 cash flows or RV associated with the GPUs after the term of the contract, we expect an unlevered IRR of low double digits. Obviously, we'll be looking to add some leverage to the capital structure for the transaction, as we also discussed. And once we take that target $2.5 billion of additional leverage into account, you're achieving a levered IRR in the order of circa 25% to 30%. Obviously, that is assuming that $2.5 billion package and it also assumes that the remaining funding is coming from equity as opposed to other sources of capital, which we might also have access to.

I'd also note that we said that the -- might well be upside on that $2.5 billion. Obviously, at a $3 billion leverage package against the GPUs on a secured financing package, you could see those -- that levered return increase by circa 10%. In terms of the RV, we've obviously -- in those numbers, we're just reflecting 0 economic value in the GPUs at the end of the term. If, for example, you were to assume a 20% RV, obviously, that has a material impact. Unlevered IRRs would increase to high teens and your levered IRRs would be somewhere between 35% to 50% depending on your leverage assumptions.

Yes. I think maybe just to jump in as well. Thanks, Anthony. That's all absolutely correct. And there are a lot of numbers in there, which is demonstrative of the amount of time we spent thinking about IRRs. So I think just to reiterate a couple of points. One is we've clearly divided out our business segments into stand-alone operations for the purposes of assessing risk return against a prospective transaction. So to be really clear, all of those AI cloud IRRs assume a colocation charge. So they assume a revenue line for our data centers. So our data centers, we've assumed to earn internally $130 per kilowatt per month escalating, which is absolutely a market rate of return, particularly considering the first 5 years is underwritten by a hyperscale credit. So that's probably the first point I'd make.

But -- it's also really important to mention that we've optimized elsewhere. So the 76,000 GPUs that we've procured for this contract at a $5.8 billion price, Dell have really looked after us to the point where they've got an in-built financing mechanism in that contract, where we don't have to pay for any GPUs until 30 days after they're shipped. So there's further enhancements there. And then the final point I'd reiterate is this 20% prepayment, which I don't believe we've seen elsewhere, accounts for 1/3 of the entire CapEx of the GPU fleet. And I guess we've been asked previously why we would prefer to do AI cloud versus colocation -- as one very single small data point, we are getting paid 1/3 of the CapEx upfront here as compared to having to give away equity -- big chunks of equity in our company to get access to a colocation deal. So we're really pleased to lead us towards that $3.4 billion in ARR by the end of 2026 on returns that are pretty attractive. Yes, it's a good result.

Anthony, Dan, I really appreciate all the detail there. One more, if I could. I was just wondering if you could give us a sense for the number of GPUs that will ultimately be deployed as part of the Microsoft deal. And then as we look out to year 6 and beyond, I mean, can you just speak to any of the kind of future-proofing you've done of the Horizon platform and what can ultimately be accommodated in the long term for future generations of chips?

I'm happy to jump in and take that one, Dan. So in terms of the number of GPUs to service this contract, I draw your attention to some of our previous releases where we've said that each phase of Horizon would accommodate 19,000 GB300s. And obviously, we're talking about 4 phases here with respect to that. In terms of future proofing of the data centers, there are a number of elements to it, but the primary one is that we have designed for rack densities here that are capable of handling well in excess of the GB300 rack architecture.

And to give you specific numbers there, the GB300s are around 135 kilowatts of rack for the GPU racks and our design at the Horizon facilities it can accommodate up to 200 kilowatts of rack. So that is the primary area where we have future-proofed the design. But as Dan also mentioned in the remarks on the presentation, we have enhanced the design in a number of ways, including effectively what is full Tier 3 equivalent concurrent maintainability. So yes, there are a number of elements that have been accommodated into the data centers to ensure that they can continue to support multiple generations of GPUs.

The next question comes from Paul Golding from Macquarie.

Congrats on the deal and all the progress with HPC. I wanted to ask, I guess, just a quick follow-on to the IRR question. Just on our back of the envelope math, it looks like pricing per GPU hour may be on the rise or at the higher end of that $2 to $3 range, assuming full utilization, so presumably potentially even higher. How should we think about the pricing dynamics in the marketplace right now on cloud given the success of this deal? And what seems to be fairly robust pricing? And then I have a follow-up.

Sure.

You go ahead, Dan.

Look, I'll let Kent talk a bit more about the market dynamic, but it is absolutely fair to say that we're seeing a lot of demand. That demand appears to increase month-on-month in terms of the specific dollars per GPU hour, we haven't specified that exactly. However, we have tried to give a level of detail in our disclosures, which allows people to work through that. I think importantly, for us, rather than focusing on dollars per GPU hour, which I think your statement is correct, is focus on the fundamental risk return proposition of any investment. And when we've got the ability to invest in an AI cloud, delivering what is likely to be in excess of 35% levered IRRs against the Microsoft credit, I mean, you kind of do that every day of the week.

Yes. Thanks, Dan. And Paul, with regard to your specific question around demand, we continue to see very good levels of demand across all the different offerings we have. The air-cooled servers that we are installing up in our facilities in Canada lend themselves very well to customers who are looking for 500 to 4,000 GPU clusters and want the ability to scale rapidly. As we've discussed before, transitioning those existing data centers over from their current use case to AI workloads is a relatively quick process, and that allows us to service the growth requirements of customers in that class very well. And case in point, we've been able to precontract for a number of the GPUs that we purchased for the Canadian facilities well in advance of them arriving out of the sites.

And this is something that customers have historically been pretty reticent to do, but that level of demand exists in the market as well as ongoing trust and credibility of our platform with both existing and new customers that is allowing us to take advantage and pre-contract a lot of that away. And then obviously, with respect to the Horizon 1 build-out for Microsoft, this is the top-tier liquid cooled capacity from NVIDIA. We continue to see extremely strong demand for that type of capacity. And the fact that we are able to offer that means that we can genuinely serve all customer classes from hyperscalers, the largest foundational AI labs and largest enterprises with that liquid cooled offering down to top-tier AI start-ups and smaller scale inference enterprise users at the BC facilities.

As a follow-up, as we look out to Sweetwater 1 energization coming up fairly soon here in April. Are you able to speak to any inbound interest you're getting on cloud at that site? I know it's early days just from a construction perspective, maybe for the facilities themselves. But any color there and maybe whether you would consider hosting at that site given the return profile and potential cash flow profile that you would get from engaging in, in the cloud business over a period of time?

Yes. In terms of the level of interest and discussions that we're having, we're seeing a strong degree of interest across all of the sites, including Sweetwater as well. Obviously, very significant capacity available at Sweetwater, as Dan mentioned, with initial energization there in April 2026, which is extremely attractive in terms of the scale and time to power. So I think it's very fair to say that we're seeing strong levels of interest across all the potential service offerings.

As it relates to GPU as a Service and colocation, as previously, we will continue to do what we think is best in terms of risk-adjusted returns. Anthony outlined the risk-adjusted returns that we're seeing in colocations -- sorry, in GPU as a Service specifically at the moment. And as we've outlined over the past number of months, that does look more attractive to us today. But as we continue to see increasing supply-demand imbalance within the industry, that may well feed through into colocation returns where it makes sense to do that in the future. But as it stands today, certainly, the return profile that we're seeing in GPU as a Service, we think is incredibly attractive.

The next question comes from Brett Knoblauch from Cantor Fitzgerald.

On the $5.8 billion, call it, order from Dell, can you maybe parse out how much of that is allocated to GPUs and the ancillary equipment? And on the ancillary equipment, say you wanted to retrofit the Horizon data centers with new GPUs in the future, do you also need to retrofit the ancillary equipment?

So out of that total order amount, I mean it's fair to say the GPUs constitute the vast majority of it, but there is some substantial amounts in there for the back-end networking for the GPU clusters. which is the top-tier InfiniBand offering that's currently available. In terms of future proofing, we'll have to see how much of that equipment may or may not be reusable for future generations of GPUs. As I was referring to earlier, the vast majority of our data center equipment and the way that we have structured the rack densities within the data center mean that the data center itself is future-proofed. But in terms of the specific equipment for this cluster, it remains to be seen whether that will be able to be reused.

Perfect. And then on the -- maybe the new 40,000 order that sounds like kind of be plugged in, in Canada. You talked about maybe a very efficient CapEx build for those data centers. Can you maybe elaborate a bit more on that? I know when the AI craze maybe first got started 18 months ago, you guys flagged that you guys are running GPUs up and sure that you built for less than $1 million a megawatt. Are we closer to that number for this? Or are we just well below maybe what the Horizon 4 cost per megawatt basis?

So in terms of the basic transition of those data centers over to AI workloads, it is relatively minimal in terms of the CapEx that is required. The vast majority of the work is removing ASICs, removing the racks that the ASICs sit on and replacing those with standard data center racks and PDUs, so the power distribution units. That can accommodate the AI servers. So that is relatively minimal. As we've discussed before, it's a matter of weeks to do that conversion. And from a CapEx perspective, it is not material. The one element that may be more material in terms of that conversion is adding redundancy if required to the data centers that would typically cost around $2 million a megawatt if we need to do that. But obviously, in the context of a full build-out like we're seeing of liquid cooled capacity at Horizon, it's extremely capital and CapEx efficient.

The next question comes from Darren Aftahi from ROTH Capital Partners.

Congrats on the Microsoft deal as well. To start, with Microsoft, was colocation ever on the table with them? Did they come to you asking for AI cloud? Or how did those negotiations sort of fall out?

Just think about the best way to answer this. So we've been talking to Microsoft for a long period of time and the nature of those conversations absolutely did evolve over time. Is there a preference for the cloud deal? Possibly. But at the end of the day, we want to focus on cloud, and that was the transaction we were comfortable with. So conversations really focused around that over the last 6 weeks or so.

I think if I may, I'd talk more generically around these hyperscale customers because obviously, we weren't just talking to Microsoft. I think there probably is a stronger preference from those to be looking at more colocation and infrastructure deals rather than cloud deals. But also is the case that there's an appetite for a combination. So it may be that we do some colocation in the future.

Yes, I think different hyperscalers have different preferences. We'll entertain them all. But given the nature of the deal we did with a 20% prepayment funding 1/3 of CapEx and a 35% plus equity IRR, we're feeling pretty good about pursuing AI cloud.

Got it. And just as a follow-up with the rest of Childress, is there any significance to the size of the Microsoft deal starting at 200 megawatts? Do they have interest in the rest of the campus? Have you talked to them about that yet?

So again, I'm going to divert the question a little bit because we've got some pretty strong confidentiality provisions. So let me talk generically. There is appetite from a number of parties in discussing cloud and other structures well above the 200 megawatts that's been signed with Microsoft.

The next question comes from John Todaro from Needham.

Congrats on the contract. I guess just one on that as we dig a little bit more in, any kind of penalties or anything related to the time line of delivering capacity? Just wondering if there's guardrails around that. And then I do have a follow-up on CapEx.

There's always a penalty, whatever you do in life, if you don't do what you promise you're going to do. So we're very comfortable with the contractual tolerances that have been negotiated, the expected dates versus contractual penalties and other consequences. I can't comment more specifically beyond that on this call. But the other thing I would reiterate is we have never ever missed a construction or commissioning date in our life as a listed company. So I think you can take a lot of comfort that if we've put something forward to Microsoft and agreed it there and if we put something forward to the market, our reputations are on the line, our track record is on the line, we're going to be very confident we can deliver it and potentially even exceed it.

Got it. Understood. And then just following up on the CapEx. That $14 million to $16 million on the -- I think it was the data center side. Just wondering if there's anything kind of additional in there that would get it north of the colo items other folks are talking about, if maybe there's some networking or cabling included in that or any contribution from tariffs are being considered there?

To give some additional color there. So yes, in terms of networking, et cetera, again, as Dan mentioned in his presentation earlier. This is designed -- the Horizon campus is designed to be able to operate 100-megawatt super clusters. Now that does raise a significant level of additional infrastructure that is required over being able to deliver smaller clusters. And so certainly, some of the costs that are in the number that you mentioned are related to the ability to do that. And that will not necessarily be a requirement of every customer moving forward. So that probably is an element that is somewhat unique.

The next question comes from Stephen Glagola from Jones Trading.

On your British Columbia GPUs, can you maybe just provide an update on where you guys stand with contracting out the remaining 12,000, I believe, GPUs of the initial 23,000 batch? And are you seeing any demand for your bare metal offering in BC outside of AI native enterprises?

Yes. Happy to give an update there. We previously put out guidance a couple of weeks ago that we had contracted 11,000 out of the 23,000 that were on order. Subsequent to that, we have contracted a bit over another 1,000 GPUs. And primarily the ones that are not yet contracted are the ones that are arriving latest in terms of delivery time lines. As I mentioned earlier, we are seeing an increased appetite from customers to precontract. But these are GPUs that are a little further out in terms of delivery schedules relative to the ones that have already been contracted.

Having said that, we continue to see very strong levels of demand, and we're in late-stage discussions around a significant portion of the capacity that has not yet been contracted and continue to see very good demand leading into the start of next year as well and are receiving an increasingly large number of inbounds from a range of different customer classes. So you mentioned AI natives. Yes, that has been a portion of the customer base that we've serviced previously. But we are also servicing a number of enterprise customers on an inference basis -- so it is a pretty wide-ranging customer class that we're servicing out of those British Columbia sites.

The next question comes from Joseph Vafi from Canaccord Genuity.

Congrats from me too on Microsoft. Just maybe, Dan, if you could kind of walk us through what you were thinking in your head. Clearly, some awesome IRRs here on the Microsoft deal. But how are you thinking about risk on a cloud deal here versus a straight colo deal, which probably wouldn't have had the return, but maybe the risk profile may be lower there? And then I have just a quick follow-up.

Thanks, Joe. Look, it's funny. I actually see risk very differently. So we've spoken about colocation deals with these hyperscalers. And if you model out a 7% to 8% starting yield on cost and run that through your financial model, what you'll generally see is that you'll struggle to get your equity back during the contracted term, and then you're relying on recontracting beyond end of that 15-year period to get any sort of equity return. So in terms of risk, I would argue that there's a far better risk proposition implicit in the deal that we've signed and going down the cloud route. And then for the shorter-term contracts on the colo side where you may not have a hyperscale credit, you're running significant GPU refresh risk against companies that don't necessarily have the balance sheet today to support confidence in that GPU refresh. So again, we think about it in business segments, we think about our data center business has got a great contract internally linked to Microsoft as a tenant. And that data center itself is future-proofed accommodating up to 200-megawatt rack densities.

And it's also the case that in 5 years, the optionality provides further downside protection. So upon expiry of the Microsoft contract, maybe we can run these GPUs for additional years, which we've seen with prior generations of GPUs like the A100s. But assuming that isn't the case, we've got a lot of optionality within that business. We could sign a colocation deal at that point. We could relaunch a new cloud offering using latest generation GPUs. So my concern with these colocation deals is what you're doing is you're transferring an interest or an exposure to an asset that is inherently linked to this exponential world of technology and demand and the upside that, that may entail and you're swapping that for a bond position in varying degrees of credit with the counterparties. So if you're swapping an asset for a bond exposure to a $1 trillion hyperscaler and you're kind of hoping you might get your equity back after the contracted period. I mean that's one way to look at it. If you're swapping your equity exposure for a bond exposure in a smaller Neo cloud without a balance sheet, then is that a good decision for shareholders? We just haven't been comfortable.

I get it, Dan. I mean we've run some DCF here and on some colo deals here in the last couple of months. And there's a lot to be learned when you do it. There's no doubt. And then just on this prepayment from Microsoft, I know you've got some strong NDAs here, but kind of a feather in your cap on getting that much in a prepayment. Any other -- anything else to say on how -- maybe your qualifications or how Microsoft perhaps and you came to the agreement to prefund the GP purchases out of the box?

Look, yes, getting the 1/3 of the CapEx funded through a prepayment from the customer is fantastic from our perspective, and we're super appreciative for Microsoft coming to the table on that. And what that allows us to do is to drive a really good IRR and return to equity for our shareholders. And again, linking back to what Anthony said earlier, we expect 35% equity IRRs from this transaction accounting after an internal data center charge. So trying to create that apples-and-apples comparison for a Neo cloud that has an infrastructure charge. Even after that, we're looking at 35% plus. And also what's really important to clarify is the equity portion of that IRR we have assumed is funded with 100% ordinary equity, which given our track record in raising convertibles, given the lack of any debt at a corporate level, is probably conservative again. So from a risk-adjusted perspective, linked to a $1 trillion credit and the ability to fund it efficiently, I mean we're really happy with the transaction. And yes, hopefully, there's more to come.

The next question comes from Michael Donovan from Compass Point.

Congrats on the progress. I was hoping you could talk more to your cloud software stack and the stickiness of your customers.

Yes, I'm happy to take that one. Yes, to date, the vast majority of our customers have required a bare metal offering, and that is their preference. These are all highly advanced AI or software companies like a Microsoft. They have significant experience in the space, and they want the raw compute and the performance benefits that, that brings, having access to a bare metal offering and then being able to layer their own orchestration platform over the top of that. So that has been by design that we have been offering a bare metal service. It lends itself exactly to what our customers are looking for.

Having said all of that, we obviously are continuing to monitor the space, continuing to look at what customers want. And we are certainly able to go up the stack and layer in additional software if it is required by customers over time. But today, as I said, we haven't really seen any material levels of demand for anything other than the bare metal service that we're currently offering.

And I think maybe just to add to that, Kent, if you step back and think about it, you contract in with some of the largest, most sophisticated technology companies on the planet that want to access to our GPUs to run their software. It's kind of upside down world to then turn around and say, "Oh, we'll do all the software and operating layer." Like clearly, they're in the position they are because they have a competitive advantage in that space. They're just looking for the bare metal. I think as the market continues to develop over coming years, it may be the case that if you want to service smaller customers that don't have that internal capability or budget, then yes, maybe you will open up smaller segments of the market. But for a business like ours that is pursuing scale and monetizing a platform that we spent the last 7 years building, it's very hard to see how you get scale by focusing on software, which is, I think everyone generally accepts is going to be commoditized anyway in coming years as compared to just selling through the bare metal and letting these guys do their thing on it.

That makes sense. I appreciate that. You mentioned design works are complete for a direct fiber loop between Sweetwater 1 and 2. How should we think about those two sites communicate with each other once they're live?

Yes. I think really the best way to think about it is it just adds an additional layer of optionality as to the customers that would be interested in that and how we contract those projects. There are a number of customers out there who are looking particularly for scale in terms of their deployments. And obviously, being able to offer 2 gigawatts that can operate as an individual campus even though the physical sites are separated is something that we think has value, and that's why we have pursued that direct fiber connection.

At this time, we're showing no further questions. I'll hand the conference back to Dan Roberts for any closing remarks.

Great. Thanks, operator. Thanks, everyone, for dialing in. Obviously, it's been an exciting couple of months and particularly last week. Our focus now turns to execution to deliver 140,000 GPUs through the end of 2026, but also continuing the ongoing dialogue with a number of different customers around monetizing the substantial power and land capacity we've got available and our ability to execute and deliver compute from that. So I appreciate everyone's support. I look forward to the next quarter.

Good day, and thank you for standing by. Welcome to the IREN FY 2025 Results Conference Call. At. Please be advised that today's conference is being recorded.

[Operator Instructions]

I would now like to hand the conference over to your speaker today, Mike Power, VP, Investor Relations.

Thank you, operator. Good afternoon, and welcome to IREN's FY 2025 Results Presentation. My name is Mike Power, VP of Investor Relations. And with me on the call today are Daniel Roberts, Co-Founder and Co-CEO; Belinda Nucifora, CFO; Anthony Lewis, Chief Capital Officer; and Kent Draper, Chief Commercial Officer.

Before we begin, please note that this call is being webcast live with a presentation. For those that have dialed in via phone, you can elect to ask a question via the moderator after our presentation. Before we begin, I would like to remind you that certain statements that we make during the conference call may constitute forward-looking statements, and IREN cautions listeners that forward-looking information and statements are based on certain assumptions and risk factors that could cause actual results to differ materially from the expectations of the company. Listeners should not place undue reliance on forward-looking information or statements. Please refer to the disclaimer on Slide 2 of the accompanying presentation for more information.

Thank you, and I will now turn the call over to Dan Roberts.

Thanks, Mike. Good afternoon, everyone, and thank you for joining our FY 2025 earnings call. So today, we will provide an update on our financial results for the fiscal year ended June 30, along with some operational highlights and strategic updates across our business verticals, we'll then end the call with Q&A. So FY '25 was a breakout year for us, both operationally and financially. We delivered record results across the board, including 10x EBITDA growth year-on-year and strong net income, which Linda will discuss shortly. Operationally, we scaled at an unprecedented pace. We increased our contracted grid-connected power by over 1/3 to nearly 3 gigawatts and more than tripled our operating data center capacity to 810 megawatts, all at a time when power, land and data center shortages continue to persist across the industry. We expanded our Bitcoin mining capacity 400% to 50 exahash and in the process, cemented our position as the most profitable large-scale public Bitcoin miner.

At the same time, we made huge strides in AI, scaling GPU deployments to support a growing roster of customers across both training and inference workloads. We also commenced construction of Horizon 1, our first directed chip liquid cooling AI data center and Sweetwater, our 2-gigawatt data center Harvey in West Texas, one of the largest data center developments in the world and a cornerstone of our future growth plans.

These achievements underscore the strength of our execution and the earnings potential of our expanding data center and compute platform. We expect this momentum to carry into FY '26 and beyond as we realize the revenue potential of our 50 exahash platform and advance our 4 AI growth initiatives.

So reflecting on current operations. Our AI cloud business is scaling rapidly with more than 10,000 GPUs online or being commissioned in the coming months backed by multiple tranches of non-dilutive single-digit GPU financing, this rollout will feature next-generation liquid-cooled Nvidia GB300-NVL72 systems at our Prince George campus. This strengthens our position as a leading AI cloud provider and a newly designated NVIDIA preferred partner. In parallel, while we have paused meaningful mining expansion, our 50 exahash platform continues to generate meaningful cash flow, over $1 billion a year in annualized revenue at the current economics, supporting our continued growth in AI.

Together, these operations are approaching annualized revenue of $1.25 billion. That's the scale we're delivering today. However, the clear visibility to continued growth ahead is something we're quite excited about on that note. Our strategy is focused on scaling across the full AI infrastructure stack from grid connected transmission line all the way down to the digital well compute.

With a strong track record building power dense data centers and operating GPU workloads, we are uniquely positioned to serve AI customers end-to-end from cloud services to turnkey location, capturing a broad and growing addressable market. Beyond the current expansion to 10,000 GPUs, our 160 megawatts of operating capacity in BC provides a path to deploy more than 60,000 NVIDIA GB-300s with Horizon 1, then offering the potential to scale that further to nearly 80,000. As we continue to assess market demand, this gives line of sight to billions in annualized revenue from our AI cloud business alone.

In terms of AI data centers, we're progressing 3 major data center projects to drive this revenue growth as well as provide scope for future expansion. At Prince George in BC, we're continuing to transition existing capacity from Bitcoin to AI workloads with retrofits for air cooled GPUs and the construction of a newly announced liquid cooled data center underway to support our GB-300 deployment. At Childress, Horizon 1 continues to remain on schedule for Q4 2025. Given strong demand signals, we have also begun site works and long lead procurement for Horizon 2, a secured liquid cooled facility. Together, these projects can support over 38,000 NVIDIA GB-300s. At Sweetwater, our flagship 2-gigawatt data center Harbes West Texas, SweetWater One remains on track for energization in April 2026. Construction is progressing well with key long-lead equipment, either on-site already or on order. Upgrades to the utility substation have now commenced. So in summary, we've delivered record performance this year. We've got a clear AI growth path with near-term milestones. And most excitingly, we continue to position our platform ahead of the curve to monetize substantial opportunities in the AI infrastructure and compute markets.

I'll now hand over to Belinda, who will walk through the FY '25 results in more detail.

Thank you, Dan. Good morning to those in Sydney, and good afternoon to those in North America. As noted in our recent disclosures, we've completed our transition to a U.S. domestic issuer status from the first of July of this year. And as such, we've reported our full year results for the period ended June 30, 2025 under U.S. GAAP and the required SEC regulations. For the fourth quarter of FY '25, we delivered a record revenue of $187 million being an increase of $42 million from the previous quarter, primarily due to the record mining bitcoin of $180 million as we operate at 50 exahash. During the quarter, we also delivered AI cloud revenue of $7 million. Our Bitcoin mining business continues to perform strongly, supported by best-in-class fleet efficiency at 15 joules per terahash and low net power costs being $0.035 per kilowatt hour in Q4. Whilst our operating expenses increased to $114 million, primarily due to overhead and depreciation costs associated with our expanded data center platform and increased Bitcoin mining and GPU hardware, we've delivered a strong bottom line of $177 million.

High-margin revenues from our Bitcoin mining operations were a key driver of this profitability with an all-in cash cost of $36,000 per Bitcoin mined versus an average realized price of $99,000. Noting that these all-in costs incorporate expenses across our entire business, including the AI verticals, underscoring the strength of our platform. We closed the financial year with approximately $565 million of cash and $2.9 billion in total assets, giving us a strong balance sheet to support the next stage of growth.

I'll now hand back to Dan to discuss the exciting growth opportunities that continue for IREN.

Thanks, Melinda. So I think it's fair to say that the market backdrop for our AI cloud business is pretty compelling. Industry reports to demonstrate accelerating enterprise adoption of AI solutions and services with the percentage of organizations leveraging AI in more than 1 business function growing from 55% to 78% in the last 12 months alone. As almost all of us would know demand is accelerating faster than supply. New model development sovereign AI programs and enterprise adoption are driving a step-up in GPU needs and the constraint is infrastructure and compute, not customer interest. Power availability, GPU ready high-density data center capacity remains scarce with customers prioritizing speed to deploy and the ability to scale. IREN is uniquely positioned to meet this demand. Our vertical integration gives us control over the key bottlenecks, significant near-term grid-connected power with data centers engineered for next-generation power dense compute.

This enables accelerated delivery time lines and rapid low-risk scaling. Because we own and operate the full end-to-end stack, we are able to deliver superior customer service, tighter control over efficiency, uptime and service quality translating directly into a better customer experience for our customers. We are leading our service with a bare metal service because it gives sophisticated developers, cloud providers and hyperscale is what they want most, direct access to compute and the flexibility to bring their own orchestration.

As and when customer needs evolve, we have the flexibility to layer in software solutions to provide additional options to the customer. Our new status as an NVIDIA preferred partner is helpful in that regard. It enhances supply access and helps broaden our customer pipeline, supporting expansion across both existing relationships and new end users, platforms and demand partners. So the market is large. It's accelerating. Supply is constrained, and we have the platform to meet market demand for AI cloud and meet that reasonably quickly.

That is why we're immediately scaling to more than 10,000 GPUs, but also now importantly, focusing on what comes next. So our 10,000 GPU expansions underway. With it, we will be positioned at the front of the Blackwell demand curve, delivering first-to-market benefits. We saw this with our initial B200 deployment several weeks ago. Upon commissioning, it was immediately contracted on a multiyear basis. Importantly, we are funding growth in a CapEx-efficient way. In the past week alone, we have secured 2 new tranches of financing which have funded 100% of the purchase price of new GPUs at single-digit rates. Anthony will touch on this shortly as well as what's next.

In terms of revenue, these GPUs will be delivered and progressively commissioned over the coming months, targeting $200 million to $250 million of annualized revenue by December this year. Approximately 1 exahash of ASICs will be displaced as a result, which we plan to reallocate to sites with available capacity, minimizing any impact to the overall 50 exahash installed hash rate.

Finally, we also expect the strong margin profile of our AI cloud business to continue, underpinned by low power costs, but importantly, full ownership of our AI data centers. eliminating any third-party co-location fees from our cost base. Our Prince George Campus will anchor this next phase of our AI cloud growth. So as I alluded to earlier, we are pleased to announce today that construction is well underway on a new 10-megawatt liquid cooled data center at Prince George, designed to support more than 4,500 Blackwell GB-300 GBUs. Following this buildout, half of Prince George's capacity will now be dedicated to AI cloud services. There has been clear runway to double capacity to more than 20,000 GPUs at this site alone. Procurement is also in progress to equip every GPU deployment at Prince George with backup generators and UPS systems Beyond Prince George, Mackenzie and Canal Flats. Our data center [indiscernible] business in each of these locations create an even larger opportunity with powered shells, existing and designed to the same architecture as Prince George, these sites offer a straightforward and replicable pathway to more than 60,000 GB-300s. Horizon 1 and our broader portfolio of data center sites in Texas opens up a further path to continued AI cloud growth.

It's fair to say we're incredibly excited by the AI cloud opportunity. It's a business line that many are simply unable to pursue due to the significant technical expertise and requirements involved with 2- to 3-year payback periods and the low-cost GPU financing structures we are securing, we see this as a highly attractive pathway to continue compounding shareholder value.

Our ability to build and operate world-class AI services, all the way down from the transmission line down to the compute layer uniquely positions IREN at the forefront of this digital AI transformation.

Now on to the major projects driving our AI expansion. Childress continues to show strong on-the-ground momentum with Horizon 1 construction, progressing according to schedule and remaining on track for this year. As you can see in the progress photos, the data center buildings are nearing completion and the installation of the liquid cooling plant on the south side of the hall is underway.

Based on customer feedback, we have also upgraded certain specifications, including introducing full Tier 3 equivalent redundancy across all critical power and cooling systems. Due to the expected timing gap before NVIDIA's Ruben GPUs are available, we have also reconfigured the design to be able to accommodate a wider range of [indiscernible] entities while preserving the flexibility to accommodate next-generation systems when they are available. Even with these adjustments, we expect to remain a very competitive build cost target, reflecting the efficiencies of our in-house design, procurement and construction model. Finally, we are also moving ahead with certain tenant scope work to derisk delivery time lines and provide additional flexibility, including the potential to monetize the capacity via our own AI cloud service.

In that regard, engagement remains active with both hyperscaler and non-hyperscale customers across both cloud and colocation opportunities, site visits, diligence, commercial discussions, documentation ongoing. Building on this strong customer traction at Horizon 1 and general overall market momentum, we are pleased to announce that we've commenced early works and long lead procurement for Horizon 2, a potential second 50-megawatt IT load liquid-cooled facility at Childress.

Together, Horizons 1 and 2 will have capacity to support over 38,000 liquid cooled GB-300 creating one of the largest clusters in the U.S. market. In saying that, it's still modest compared to the capacity of our Sweetwater hub, which could support over 600,000 GB-300s, which is a good segue, Sweetwater. So both construction and commercial momentum continues to build at the 1.4 gigawatt Sweetwater 1 site, still scheduled for energization in April 2026.

As you can see in the progress photo, construction of the high-voltage bulk substation is underway and key long lead equipment continues to arrive at site. On the commercial front, we're advancing discussions with prospective customers for different structures. The campus is inherently flexible by design so we can meet demand across the entire AI infrastructure stack, powered shells with partners who want to self-operate, turnkey colocation for customers seeking speed and cloud services for those who would like us to run it end to end.

While we have a multitude of other exciting growth opportunities preceding this, Sweetwater's combination of scale, certainty and flexibility positions it as yet another growth engine for IREN in the accelerating wave of AI compute. So where do we sit today? Industry estimates call for more than 125 gigawatts of new AI data center capacity over the next 5 years with hyperscale CapEx forecast supporting the credibility of that trajectory. Yet, as most of us know, existing green capacity is well documented, has been far from sufficient to meet this demand.

Against that backdrop, we have expanded our secured power capacity more than 100x since IPO. We've built over 810 megawatts of operational next-generation data centers. In the process, demonstrating our ability to not only secure our valuable powered land, but also deliver next-generation data centers and compute at scale in some of the most demanding markets. It's a really exciting time for the industry, and it's a really exciting time for us. With that, hopefully providing a reasonably comprehensive overview of the opportunity in front of us, I'll hand over to our newly appointed Chief Capital Officer, Anthony Lewis, to discuss financing.

Thanks, Dan, and good morning or good evening, everyone, as the case may be. This slide highlights how we are funding growth across our verticals through a combination of strategic financings and strong cash flows from existing operations. To the right -- the table to the right, which many of you will be familiar with shows the illustrative cash flows from our existing Bitcoin mining operations. At the current network hash rate and 115,000 Bitcoin price, we show over $1 billion in mining revenue. And after subtracting all costs and overheads of our entire business, we arrive at close to $650 million of adjusted EBITDA. There is then a further $200 million to $250 million of annualized revenue on top of this expected to come from the AI cloud business expansion with an increasing contribution from that business over time. There is clearly some sensitivity to the relevant assumptions here, but the key message is we expect significant operating cash flow to invest in our growth initiatives over a range of operating conditions with our position enhanced by low-cost power and best-in-class hardware forwards.

These cash flows, together with existing cash and recent financing initiatives, which I'll touch on shortly, fully fund our near-term CapEx including the cloud expansion discussed with liquid cooling and power redundancy at Prince George taking GPUs to 10,900 completing Horizon 1 and energizing Sweetwater 1 substations.

Let me now turn to our funding strategy more generally. As a capital business -- capital-intensive business growing quickly, we are clearly focused on diversifying our sources of capital so that we maintain a resilient and efficient balance sheet. The $200 million of GPU financings we announced this week are a recent example of that. These transactions had 100% of the upfront GPU CapEx financed allowing us to accelerate the growth flywheel for our AI cloud business and an attractive cost of capital, and they pay down over 2 to 3 years, matching well against the accelerated paybacks on the underlying hardware.

End of lease term options, instructions like these also give us added operational flexibility. We're also seeing strong institutional demand for asset-backed and infrastructure lending in the AI sector. And with our existing portfolio of assets and the growth opportunities in front of us, we think IREN is well placed to access that capital. We are currently advancing a range of financing work streams, which could support further growth. This could include further asset-backed financing, project level and corporate level debt. We've also proven good access to the convertible bond market with 2 well-supported transactions over the course of the financial year and that remains a further sort of funding potential for us. Of course, we'll also be focused on maintaining a prudent level of equity capital as we continue to scale, ensuring continued balance sheet resilience.

So in closing, with the foundation in strong operating cash flow from existing operations and a broad range of capital sources available to us. We feel we are well placed to fund the next stage of growth. With that, we'll now -- I'll turn the call over to Q&A.

[Operator Instructions]

Our first question comes from Paul Golding with Macquarie.

[Audio Gap]

I wanted to ask on efficiency at these sites. I noticed that PUE in the British Columbia sites is down at 1.1, which is a very impressive efficiency ratio versus Sweetwater being about 1.4. Those may be peak numbers as opposed to average. But I was wondering if you could give some color around how that might influence the thought process around rollout or concentration of sites receiving GPUs initially versus others as you think about the efficiency? And then also just along the lines of this infrastructure being developed with PUE at low being cited, how are you thinking about the backup generation [indiscernible] and for the existing pods that you have outstanding. I only asked that question in relation to the on-demand versus contracted customer dynamic and how you're seeing that evolve.

Paul, happy to jump in and take that one. So as you mentioned, across the BC sites, we're operating at a PUE 1.1. That's on an air cooled basis. Once we install the liquid-cooled facilities there, we expect that to be operating on an average slightly higher than that, but still well under 1.2 across the year. At Childress, the Horizon 1 liquid cooled installation, the number that you mentioned is much closer to a peak PUE number, although we actually expect it to be less than 1.4 and the average PUE over the year to be around 1.2. In all cases, I think those are extremely competitive numbers across the industry. We are more led in terms of our deployments across the different sites by what our customers are ultimately demanding within British Columbia, the ability to scale extremely quickly on an air cooled basis has been a significant driver of demand for us. And again, that level is extremely competitive regardless. And so that is where we are seeing some of the primary interest from our customer base. At Horizon 1, that liquid cooled capacity, in particular, is extremely scarce in the industry at the moment and the ability to locate a single cluster of up to 19,000 -- or just over 19,000 GB-300s is significantly attractive and driving high levels of customer interest.

So I think less driven by PUE overall in terms of deployments and more driven by the customer side of the equation. To your question on redundancy, as Dan mentioned in his remarks, we're introducing redundancy across the entire fleet of GPUs that we have in our existing operating business as well as for the new GPUs that we purchased. While we believe that for many of the applications that these customers are used for, it's not necessarily the case that it's required to have redundancy. We have seen some of our customers wanting that redundancy. And for us, we ultimately want to be driven by providing the best customer service, and that's really what's driving us to install that redundancy across the fleet.

And if I could ask one quick follow-up on the GB-300 NVL72 capability that has been incorporated or retrofitted into the original plan for Horizon 1, I believe. If you could just give us any incremental color around what that may have entailed and any impact that may have had on how financing availability or future financing plans may be impacted as you think about incremental cost for that density and in particular, maybe as you plan for Ruben given this preferred partner status now?

Yes. So I think what you're referring to with Dan's comments around introducing flexibility for a wider range of densities. And for us, that actually comes more towards lower densities, so being able to operate at densities that are under what the [indiscernible] rubins would require. So the base design, as we had it, could handle up to 200 kilowatts of rack easily able to accommodate the next iteration of GPUs. But what we're seeing in the market today is that many customers actually want flexibility to be able to operate not only at the right densities for GB-300s, which are around 135 kilowatts a rack, but actually even lower densities to accommodate additional types of compute within the data center infrastructure. And so what we've done is gone back and reworked some of the electrical and mechanical equipment to be able to actually accommodate lower rack densities. So as it relates to accommodating Rubens in the future, no change from our perspective.

Our next question comes from John Todaro with Needham.

Congrats on a very strong quarter. First question on the cloud business. And apologies if I missed this, but just the average duration of the contract, kind of trying to determine with the 3-year payback with the GPUs plus infrastructure, the overlap there with the customer contract division? And then I also follow up on the HPC side of things.

Yes. We've got a range of contract length across our existing asset base today, all the way from 1 month rolling contracts out to 3-year contracts. For the newer gen equipment, including the Blackwell purchases that we've made, we've typically seen demand in slightly longer contract lends whilst those Blackwells are new equipment on the market, and so a good indication of that is the initial portion of our B200 that, as Dan mentioned, as soon IREN were installed, we were able to contract them on a multiyear basis. So we do have contracts across the spectrum, but we are seeing fewer gen equipment often longer-term contracts being available.

Got it. That's great. And then just with the success you're having so far in the cloud business, you could take a step back and think, do we need to sign HPC colo capacity? Would you be more comfortable kind of continuing at this at even a bigger scale? And then as it relates to just kind of thoughts on the CapEx to get you there? Any targeted leverage ratio or a threshold on debt, too.

Yes. We're constantly evaluating the opportunities as it relates to both colocation and cloud. I think we're uniquely positioned in the sense that we are able to take advantage of both opportunities which we think is quite differentiated to a number of others in the industry. They obviously have very different profiles in terms of the risk-adjusted returns. So co-location longer-dated contracts, typically in the range of 5 to 20 years, but lower payback periods often higher than 7 years before you can get your capital back. And in many cases, because of the nature of the debt financing associated with those, there's very little actual cash flow coming out of the business during that finance period whereas cloud shorter-dated contracts, but much stronger margins and shorter overall payback period. So we typically see around 2-year payback periods on the GPUs alone and 3 to 4 years on the GPUs plus data center infrastructure.

So it is something that we're constantly evaluating. And overall, we're looking to maximize risk-adjusted returns across both models. I think you can tell from the comments today, as it stands, we do find the cloud opportunity extremely compelling. Anthony, did you want to touch on the comments around financing?

Thanks, Kent. Yes, obviously, we have a very modest debt servicing requirements today. And I guess as we scale the business, obviously, where those opportunities have developed and the nature of the cash flows and the security of those cash flows will ultimately drive what an appropriate level of leverage is for the business. So the capital structure will continue to evolve as we continue to grow, but we'll obviously be focused on maintaining a strong and resilient balance sheet as well as an efficient cost of capital.

Our next question comes from Darren Aftahi with ROTH.

Congrats on all the progress. A couple if I may. So on Horizon 1 and 2. I guess there's commentary in the press release about what theoretically Horizon 1 could support in terms of GPUs, but you kind of left the door open that there may be other uses. So I'm kind of curious on strategic thinking there. And then on Horizon 2, I think if my math is right, you guys only have 25 megawatts left at Childress and you're talking about, I guess, 50 megawatts of critical load, will you be borrowing from your Bitcoin business to kind of get there? And are there expansion opportunities beyond that? Second question, I guess, on Slide 9, you have one of your demand partners is Fluidstack. I'm more curious on the neo cloud side and maybe that entity in particular, given one of your peers signed a deal with them and another partner there, just kind of what the demand drivers are with Fluidstack in particular?

Thanks, Darren. Appreciate that. So 3 questions I hear in there. Horizon, we mentioned 19,000. It's just takeover based on the NVL 72 configuration, GB-300s. The project has been engineered specifically for liquid cooled GPU. So there is no other use case as an end market other than that in saying that there's a couple of different ways we might monetize that capacity. A is through different types of GPUs. So as we mentioned during the presentation and Kent reiterated, we've now introduced the flexibility to accommodate a wider range of rack entities. We actually discovered building this, that the issue is we're building rack entities that are too dense for where the industry is today.

So we've had to dial it back a little bit. So accommodating lower rack density gives us the ability to accommodate a wider range of different GPUs whilst preserving the ability to service the Virarubins as and when they're released and potentially beyond that. So that's exciting. In terms of monetizing the capacity, there's been co-location versus cloud. So we may buy on operate the 19,000 GPUs, and we're having conversations with a variety of potential partners for that, including hyperscale customers. We're progressing financing work streams in parallel. That's a real option. If the risk return balance is right, as Kent mentioned, then absolutely, we're in a unique position where not many people can build on and operate a cloud service. So we're pursuing that, and we're excited about that. But equally, we're seeing a lot of demand for colocation and that would deliver more of an infrastructure return on capital, and we'll remain open to that structure, but we want to see a risk-return framework that is compelling. And to date, I guess, we haven't yet seen that.

In terms of potentially displacing additional mining capacity. You referenced 25 megawatts potentially being displaced from Childress. Look, that's a cost of business. As we said 7 years ago when we started this business, Bitcoin mining will help bootstrap the business, help us build out the data center capacity as and when higher, better value use cases coming along, then we have the ability and the flexibility to swap those in and monetize our data center capacity differently. In saying that we don't envisage stopping building new data centers. We've got 2 gigawatts of Sweetwater.

So it may simply be the case where we just reallocate capacity across different sites. And perhaps there's some relocated action of mining capacity to sweet water at some point, but that's something we're working through.

Finally, Fluidstack. Yes, we've known the Fluidstack guys for quite some time. We've got a good relationship there. We speak to them. We speak to Google. We know what deals have been done. We look at the deals. As of today, we find a 3-year payback on data center and GPU infrastructure pretty compelling, particularly when Anthony is lining up 100% GPU financing at single-digit interest rates. So look, we'll remain open to colocation opportunities, but the devil is in the detail. And the high level is not always what you end up carrying over a longer period of time, so I might leave that there.

Our next question comes from Joseph Vafi with Canaccord Genuity.

Everyone and congrats on all the progress here in fiscal Q4 and quarter-to-date, really great progress. Just really one question for me, maybe just a 2-part but a single question. I just want to drill down a little bit more on the financing on the black wells I know that you mentioned there's some optionality at the end of the lease financing period. I thought maybe we could kind of go into what you're thinking at the end of those. At the end of the lease financing period, what may just what may be a factor in how do you decide what to do next with those?

And then just as a follow-up, it does seem like at least initially, the -- building your own clusters with this financing does look attractive on a payback and time value of money basis. Just wondering how much financing do you think is available in this market versus the kind of project financing that maybe yourself and others have discussed for for a broader colocation type project.

Yes. Thanks for the question. In terms of the -- you're probably familiar with the various types of leasing structures you can see in the market, some of them are structured as more classic full payout finance leases. Others are sort of more sort of tech rotation style where you have fixed committed lease payments and then you have an FMV option to acquire at the end of -- often capped at a percentage of the day 1 price. So that obviously allows you the flexibility to potentially return the equipment if we wanted to reinvest in, for example, the next generation of GPUs at that time or obviously continue to own and operate the equipment depending on the conditions that we see. Sorry, could you just remind me of the second part of your question?

Just the amount of financing capacity you see out there on the GPU side versus co-location.

Yes, I think they are obviously quite different asset profiles and the amount of, obviously, leverage and the cost of that leverage depends greatly on the specific situations on -- on the cloud side, it's obviously focused on the underlying portfolio of customers, the diversity in the customer mix, the credit quality, the duration of the contracts that will all drive both the sort of pricing and leverage that you can secure. And I guess, similarly on the colocation side, obviously, you can obtain very attractive cost of funds and very meaningful leverage against high-quality offtake such as high-scale offtakes. And as you come down the credit spectrum or the duration of the contract, that will obviously flow through into the cost of the finance and the leverage that you can obtain.

And maybe just to add to that, Anthony. Joe, the two are not mutually exclusive cloud and colocation in the sense that we are arranging these 100% financing lease structures, as Anthony mentioned, over the GPUs, but that doesn't preclude us then financing the asset base and the infrastructure base at a data center level similar to how you would finance the colocation. It just happens to be the case that the colocation partner is an internalized IREN entity. So that market is open. We're talking to a number of a -- vast number of potential providers of capital for that. But as Anthony mentioned, we're looking up and down the entire capital stack to optimize cost of capital at a group level. So you've got these asset level options, but then you've got corporate options as well.

We mentioned the buoyant convertible note market that continues to look quite prospective. We've been prosecuting bond type structures at a corporate level as well. So there's a whole different array and every week, depending on level of demand, our revenue profile, how we're building out different elements of the business, the jigsaw puzzle from a financing perspective kind of falls into place and help support that. So it's that reflective wheel of sources and uses of capital, and that's the benefit of now having Antonio and dedicated full time to optimizing cost of capital while Kent runs around North America looking to deploy it.

Our next question comes from Reggie Smith with JPMorgan.

This is Charlie on for Reggie. Can you talk a bit more about some of the key hires you've made in building out the cloud and colocation businesses and where, if anywhere, there is still some room to go. And then as a follow-up, digging in a bit more on the sales side, can you provide a bit more on how you're getting in front of and winning some of the AI clients that you called out in the slides.

Yes. Happy to jump in there on the resourcing question. So we've been hiring across the stack as Dan made clear at a level of vertical integration that we have, we continue to need resources across all areas, including data center operations, networking, InfiniBand experts, developers on the software side. We also continue to build out our go-to-market function. So that consists of hiring additional sales executives as well as solutions architects, and we're also expanding the marketing team in parallel with that. So there is an ongoing level of hiring across the business to support the additional customer-facing work that we're doing. And sorry, there was a last part to your question that I missed, it was breaking up a little.

Yes. Just more on the sales side, like how you're getting in front of the clients, what are you competing on? Why are they choosing IREN, things like that?

Yes. So we get a mix of inbound and outbound customer demand drivers. We have been active recently in the conference space. So we have been getting out telling our story showing why we are differentiated. As I mentioned, we've been expanding the marketing team and our efforts there to help drive inbound particularly our activities across all social platforms have been ramping over the past 12 months, in particular.

And we're seeing a high degree of interest there. And as that gets out into the public sphere as well as our ongoing provision of cloud services and customer word of mouth, we are starting to see more inbound inquiries as well around both our cloud services platform and the potential co-location platform. So it is a bit of a mix there in terms of what we're seeing.

And I think maybe just to add to that as well. Like this is exactly the point, the whole demand-supply equation in this industry is imbalanced, but there is little supply. So the demand when they need something -- when people need something, they tend to find it particularly when it's cars word-of-mouth through these demand brokers, conferences, existing customers word does get out. And we do have 3 pretty unique competitive advantages compared to other competition around Neo clouds like a scale, we control the infrastructure end-to-end. We can scale up capacity up and down across our existing data center footprint let alone the new footprint and building into that growth.

Performance. vertical integration is really important because it gives us direct oversight of every single layer in the stack. So we've got tighter control of performance, reliability, service and they get higher uptime as a result because there's no colocation partners, there's no SLAs with data centers that restrain and constrain your ability to update GPUs and get your hands on them. And then finally, from a cost perspective, we've got no colocation fees and greater operational efficiency as a result. So we're in a really good spot. And this also translates to sales force and marketing support and general cloud support because we are in the industry, we are doing stuff.

We've got available capacity. There's significant interest in joining IREN because we have capacity to sell as distinct from other providers who have no capacity and salespeople are sitting there with not a lot to do.

Our next question comes from Brett Knoblauch with Cantor Fitzgerald.

Maybe on the cloud services front. Is the strategy to go out and order or purchase GPUs with a customer already in mind? Or are you buying those GPUs and and then trying to find a customer. And then could you maybe just elaborate on the power dynamics per GPU? I think the 19,000 GB-300s for Horizon 1 implies it can be 380 of them per megawatts of critical ICU load so you have like maybe a similar metric or GB-300s or GB-200s if you can provide any color there, that would be helpful as well.

So I might take the first half. Kent, if you in the second half. The prospect of ordering GPUs before or after a contract, this is the nature of the industry where companies want compute, they want it now, but they don't want to wait 2 to 3 months. Do you think about an enterprise that's made the decision. You think about an AI scale-up or start-up that's raised a bunch of capital. Very few companies are in a position where they can plan out and map out a 2- to 3-year time line of GPU needs.

Often it's we need GPUs, we need them for a project, we need them for today. So the world wants on-demand compute -- and we've almost used this as a universal motherhood statement to guide what we do. The world doesn't really want data infrastructure. The world is call wants compute and it wants it now and when it needs it. That's the first element. The second element is I feel like it's groundhog day. We're back in this world, and it takes me back to bitcoin mining where every man and their dog promises certain amounts of capacity online by a certain date, and no one does it. No one hits the schedules, everyone revises them downwards, stretches them out, cost blowouts, et cetera, because the real world is hard dealing with large-scale infrastructure projects, large-scale workforces, complex project delivery, safety, like it takes a lot of work and systems and structures to deliver that.

This is why we're in such a good position. We never missed a milestone of Bitcoin mining, where the most profitable if only profitable Bitcoin miner because we did things properly from the start. And we're now sitting here. And as I said, it's groundhog day with the cloud business, where again, all these companies, neo clouds and otherwise promise capacity online by a certain date, and they really hit it. And as a result, customers got a bit gun shy. So the best thing you can do is to continue ordering the hardware. If it's snapped up [indiscernible] as soon as it's commissioned, that's a pretty good sign that you're doing the right thing. And as and when we install hardware and the sales cycle starts slowing down, then you know, okay, well, maybe we've just got to slow down on the orders. But each incremental order from here is a relatively small portion of our overall risk so we can afford to take it.

And with respect to the power question, yes, we do continue to see the overall power usage per GPU ticking up with each incremental release from NVIDIA and the other manufacturers. I think using some of the examples of the numbers that were presented earlier in the presentation on an air cooled basis for B200s, we can fit over 20,000 GPUs into the Prince George site, which is 50 megawatts. At Horizon 1, 50 megawatts of IT load, you're looking at around 19,000 GB-300s. So yes, it's not exact math there, but it does give you an idea of what we're seeing in terms of the amount of power per GPU going up over time.

Our next question comes from Nick Giles with B. Riley.

I wanted to go back to how the Horizon 1 capacity will be utilized and you're closing in on that 4Q completion. So -- at what point would you make the decision to fill Horizon 1 with your own GPUs versus pursue or colocation dealer. Maybe said differently, and I think Dan alluded to this from a financing perspective, but if you were to fill it with GPU, should we expect that to be the case for the entire capacity? Or could we see you co-locate between your own GPUs and a third party?

Yes. I think that's one of the advantages of where we're at is they're not mutually exclusive options for us. So as we mentioned earlier, we are in a unique position that we can monetize that data center capacity in a number of ways, and it doesn't have to be 1s or 0s. We don't need to do all of it as cloud or all of it is colocation. It could be a combination within Horizon 1.

As Dan mentioned, we've started building out Horizon Again, that gives us significant optionality where we could potentially do Horizon 1 under 1 methodology and 1 type of monetization, Horizon 2 under another, but what we will continue to do over time is try and maximize the risk-adjusted returns for how we monetize the assets. And that may fluctuate over time. We're in an obviously incredibly dynamic industry here. And at different points in time, we may see a very different risk/reward proposition in co-location versus cloud, but we do have significant flexibility as to how we utilize the capacity.

Thanks for that, Kent. Just on the cloud services, you're focused on bare metals today, but I think you did make some comments that you could expand your software offerings or integrate, if needed. What should we be looking for there? Or what would the incremental revenue opportunities be if you were to integrate?

Yes. Today, as Dan mentioned, the vast majority of the customers that we are dealing with, which make up the majority of the compute market. These are highly experienced AI players, hyperscalers, developers. They are, for the most part, demanding bare metal because it actually suits them better to be able to bring their own orchestration layer where we see benefits over time from adding incrementally to the software layer is being able to serve a slightly different customer class, which might be smaller AI start-ups or enterprise customers who are looking for a simple single click spin up, spin down type service. But today, where we see the demand supply imbalance, that bare metal offering that we have has a significant level of demand for it. And so we feel like we're well positioned where we're at today.

I think, again, look, just to reiterate, this notion that software is required and these large sophisticated end users of GPUs want a third-party provider to staple their own software and make them use it like these guys are sophisticated. They just walk compute. They want to run their own stuff. And at the end of the day, software is eating the world. We know that. Software is not difficult to overlay. The large customers don't want your software. They want their own software. And we are hearing it also firsthand from executives and employees at some of these companies that offer their own software, that it's a nightmare because every time the GPUs change, they need to update the software and rewrite it.

And it's this constant evolution of code bugs, rewriting, updating, et cetera, or for an area of the market that, yes, it might seem good as a narrative, but fundamentally and substantially in terms of revenue opportunity is quite small today.

Our next question comes from Stephen Glagola with JonesTrading.

As IREN is now recognized as the preferred cloud partner on NVIDIA's website. I was hoping, Dan and Kent, maybe you could provide more detail on your participation in the DGX cloud [indiscernible] marketplace. And specifically, how do the economics of working through the [indiscernible] marketplace compared to maybe operating your own independent cloud offering? What advantages does IREN get from being on that platform? And any insights into sort of NVIDIA's fee structure or take rate for participants there.

Yes. Happy to give some more color there. So we're not currently participating in the Lepton marketplace. But as an NVIDIA preferred partner, we continue to evaluate platforms like that that could expand how we're able to get customers access to our infrastructure. So it may offer us broader reach into developer communities, simple onboarding. So again, to come back to the previous comments that I made on software may open up some of the smaller areas of the market, with smaller AI start-ups and the enterprise customers who are looking for a simpler solution. So we continue to monitor this. We are seeing an increasing number of these type of offerings coming to market. And for us, we think it will be an additional demand driver for the underlying compute layer that we are providing.

And I could just ask one more on Horizon 1. Is there any -- is the growth of your cloud services business, is that influenced which partners you're willing to consider for colocation potentially at Horizon 1, given arguably they can be competitors?

Yes. I mean it's something that we continue to evaluate in terms of the mix. And I think what you're probably referring to are Neo Cloud customers on the colocation side. Now the the majority of Neo clouds have a very different profile to hyperscalers in terms of colocation. So even within the broader colocation market, there is a significant degree of differentiation. If you think of hyperscalers, they're typically looking for longer-term contracts, often 10 to 20 years, extremely creditworthy but drive a hard bargain in terms of the financials and the economic returns that you're able to achieve with Neo clouds, we often see shorter-term requirement.

So typically, it might be 5 to 15 years, less creditworthy than the hyperscalers. So it's all something that we factor in in terms of that risk-reward element that we discussed earlier. But in terms of -- because we have heard from a number of people whether the fact that we're offering a cloud service limits our ability to do co-location, I would actually say quite the opposite. Most of the colocation customers that we're talking to significantly value the fact that we understand how to operate these clusters at scale, that we have the data center knowledge.

We know how to design data centers to operate these clusters, and we proved out through our own cloud service that we can operate them at a very efficient level. So I don't see any kind of conflict there, and it hasn't been a particular issue for us over time.

Sorry, just to jump in on the Lepton cloud as well. It hasn't really been live functionally. So NVIDIA has been working through a number of items in relation to making that available. I think some of it is now live early access, and we're in direct conversation with them about integration at the moment. So it is a demand partner that we can absolutely envisage using.

Our next question comes from Ben Summers with BTIG.

So kind of more on the colocation side, just curious what went into the decision to start developing Horizon 2. And if that was a lot of potential customers were thinking about potentially scaling beyond the initial 50 megawatts of Horizon 1. And then I think kind of more broader picture, as we progress towards getting Sweetwater online. What's the different customer profile, if any, for more larger scale sites versus potentially just wanting 50 megawatts or 100 megawatts and just kind of any color on the counterparties that you're having conversations with.

So we haven't committed the full CapEx to build in our Horizon 2. So importantly, over the last 7 years, our whole business model has been around cheap optionality. And sitting here right now today looking at the bigger picture, and I can drill into that, it just makes sense to order long lead items and start moving the ball ahead on a potential commissioning of facility. So a lot of the way the S curve works for CapEx in respect to these facilities is you've got long time and small cash outlays that build up over time before the larger CapEx commitments come in.

So it makes sense to put down deposits on long lead items, get the ball rolling, so that we can maintain a really competitive fast time to power for Horizon 2. Now sitting here today relative to 3, 6 months ago, we're seeing further validation of a decision to spec a relatively small amount of capital. We are seeing demand take up for AI cloud. We're seeing the number of inbounds for colocation. We're seeing better visibility on the overall demand supply imbalance for liquid cooled chips. So it's a bit of a no-brainer to be honest. And in terms of committing full CapEx to that, we've got time, and we'll just continue to monitor the market life because things are changing week-to-week in this industry. And that flexibility, having a governance structure of its founder led the ability to make quick decisions, work with the Board and adapt to where the market is going is really important because it is super dynamic.

Thank you. I would now like to turn the call back over to Daniel Roberts for any closing remarks.

Thank you very much. Thanks, everyone, for dialing in. It's obviously been an exciting quarter and exciting year. We're thrilled about expanding to 10,900 GPUs in the coming months and really putting our AI cloud service further on the map. But for us, most of our time is now focused on what lies beyond that. So expanding our 3 gigawatt power portfolio we're working hard on. That's exciting. That's many years away, but it was many years away the 3 gigawatts where we started 7 years ago. So continuing to position ourselves ahead of the curve in every respect is just critical. And it's really important when you're fighting this real-world digital world imbalance where digital demand increases overnight, it goes exponential. Your ability to service that demand with real-world infrastructure and compute works in a linear fashion. It's harder, it takes longer. So the ability to preempt those digital demands and build for tomorrow, position for tomorrow rather than where we are today is a key competitive advantage and something we will maintain. And it manifests itself in us building 200-kilowatt racks, but the industry can't support 200-kilowatt racks.

So Horizon 1, we're having to reconfigure to make it smaller. So we'll continue to keep that in mind. We're excited about the future. We appreciate all of your support and can't wait for the next quarterly earnings. Thanks, everyone.

Thank you. This concludes the conference. Thank you for your participation. You may now disconnect.