IREN, a prominent player in the Bitcoin mining and AI cloud services sectors, reported its fiscal year 2024 results, showcasing significant expansion and a strong financial position.
The company announced a notable increase in its Bitcoin mining capacity, reaching 15 exahash. It has ambitious plans to hit 20 exahash by the end of next month and 30 exahash by the end of the year. IREN's AI cloud service segment continues to demonstrate robust revenue growth, and it plans to launch a GPU pilot at the Childress site.
The company also highlighted its solid cash position and total equity of $1 billion, alongside a powerful balance sheet with total assets of $1.2 billion, which provides ample flexibility for future growth.
Key Takeaways
- IREN's installed capacity for Bitcoin mining has hit 15 exahash, targeting 20 exahash next month and 30 exahash by year-end.
- The AI cloud service segment is showing strong revenue growth and high utilization.
- IREN has secured 2,310 megawatts of grid-connected capacity and over a gigawatt of additional development sites.
- The company purchased an extra 10 exahash of Bitmain miners and has plans for expanding data center capacity.
- A contract for direct spot pricing on power has been negotiated, leading to cost savings and increased flexibility.
- IREN reported an adjusted EBITDA of $54.7 million, with Bitcoin mining revenue at $184 million and an average realized price of $44k per Bitcoin.
- The company's closing cash position stood at $404.6 million, with a total equity of $1 billion and a strong balance sheet with total assets of $1.2 billion.
Company Outlook
- IREN is optimistic about its growth prospects, aiming to double its Bitcoin mining capacity in the next four months.
- The AI cloud service business is expected to grow, but the company will avoid suboptimal deals.
- The demand for power and land is anticipated to grow exponentially, with IREN actively acquiring sites for future expansion.
Bearish Highlights
- The company faces complexity and nuance in the AI side when dealing with partners and counterparties.
- Concerns around the contracting structure, demand, and the impact of new technology on the supply side are being cautiously approached.
Bullish Highlights
- IREN is fully funded to increase its mining capacity to 30 exahash.
- The AI HPC pilot in BC shows a two-year payback on GPUs, indicating strong return dynamics.
- The company has a robust pipeline for AI cloud service growth and is engaged in active conversations for redeploying capacity.
Misses
- The company mentioned manufacturing defects with Bitmain's T21 miners, with around 1.5 exahash of capacity needing replacement.
- Specific details on returns or the size of the AI HPC pilot at Childress were not provided.
Q&A Highlights
- The pilot project in BC is small, with about 16 GPUs, and while the return decision remains unchanged, there is caution over the extended payback period for newer models.
- The company is optimistic about future growth in Bitcoin mining and AI, with a strong financial position to support its ambitions.
IREN's earnings call revealed a company in a strong position to capitalize on the growth of both Bitcoin mining and AI cloud services. With a clear strategy for expansion, cost optimization, and a focus on financial health, IREN is well-positioned to navigate the dynamic and evolving markets it operates in.
InvestingPro Insights
IREN has demonstrated a commitment to growth and financial stability in the Bitcoin mining and AI cloud services sectors, underscored by its recent fiscal year 2024 results. InvestingPro data and tips provide additional context to the company's performance and outlook.
InvestingPro Data reveals that IREN has a market capitalization of approximately $1.47 billion, with a substantial revenue growth of 379.82% in the last quarter. This is in line with the company's reported expansion in Bitcoin mining capacity and AI cloud services. Notably, the gross profit margin is impressively high at 88.16%, suggesting that IREN is effectively managing its cost of sales in relation to its revenue.
Two InvestingPro Tips offer further insight into IREN's financial health and market performance. Firstly, analysts anticipate sales growth in the current year, which aligns with the company's ambitious expansion plans. Secondly, despite recent volatility, with a stock price decrease over the last month, IREN holds more cash than debt on its balance sheet, providing a cushion against market fluctuations and enabling continued investment in growth initiatives.
For those interested in deeper analysis, InvestingPro provides additional tips on IREN, including predictions on profitability and stock price movements. There are 12 more InvestingPro Tips available for IREN, which can be accessed at https://www.investing.com/pro/IREN, offering valuable insights for potential investors and stakeholders considering the company's future prospects.
Full transcript - Iris Energy Ltd (NASDAQ:IREN) Q4 2024:
Operator: Good day and welcome to IREN's Fiscal Year 2024 Results Conference Call. At this time, all participants are in a listen-only mode. After the speakers' presentation, there will be a question-and-answer session. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference over to your speaker Mr. Lincoln Tan, Director of Investor Relations. Please go ahead.
Lincoln Tan: Good afternoon to all of our North American participants and good morning to those joining us from Australia. Welcome to IREN's FY'24 Results Conference Call. I'm Lincoln Tan, Director of Investor Relations and I'm joined today by Daniel Roberts, Co-Founder and Co-CEO; and Belinda Nucifora, CFO. Please note that certain statements that we make during this call may constitute forward-looking statements. These statements are based on certain assumptions and risk factors that could cause actual results to differ materially from our expectations. Listeners should not place undue reliance on forward-looking information or statements. For further details, please refer to the disclaimer on Slide 2 of the accompanying presentation. Thank you for joining us. I'll now hand the call over to Dan Roberts.
Daniel Roberts: Thank you, Lincoln. Welcome everyone and good afternoon. So moving through the presentation, obviously, this is our full-year results. We have a year end of 30 June, so very pleased to report on what's been an exciting 12 months for IREN and its shareholders and, obviously, an exciting outlook over the coming months and years. So jumping straight into it, disclaimer, please all read it, very important. And then into a quick overview of where we are as a business right here, right now. So we're pleased to announce that we have increased our installed capacity Bitcoin mining to 15 exahash. So the last reported update was 10.5. And we reiterate our original unchanged guidance that we will hit 20 exahash by the end of next month and 30 exahash by the end of this year. So in about three and a half months' time, three months' time, time's flying. So that's exciting. We're on track and the team's doing an absolutely tremendous job. On the AI cloud service front, we continue to build good, strong momentum there, month-on-month revenue growth. Our fleet continues to be fully utilized and the pipeline looks promising to further expand that business. And we'll get into that in a little bit more detail. In terms of our data centers that are underpinning this growth. We're up to 300 megawatts now of energized, commissioned and now operating data centers. And that's from a target of 510 megawatts by the end of December. Again unchanged. Reiterating that guidance, we continue on the trajectory of where we said we would be. So all very good, all very positive, and again, huge effort from the team, but everything proceeding according to the original plan. And then, finally, a quick overview of our current power and land portfolio. As most of you would be aware, we have 2,310 megawatts of secure grid-connected capacity and the land to go with that. In addition, we've got over a gigawatt of additional development sites around the world, which we continue to progress with even more vigor over the last few months, given the scarcity around power and land and the demand for that, to underpin not only our own growth ambitions, but also potentially to leverage into, I guess, others growth, which we'll get onto shortly. So, jumping ahead to expand a little bit more on the Bitcoin mining side of the business. So again, a little bit of a recap of where we've been. It was only the beginning of last year, we were operating with 1.7 exahash. Doesn't sound super impressive today, given the numbers that we're talking, but it is good to reflect and for us, for me, for Will to congratulate the internal team on the tremendous amount of growth and execution that they've achieved in that relatively short time period. So to hit 15 exahash today and to continue on a trajectory to double that again over the coming three months is super exciting. We're also pleased to announce that we have purchased an additional 10 exahash of latest generation Bitmain miners, the S21 XPs. Their efficiency is 13.5 joules per terrahash and that will drop our overall fleet efficiency based on the 30 exahash to 15 joules per terrahash. So again, very positive, positions us very well as a low-cost miner, particularly combined with some updates on our power strategy, which I'll come to in Texas in a minute. So, again, just reiterating, we have retained the current 20 exahash of miner options, which are exercisable in 2025, which provides us a pathway to expand to 50 exahash next year. So again, 30 exahash by the end of December and 50 exahash next year. So these miners, they require data centers and real-world construction. And again, the team continues to perform extremely well on that front. You can see the growth since our IPO, in 2021, obviously, it was a challenging period early on IPOing in November 2021 and where the markets went shortly thereafter. But as the markets have recovered, we've been able to capitalize on a lot of the work that we've done in the investment in land planning, grid connections, et cetera to really build some strong cadence and momentum into the end of this year and beyond. The team has done a fantastic job in continually trying to challenge how we procure items, how we build, how do we design, and that's resulting in us achieving significant efficiencies. The first efficiency is we're now building 25 megawatt data centers rather than 20 megawatt data centers and we're now targeting two of those every month. So 50 megawatts of additional data center capacity every month is what we're targeting internally and that's driving significantly rapid expansion, as we're seeing illustrated in these charts. In addition, the other efficiency we're seeing is cost savings, which is always very welcome. So, previously, many of you would recall, we've been guiding towards $750,000 per megawatt. As a result of all these optimizations that we're running, we're happy to report that we're now running at around $650,000 per megawatt. So again, the guys have reviewed all elements of the building design to optimize for both speed and cost without compromising quality. We've worked closely and continuing to deepen our relationships with the supplier network and reduce those lead times, reduced bottlenecks, and again, improving speed and flexibility through construction through internally developed and IP designs around miner racking, filter banks, et cetera. So, again, we're continuing to challenge how we do things. We're not sitting here complacent. We want to continue to get better and we're going to continue to try and do faster and better than what we're doing today. So that underpins a significant growth pathway. But if you haven't got the power and land, then it's very difficult to grow, which is a great segue onto my next slide. We have single site expansion at Childress, which we continue to build into. As you can see on the left-hand side, it's been a fantastic trajectory to-date. On the top left, that is a picture, an image from August last year. A lot of empty space there. And if you scroll down to the bottom left-hand side, you can see considerable amount of construction and data center progress. If we look at the image on the right, it shows you a larger perspective of our overall landholding in Childress available for us to expand into. So, based on the current 750 megawatts available at Childress, you can see the little green area that we're utilizing for our 350 megawatts, which is being built and commissioned over the next four months, into the end of December and then we have a rough guide as to where the next 400 megawatts will be located. At the moment, the initial plan says for that to be continued to be built for Bitcoin mining as the primary use case. But equally there's more opportunities arising on the AI side, potential colocation, potential cloud, which I'll come to shortly. So an update on our pricing and power strategy for Childress. So as many of you are aware, in July, our power costs were high. It was a result of a hedge contract that we had in place. And historically, we have had to hedge, we have had no choice. And a lot of that is down to retailers being very nervous, historically, around allowing Bitcoin miners to take spot power. The risk with that for them and for the miner is that their curtailment systems fail and do not work when power prices are high. So, for example, ERCOT can be very volatile. We can see $5,000 per megawatt hour pricing. If you are operating during that then you're accruing a considerable liability. So the history of miners in this sector has resulted in retailers being very nervous about that. The good thing about July is it brought it to ahead. It allowed us to have those conversations on a very assertive basis and point to the demonstrated track record of our curtailment systems, the scale of who we are now, and our ability to curtail automatically multiple levels of failsafe, and successfully negotiate a contract that allows us to take direct spot pricing from the market from August the 1st. So very exciting for us. It's been something in our pipeline ever since we launched at Childress. But finally, now, off the back of the events in July, off of the back of our demonstrated track record in systems, we've been able to successfully negotiate that. So the benefit of spot pricing is it allows us to optimize our power costs in real time and simply avoid the high-price time periods without incurring the costs and risks associated with entering into near-term hedges. So there's one-off cost of around $7 million to close out the existing hedges. So that is done. That will be booked as a one-off cost in the accounts. And for August month to-date at the time of this presentation, our power price has been 3.1 cents all in at Childress, which equates roughly to about $23,000 per bitcoin in terms of cost per bitcoin mined. So on the right-hand side, you can see our power prices, the average monthly cost since inception. On the left-hand side, that bar chart shows the 4.3 cents, which is our average historically actually. So that encompasses all the hedging costs. On the right-hand side, the 3.5 cents is illustrative of what power price we would have achieved if there was no hedge. So, simply, if all we did was operate on the same profile, with the same curtailment, and took spot market power, 3.5 cents would have been the resultant price. So it's a material cost saving going forward. It gives us material flexibility to adjust to market conditions going forward. At the moment, we're able to set our parameters dynamically around where we want to trade power. So if Bitcoin mining profitability increases, we're able to increase that threshold to divert more of our electrons into the Bitcoin network. If, on the other hand, Bitcoin mining profitability falls, we're able to sell more power back into the ERCOT market, because at that point, it's going to be more profitable to do so. So the profit maximization opportunity is only enhanced through this change in our contracting structure. So, very positive development, something we're excited about, particularly as we now start to hit real scale at Childress and get momentum, building a significant amount of additional capacity there over the coming months. So onto a quick update around our AI Cloud Service. So we continue to service multiple customers with our 816 NVIDIA (NASDAQ:NVDA) H100s. As you can see on the right-hand side, it's been a really good experience year-to-date since we launched in February. Month-on-month revenue growth, month-on-month growth in the number of customers that we are serving, which is equally as important. The opportunity to get exposure to more players in the industry, more companies doing different applications, approaching their systems, their operations in different ways has been fantastic opportunity for us and the team to service them and continue to build our profile through the industry. We're now launching our GPU pilot at Childress in the second half of this year, which is really exciting, the opportunity to see the flexibility across different jurisdictions, different climates. So we look forward to reporting back on the results of how that goes. Poolside, recently extended their contract again, that's now expected to roll off at the end of August. They're looking to consolidate their cluster or clusters into a more larger cluster and the pathway they're going down is for strategic reasons. Confidentiality, we're not able to say too much more, and they haven't told us too much more, but it has been clear that they remain very positive, on IREN, they look forward to continuing the dialogue as their growth continues. And most importantly, they continue to be one of our best salespeople in the market. So the number of testimonials, the number of direct customer referrals that they are providing to us is really helping us continuing to grow that business. So, again, it's a testament to the internal team, the systems, the product that we've launched, the ability to look after every customer that comes through our clusters and drive that market awareness, word of mouth, to ensure that we're recognized as a good operator in this space. Our pipeline remains strong. So at the end of the August, we look forward to redeploying our capacity that will be freed up from Poolside. There's a number of active conversations going on there. We would expect that capacity to be redeployed in relatively swift order, but it may take a little bit of time. Following that, we will continue to assess the opportunity to grow that fleet, but we're very mindful of the capital intensity of growing out our GPU fleet. They are expensive and that's not to say that we won't, but equally, we're now very focused on things like customer credit-worthiness, the amount of prepayments, the tenor of contracts. We've proven the concept. We've grown the business from a standing start six months ago. And now really it's about optimizing the capital allocation to this business and we look forward to growing it sustainably and prudently into the future. Power and land. So a quick update and a recap perhaps to start on what we have when it comes to power and land. We've got 510 megawatts of operating data centers that will be built and online by the end of December this year. That is part of a 2.3 gigawatt overall portfolio of secured grid-connected power. Again grid-connected being very important. You're directly connected into these large-scale transmission lines and get access to wholesale markets. We're not behind the meter. We're not reliant on counterparties. We have our destiny in our own hands. We're connected into public infrastructure and have access to wholesale markets. Very important in terms of risk management. In addition, we continue to develop our 1 gigawatt global development pipeline. We're progressing connection agreements, land options, design, planning and approval for a variety of sites around the world. We look forward to providing further updates on these in due course. But this power and land really underwrites our ability to continue that hyper growth profile that we outlined earlier and the growth that we've experienced over the last 18 months to continue building, to continue to deliver shareholder value. A quick update on other opportunities to monetize this portfolio. So last month, July, we announced that we had appointed Morgan Stanley to evaluate AI data center opportunities for our 1.4 gigawatt West Texas site. That process continues to play out. We're signing NDAs. We're providing information and data room access to interested parties, so that's exciting. It's a process that will take some time. But equally, the traction that we're getting early on, the early conversations, the engagement looks promising, that there will be robust conversations around opportunities for that site. Where it ends up, I don't know, but it's exciting to go through the process. And the fact that we're engaged with some of the largest companies in the world, active in this space is optimistic for the asset base that we've built and the opportunity set that lies ahead. In addition, these conversations, in conjunction with existing relationships that we have are also opening up additional conversations around our existing sites in Canada in Childress, where a range of structures have been discussed with potential partners around AI cloud services, around colocation and a variety of other options. And again, to recap, the attractiveness of these existing sites is they have grid-connected power right now. We have the ability to retrofit additional generation and batteries. We have the ability to retrofit liquid cooling if that's required and suitable for customer clusters. And finally, the latency. We've got sub 10 milliseconds from our Texas sites into major hubs. There are 1,000 milliseconds in a second. We've got 10 milliseconds. So latency is great. It should support a vast variety of applications and really not be a constraining factor. In Canada, 20 milliseconds, still more than sufficient for the majority of the applications that we're seeing. So we look forward to providing further updates on all this in due course and to some Q&A at the end. But in the meantime, I'll pass over to Belinda to give us an update on the numbers. Thanks, Belinda.
Belinda Nucifora: Thank you, Dan and good morning to those in Sydney and good afternoon to those in North America. Thank you for joining us for our full-year earnings update and I'm pleased to report for the year ended 30 June, 2024, the adjusted EBITDA is $54.7 million, being a significant year-on-year increase of $53.3 million and the highest recorded EBITDA for the company. With Bitcoin mining revenue increasing from $75 million to $184 million as the average operating hash rate increase from 5.6 exahash to 9.4 exahash, resulting in 4,191 Bitcoin mined and an average realized price of 44k being an 89% increase in price year-on-year. During the financial year, as Dan mentioned, we commenced our AI Cloud Services business and we recorded revenue in relation to that business of $3.1 million. Average net electricity costs per Bitcoin mined increased from 11k to 18.1k, primarily due to the increase in global hash rate and the impact of the halving event in April 2024. Our other costs increased from $38 million to $56 million. These include employee benefit expenses of $22 million, site expenses of $8.7 million and these include the procurement of RECs which is consistent with our commitment to utilising 100% renewable energy. We had insurance costs of $7 million, professional fees of $6 million and we also had our provision for Canadian non-refundable sales tax of 6.3 which is an ongoing CRA audit. The increase in the OpEx reflects a larger business today than FY'23, including the expansion of our Childress site and the commencement of the AI Cloud Service businesses as we recruit resources to support that growth which has delivered significant growth during the financial year and we project that growth to continue over coming years. We also have increased costs in relation to expanded risk compliance and reporting obligations which come with being a listed NASDAQ company. Depreciation increased from $30.9 million to $51 million due to commissioning of assets at Childress as well as accelerated depreciation for the S19 J Pro miners scheduled to be sold in quarter one of FY'25 due to our recent fleet upgrade. So I might now turn to our cash flows for the year. We had very strong cash flows with a net increase of $336 million which resulted in a closing cash position of $404.6 million. That increase in net cash came from primarily operating activities resulting in net cash of $53 million and included $184 million of receipts from our daily liquidation of our Bitcoin mined. These positive operating cash flows highlight the quality of our underlining operations and are reinvested to support our ongoing expansion plans. Increase in net cash used in investing activities was $499 million and is due to the expansion of the Childress data center as well as our purchase of the Bitcoin, sorry, the Bitmain S21 Pro and T21 miners as part of our pathway to 30 exahash in 2024 and 40 exahash beyond as well as our purchase of the NVIDIA H100 GPUs of 816. Our increase in cash from financing activities was $782 million with the sale of 108.1 million shares sold under the ATM and 12.9 million of shares sold under the ELOC facility. So might move on to now our balance sheet and as mentioned in our cash flows, we have a very strong closing cash at 30 June of $404.6 million. We continue to have no external debt, so no debt facilities, and we have strong operating cash flows as discussed on the previous slide. Our total equity increased to $1 billion with gross proceeds of $823 million from the sale of 121 million shares sold under the ATM and the ELOC. This excludes 463,000 shares settled in early July, which raised a further $5.2 million of cash proceeds. Effective today, we have a remaining 223 million on our ATM facility. We have a strong balance sheet with total assets of $1.2 billion, which provides flexibility to fund our future growth. So that's a highlight of the key earnings highlights for the FY'24 year against FY'23. I think now we're going to turn to Q&A.
Operator: Thank you. [Operator Instructions] One moment while we compile the Q&A roster. Our first question comes from the line of Joseph Vafi with Canaccord Genuity. Your line is open.
Joseph Vafi: Hey, everyone. Good morning. Thanks for all the updates here, very thorough. Just maybe we'll start off, Dan, you did a New York analyst update for us all a few weeks ago. Anything kind of material since then or incremental to add in, I guess, in both strategies. And what's your latest thoughts here on attractiveness in Bitcoin mining versus perhaps putting more CapEx into the hosting side shorter term? And I have a follow-up.
Daniel Roberts: Yeah, hi, Joe. A few short weeks ago in New York, it was a really good event and appreciate you and 100 others in the room taking the time. Look, things continue to evolve like every day, every week. Will and I are learning more of the team and continue to iterate our thinking. It doesn't result in materially different decisions, but at the margins it does start to refine your thinking. What I will say about Bitcoin mining is we are in control. We're the masters of our own destiny. With 2.3 gigawatts of power, we can plan to build a lot of Bitcoin mining capacity over the coming few years. So that's a fantastic position to be in, particularly when we can do that accretively, particularly when capital markets are rewarding for that growth. So that's really exciting because it's something that we can control and plan for. On the AI side, so the non-Bitcoin applications, frankly, the challenge there is you're dealing with partners, you're dealing with counterparties, and that takes more time, there's more complexity, there's more nuance. The Cloud Service is essentially a commodity in the sense that you're selling GPU hours online. So we provide a VPN, people log in, they use the GPUs for their own purpose. And the contracting structure is relatively simple. When it comes to larger-scale deals, whether it's on the AI cloud or colocation or perhaps leasing a site perhaps selling a site if we got the right price, obviously, you're dealing with a different scale and a different set of complexities involving counterparties. So when we talk to partners or prospective partners, I should say, around potentially building them out, data center facilities for them to host their own GPUs in, there's a whole variety of considerations around what those data centers need to look like. The technical specs, things around backup generation, things around chillers, things around how the rack density looks and the internal network in the software. So there's a lot to it, and it means that in a few weeks, there's not a material update on that, because these things take time. But the best thing for us that gives us so much excitement about this business is we are the masters of our own destiny, where Bitcoin can continue to underwrite significant growth in this business. And this was the plan six years ago when we set the business up. Bitcoin bootstraps the business. We build multifunctional data centers and then we can leverage into other applications. Once we have the opportunity to sign the right type of contract to grow the business in the right way on non-Bitcoin application, absolutely we're going to do it. But equally, it's not like we sit here feeling pressure to do a deal that's suboptimal, we're able to really control and throttle both sides of the business in a way that makes sense.
Joseph Vafi: Sure. Thanks for that color, Dan. And then you've got a pretty big portfolio now, but you're still kind of in site acquisition mode with a big pipeline. Just kind of maybe kind of, again, talk about your thought process of further site acquisition, given how large your portfolio is now and CapEx for more site acquisition versus CapEx for site build-out? Thanks a lot guys.
Daniel Roberts: Yeah, absolutely. I mean, things change so quickly, and this is part of the original thesis, the scarcity that we see come in over the next 5, 10, 20 years for power and land as the world continues to be digitized. And, look, it was probably only six months that we had analysts very skeptical of our ability to ever build out 600 megawatts at Childress. But the world changes really quickly, and it comes back to the original thesis six years ago when Will and I set up this business which is the dislocation between the digital world and the real world. The real world cannot and will not keep up because the digital world moves in these exponential functions. You've got adoption curves that go vertical, this insatiable demand, the things that pop up in relative terms overnight, Bitcoin, AI, we don't know what's next. But the point is those demand drivers are exponential, whereas the ability to service that in the real world, it's hard and it takes a long time and it takes longer and harder, the bigger that digital world gets because the low-hanging fruit gets taken early, and then it gets harder and harder and the order of magnitude grows to get more power, more land, navigate supply chains, build stuff in the real world. And that's the vision for the business originally and that's the thesis that's playing out. So we're not going to stop. Absolutely not. I want gigawatts more of power because I fundamentally believe in 10 years, the scarcity dynamic that we're seeing in the market now is nothing compared to what's probably going to come. And we talk about CapEx. It doesn't really involve CapEx like you sign in land options for $20,000 here, $30,000 there. Yes, we've got a big team internally that costs us a little bit of overheads to go and do that, but most of it is time, relatively small amounts of money, and building up this incredible optionality on a future that we still fundamentally believe in.
Joseph Vafi: Great. A lot to look forward to. Thanks a lot for those comments, Dan.
Daniel Roberts: Thanks, Joe.
Operator: Thank you. One moment for our next question. And that will come from the line of Lucas Pipes with B. Riley Securities. Your line is open.
Nick Giles: Thank you very much, operator. This is Nick Giles on for Lucas. You've made some key hires related to HPC here recently and so I was wondering if you could expand on what these teams are focused on today. Should we see this as a read through for more substantial GPU purchases in the near-term? Thank you very much.
Daniel Roberts: Potentially, if we get the right structure, I mean, my appetite to go and dilute shareholders and take on debt, to spend hundreds of millions of dollars on spec to build out an AI Cloud Service business is frankly quite low. The market should speak to that and we should be able to negotiate contracts that in our minds have the right risk return. If we never do another contract because the risk-return doesn't make sense then I'm cool with that as well. We're not going to pursue it just for the sake of pursuing it. But as we prove out our credentials, we prove out our operating performance data and we provide those metrics to customers and we have an increasing amount of current and past customers providing those referrals, those testimonials, I'm optimistic that we're in a really good position to grow that business. So it's something we're really optimistic on, but equally, I'm not going to sit here and just grow for the sake of growing. The investment in people and platform within our business is really important. And we've always said we need to invest ahead of the curve. You don't grow a business by focusing on every dollar of your expense line when you're growing in a hyper growth manner. I mean, we're 10x in our Bitcoin mining capacity from 18 months, 20 months ago, right. We've grown an AI Cloud Service business that was zero six months ago that's now doing annualized revenue of $15 million, servicing half a dozen customers, and the ability to continue investing in people systems processes. Given the optionality in the business that we've created with all that power and land, the data center design and the existing team, it's a really valuable investment for us. And it covers a whole span of different capabilities from network engineering to AI sales, to general commercial, investment banking funds, management backgrounds helping us negotiate prospective structures and large-scale transactions all the way through to people that have had development experience with batteries, wind, solar. As I said in response to Joe's question, we're not stopping. We're going to continue pursuing this hyper growth because we fundamentally believe that the growth and demand for renewable energy power compute is not only not going away, it really just started the exponential growth curve.
Nick Giles: Dan, I really appreciate all that color. My follow-up would be, do you have any target levels of prepayments in mind, target contract duration and then maybe any additional color you could add around the GPU pilot at Childress, the capital spend there, how many incremental GPUs you would add? Thank you very much.
Daniel Roberts: So the hard thing about giving guidance on contract structure like it's a great question and it's really relevant. The hard thing is every situation is slightly different. So as a base case, we'd love a reasonable proportion of prepayment, we'd love contract terms, one, two, three years. And that's got to be the base case target. But equally, every circumstance is different. Every customer is different. There are strategic reasons why you might do something different. For example, it might be a trial, it might be a small opportunity to give them a small cluster, seeing that they've got large growth ambitions, and there could be a larger cluster coming thereafter once you've proved yourself. So it's really hard to give specific guidance. But I think we're talking 10 or 12 months plus, ideally, and perhaps a bit longer, and yeah, several months of prepayments, if not a little bit more. But again like please don't hold me to that, because we want to remain flexible and make the right decisions and circumstances are always unique when it comes to these decisions.
Nick Giles: Dan, that certainly won't hold you to it, but congratulations to you and the team so far and continued best of luck.
Daniel Roberts: Thank you.
Operator: Thank you. One moment for our next question. And that will come from the line of Paul Golding with Macquarie. Your line is open.
Paul Golding: Thanks so much. Just a couple from me. First on the Poolside capacity that sounds like you're going back to market with to find a client for it. Could you speak a bit to how you're deciding between going back to a single customer to take a significant swath of this capacity versus putting it on the market for on-demand customers to use? How do you think about that equation, now that this will become available at the end of the month?
Daniel Roberts: Yeah, it's a great question, and we think about this a lot, Paul, and the short answer is we want to do all of it and retain flexibility. And it's kind of funny like Poolside extended the contract again that was due, so they've doubled down, and extended a couple of times now, but the last one to roll off was earlier this month and they extended again to the end of August. And then they said, we're consolidating elsewhere, and gave us a little bit of color. But essentially it makes sense from their side. And initially, our reaction was, oh, that's pretty disappointing. We enjoyed working with Poolside. They're a great name, they've been a great thing for launching this business. But the funny thing is now having that available capacity immediately is opening up all the opportunity to pursue other avenues and provide people the opportunity to use a small amount of these 500 GPUs that become available as a segue and hopeful pathway to deploying larger scale capacity with it. So it's ended up being a bit of positivity coming out of it. And then in terms of long-term contracts versus spot and on-demand, we're pursuing both. So we're developing additional on-demand software capabilities. Internally, we're looking to provide some of that into the market at the moment, but equally in parallel, we're pursuing longer-term contracts. So it's just getting that mix right and retaining the flexibility. Do I envisage having thousands of GPUs operating on an on-demand basis in the near-term where we've got customers cycling through on a lifetime basis and not a huge contract bank for it? No. But do we want to continue building the flip capability, continuing servicing that market, learning about the depth of that market and what it might look like in the future? Absolutely.
Paul Golding: Thanks, Dan. And then just one more on the data centers themselves, the structures. It sounds like you've iterated on the design. In Texas, you're now down to $650,000 per megawatt cost, and you've got an extra 5 megawatts for structure. Could you give some extra color around what is enabling that extra 5 megawatts? Is that something where you can go retroactively to the six or so buildings that you've already constructed and flow that through there as well on a like-for-like basis or is this more of just a go-forward? Thank you.
Daniel Roberts: Yeah, sure. No, it's just a go-forward. I mean all these existing -- the existing 20 megawatt buildings were designed for 20 megawatts. So the form factor, the size, the low-voltage electricals, the network cables, et cetera, so they're optimized for 20 megawatts. So simply, the 25 megawatt design is just a bigger version of the 20 megawatts with some increases in density, efficiency, et cetera and the way that we do it. But really it's a go-forward thing that's allowed us to save a material amount of CapEx on the data centers and also build out megawatts on a faster basis.
Paul Golding: Great. Is even larger than 25 megawatts on the table or is 25 what we should expect for the foreseeable future?
Daniel Roberts: Everything's always on the table, Paul. So we're looking at that and a lot of other options, but for now, we're pretty happy with 25 megawatts. We'll keep punching them out and grow into it. But there's absolutely, there's lots of innovation challenge ideas being worked on in the background. So I would expect us not to sit still and just be sitting on that one design for the long-term.
Paul Golding: Great. Thanks so much and congratulations.
Daniel Roberts: Thanks, Paul.
Operator: Thank you. One moment for our next question. And that will come from the line of Mike Colonnese with H.C. Wainwright. Your line is open.
Mike Colonnese: Hi. Good morning, Dan and team, and congrats on all the progress at Childress. First one, for me, Dan, you guys always had a really great perspective on some of these market dynamics, so it would be great to get your outlook for hash prices here, which, as we know, have been really trending around all-time lows. Do you think we've bottomed out here? And when do you think we can reasonably see a breakout above this 4.5 cents per terrahash range?
Daniel Roberts: Thanks, Mike. Easy question. I mean, it's a big world out there and there's lots of different miners and all have got their own kind of cost structures, availability of capital, power, land, and it just gets fed into that equation. So the key driver is the price of Bitcoin, right? Bitcoin rallies, hash price skyrockets, and everyone's making money, and everyone's scrambling to invest and bring online more capacity. So that is the absolute key driver. What the price of Bitcoin does day-to-day? I've got no idea. What it does long-term? Feels like it's inevitable. It's going significantly higher. So the key for our business is positioning that we can survive extended periods where the Bitcoin price isn't that exciting and be really well positioned when it does run. So when you look at our cost per Bitcoin mine at Childress this month, in the low 20s Bitcoin mined, the increases in efficiency come in. We're on good gross profit margins. We're feeling good at the current Bitcoin price levels. Now, it's a good business. That's why we're continuing to put more capital in. It's accretive, it's generating significant cash flow, but not everyone can do it. And that's part of the hurdle and I guess the moat that we've got and the opportunity and challenge for others in this space. So, ultimately, if that hash price wants to grow, then you need the price of Bitcoin to move, and then people will try and build into it, but the challenge is how you do that. So we're probably at, what, 20 gigawatts to 25 gigawatts of power globally today dedicated to Bitcoin mining capacity, but that's no small feat. I think the global data center industry is maybe 25 gigawatts to 30 gigawatts, I haven't seen the latest numbers, but that is a lot of power. And the whole thing comes back to, I keep repeating around the digital world, real-world dislocation, you can't click your fingers and respond to these exponential functions that are moving in the digital world. And if you look at the price of Bitcoin today, Bitcoin happens to double over the coming period then how do you possibly find 20 gigawatts to 25 gigawatts of power in any timeframe that's in the short, medium term, to try and normalize mining profits or the hash price to where it is today? So that's the whole opportunity in saying that expect the hash rate, the global network to grow, but it's getting harder to grow because of that access to power and land. And really, it's probably more going to be growing through a function of increased efficiency on the chip side, which, again, is going to be slow, again is going to require capital. Not all miners have access to that capital.
Mike Colonnese: Got it. I appreciate the color there. And you cited in your July production update that there were some T21 minor performance issues, which we've seen with a couple other miners as well. I'm just curious if you can specify what those issues were for you and how much exahash is being replaced and when Bitmain should get those replacements to you?
Daniel Roberts: Yeah, I'm not sure how granular we've been on the details. And Bitmain are a really good partner of ours. So I'm reluctant to speak too negatively about them because they've been fantastic for us over the long-term. But there were some manufacturing defects around heat sinks and ability to operate in the higher temperatures. And just general construction and assembly quality relative to their high standards that we see across other units. But again, testament to Bitmain, their response times and them accepting this issue and taking 100% responsibility for it has been fantastic. So again, everyone makes mistakes. I make mistakes all the time. There's always issues. It's how you respond, it's how you fix things. And Bitmain have been absolutely fantastic. It's around about potentially 1.5 exahash of capacity and that is being replaced lifetime. So you would expect to see that resolved over the coming months.
Mike Colonnese: Great. Thanks for taking my questions, Dan.
Operator: One moment for our next question. And that will come from the line of Joe Flynn with Compass Point Research. Your line is open.
Joe Flynn: Hi. Thanks for the question. The question regarding the 1.4 gigawatt West Texas site. And you mentioned in the past you've been chipping away at the CapEx there, but can you provide some color on what actually is on site, whether there be transformers, substation, long haul fiber, et cetera, and also if there's any remaining ERCOT approvals? Thanks.
Daniel Roberts: So there's nothing on site at the moment. It's a greenfield site. We're in the design planning phase around mapping out how we will build there. But part of us coming to market on that site with Morgan Stanley is we're at that decision point now where energization isn't far away and the utility is working on that energization, it's into an existing network substation. So in relative terms, it's not an overly complicated connection. So they're working through their construction and procurement processes. And then for us, it's really design at this stage and coming to market now is because we need to crack on with it. We need to make decisions because the real world takes time and there's long lead items. So we're thinking already and working through the process for procuring high-voltage transformers and key lead items. But there's nothing material to report at this particular juncture.
Joe Flynn: And if you would just comment on that, like, in your conversations with Morgan Stanley and potential, customers or deal structures. I mean, have you seen kind of valuations from just dollar per megawatt basis start to improve in areas like West Texas and just anything you could comment on there would be great.
Daniel Roberts: Yeah, I've got nothing to say. We're just not at that point with the Morgan Stanley process. We've had kind of indicative conversations around what colocation and AI Cloud looks like at a smaller scale, the 1.4 gigawatts, but the reality is it's such a dynamic market. You look at how it's kind of evolved over the last six to nine months. Every month that goes by, the whole sector is thinking about, well, how real is this power and land scarcity dynamic. What are these things worth? What's the price that I'm willing to pay? So again, we've seen market benchmarks and there's a couple of them out there which we can point to and others can point to, which are informative in these negotiations and discussions. But again, it's so dynamic that really from our perspective, it's just looking at every opportunity on its face and saying, well, does the risk-return make sense if we're allocating capital. If we're not allocating capital, what's the opportunity cost of giving up that power and land. Recognizing we've got a lot of it, we've got a lot more coming with our development pipeline. So again, it's, yeah, I wish I could give more color, but it's nuanced, it's dynamic. And until we get to that actual decision point, it's really hard to give firm guidance because whatever I say will be wrong. Like that's the reality. Everything's just so dynamic.
Joe Flynn: Understood. Thanks for the questions.
Operator: Thank you. One moment for our next question. And that will come from the line of Brett Knoblauch with Cantor Fitzgerald. Your line is open.
Brett Knoblauch: Hey, guys, thanks for taking my question and congrats on the quarter and thanks for all the updates. Maybe starting with the new 10 exahash of Bitmain mining machines, should we have those machines being plugged in this year or those machines will be plugged in next year?
Daniel Roberts: Hey, Brett. This year. So a key part of that negotiation was trying to ensure that they'll be delivered this year. Some of it might be a little bit tight going into December, but that's a good challenge for us. But the whole idea of that additional purchase was to bring it forward into this year. Given that we're making the investment decision to go to 30 exahash, we're fully funded to do so, and then the ability to roll the existing optionality, the existing 20 exahash of options into next year. And I guess kick the can on that decision point was a key part of that decision for us.
Brett Knoblauch: Awesome. Appreciate that. And then on the AI HPC pilot at Childress that you're expecting to finish in the second half of this year, could you maybe just first broadly talk about what type of kind of returns you're seeing from what you're doing right now in BC, how much did you spend on the GPUs? And I guess just talk through the return dynamics of that business? And maybe just another follow-up on the size of this pilot, should we expect something similar in size to what you guys are doing in BC?
Daniel Roberts: Yeah, and apologies to Nick. He actually asked me about this earlier from B. Riley and I forgot to answer it. So the size of the pilot in BC is small. Like it's about 16 GPUs, I think. So just under a million dollars' worth. And, look, we're looking to start training and utilizing it imminently, so that's happening. So that's exciting. We'll learn a lot very quickly around that. But in terms of the return decision, again, it's pretty much unchanged from prior guidance in the sense that you're looking at roughly two-year paybacks on these GPUs. Maybe it's extended a little bit to 24 to 30 months for the H100s, as the Blackwells, the H200 start to come to market. So you've seen that impact of better technology coming through. But again, you've got to weigh up that breakeven point in terms of the CapEx against how much capital are we putting to work, i.e., there are other prepayments at work, what's the contract tenor, what's the level of risk behind that contract, et cetera. So like prima facie, it looks like a great business, the near-term paybacks, but you've got to get the contracting structure right, you've got to be mindful of the demand. And it's an uncertain world so far as AI training and the demand for these GPUs over the coming years. It all sounds and looks really exciting and we think we're on that exponential curve and we probably are inevitably at some point. But there's always bumps in the roads, so we don't want to be caught out there. And then on the supply side, the impact of new technology coming to market, the iteration and the efficiency of these GPUs. Again, you just don't want to get caught holding a whole bunch of older GPUs where the revenue profile of those GPUs falls a little bit faster than what you're expecting.
Brett Knoblauch: Perfect. Thanks, guys. Really appreciate it.
Operator: Thank you. I'm showing no further questions in the queue at this time. I would now like to turn the call back over to Co-Founder and Co-CEO, Mr. Dan Roberts.
Daniel Roberts: Thank you. I feel like I've done a lot of talking and sorry Lincoln, Belinda, but thanks everyone for dialing in. It's been a great 12 months behind us, but probably a more exciting three months to six months plus ahead of us. We feel like we're in a really good position, billion plus of equity on the balance sheet, $400 million of cash, zero debt. So a really robust base platform. And we continue to prove our ability to grow significantly. I mean, we've grown to our Bitcoin mining capacity 8x since the start of last year. We've built it 2.5x since the start of this year and we're going another 2x again in the next four months, all fully funded. So to go from 1.7 exahash started last year to 30 exahash at the end of this year, it's just great. Like, we're super excited about it. But for us, 2025 is not far away and the world doesn't end on 31 December. So the opportunity to grow substantially into next year and beyond both on Bitcoin mining and the AI side of the business is something we're really excited about. So appreciate everyone's support. Thanks for dialing in and have a good evening.
Operator: This concludes today's program. Thank you all for participating. You may now disconnect.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.