IREN Limited (IREN) Q3 2024 Earnings Report, Transcript and Summary

Good day and thank you for standing by. Welcome to IREN's Third Quarter FY 2024 Investor Update. At this time, all participants are in a listen-only mode. Please be advised that today's conference is being recorded. After the speakers’ presentation, there will be a question-and-answer session. [Operator Instructions] I would now like to hand the conference over to your speaker today, Lincoln Tan, Director of Investor Relations.

Thank you. Good afternoon all to those of you in North America and good morning to those of you in Australia and welcome to IREN Third Quarter FY’24 Results Presentation. My name is Lincoln Tan, Director of Investor Relations and joining me on the call today are Daniel Roberts, Co-Founder and Co-CEO and Belinda Nucifora, CFO. Before we begin, please note that this call is being webcast live with an accompanying presentation. For those that have dialed-in via phone, you can elect to ask a question via the moderator after our presentation. I would like to remind you that certain statements that we make during this 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 different 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 within the accompanying presentation. Thank you, and I will now turn the call over to Dan Roberts. Over to you, Dan.

[technical difficulty] Please stand by, your conference will begin momentarily.

The connection will be reconnecting very shortly.

Dan, you may begin whenever you're ready.

Excellent. I think we're back on. Apologies for that technical issues but we're ready to go. So thanks everyone for dialing in, thanks Lincoln for the introduction. We're thrilled to be here for our quarterly earnings call. There's a little bit to get through, so let's jump straight into it. So first of all, it's probably worth recapping where we're at in terms of our business and our priorities for 2024 and beyond. And I think working from left to right on the slide deck in front of you, the big news is we're now going to 30 exahash this calendar year. So we're now at 10 exahash of current Bitcoin mining operating capacity. We told guidance and provided guidance around 30 June for that milestone. We've hit it in mid-May and we're now upgrading our end of year. So over the next seven months and a bit we're going to 30 exahash. That's as a result of bringing forward the construction of some infrastructure, so some additional data centers that we will build by the end of this year, along with a transaction that we're pleased to announce today recently completed with Bitmain. So a bit more on that to come but that's obviously super exciting and will take us up into that very top-tier of listed Bitcoin miners. The next part which is also exciting is the continual development of our AI cloud service business. We've provided updates over the course of this year. We are looking at ways to expand this, so we look forward to giving you a further update during the course of this presentation. And then there's been a lot of commentary publicly in the media, social media, about the value of power and land in the context of the data center and AI crunch. As most of you know we've got over 3,000 megawatts of power and land secured. This not only provides us with fantastic optionality and flexibility to continue our organic growth profile but it's also giving rise to some interesting nearer term opportunities to potentially bring forward monetization of some of that portfolio. Again we'll get into that in a little bit more detail but it's an exciting time, there's a lot going on and without further ado let's jump in to the Bitcoin mining side of things. We're now in a position to go to 30 exahash in 2024 and in fact 20 exahash by the end of Q3. So over the next few months we anticipate quite a large swift ramp up in our capacity. We were the fastest growing Bitcoin miner last year in the listed-market and as a result of this we anticipate being similar this year. So 30 exahash of capacity, the majority of that is via S21 Pro from Bitmain which have an average efficiency of 15 joules per terahash which by the end of it all being installed from the 30 exahash this year we anticipate 16 joules per terahash in terms of nameplate efficiency and a $17,000 per Bitcoin electricity cost. So in terms of size, scale, low cost, we're very pleased to be in a position where we think that we're on course to be one of the real industry leaders in this. So again, this is as a result of bringing forward the construction of the infrastructure, the operations team, the construction team are really accelerating their work around this. As we've spoken about in the past. We've got a standardized data center design that we've been rolling out for a number of years now. We recently upgraded that from the 20 megawatt design to 25 megawatt design. That's indeed helped but the fact that it is standardized, the fact that we have been doing this for a number of years, the team is simply getting better and faster at it. So that's putting us in a great position to continue to build. What I also should say is that we're not stopping at 30 exahash. We've got options for another 10 exahash, as we'll see on the next slide. So as you can see we're pleased to announce that we're at 10 exahash in terms of current operating capacity. We anticipate hitting 20 exahash over the course of the next few months, i.e. by the end of September, and then by the end of December hitting that 30 exahash mark that I've mentioned a few times now. In addition, we've got fixed price options for an additional 10 exahash to potentially deploy early in 2025. And I think, what's really important about all of this, is the fact that we've got the land, we've got the power. So six years ago, when Will, my brother and I set this business up, our belief was firmly that power and land would be [scarce] (ph) and the fact that it takes multi-year timeframes to develop and many of you on this call invested in our early rounds where we spoke about things like The Matrix, Ready Player One, Wreck-It Ralph, all these sci-fi movies where it portrayed humanity in a certain direction and as a result of that how much computing power it would need. Now who knows where we end up? Is it that way? Is it a different trajectory? But our belief was over the next 10, 20, 30 years, the world is going to have this insatiable amount of demand for computing power and we're going to have this real world, digital world dislocation that we've spoken about many times before. The fact that the revenue line, the demand drivers are largely driven by the digital world, whether it's the price of Bitcoin, whether it's demand for AI, whether it's some sort of new computing application in the future, the reality is once the real world gets to a critical mass it's very hard to scale up. And this is why at the moment we're seeing such demand for new power, both from Bitcoin miners, data centers, hyperscalers, et cetera. And the fact is you just cannot bring forward these power projects for various reasons we'll go into. So we're in a really good position, where it's all organic growth. We've spoken to you for years about the platform and the foundation that we are laying down in terms of that organic growth and over the next six months to nine months we believe that you'll really see the fruits of that investment as we scale up our business extremely quickly. So I might pass over to Lincoln.

Thanks Dan. So just to provide context, the next couple of slides will just talk through the CapEx requirements, funding requirements as well as some further detail around the S21 Pro agreements that we've just signed. So turning to this page, in terms of the CapEx as we previously mentioned, the expansion to 20 exahash which is now being pulled forward to end of September this year, that is already fully funded. And the focus on this page is really talking about the additional CapEx and growth to get to 30 exahash by the end of this year. As we saw on the previous slide, the step up from 20 exahash to 30 exahash requires an additional 10 exahash for miners and an additional 150 megawatts of data center capacity. So really, at Childress, by the end of this year, we will have 350 megawatts of operating data center capacity at Childress. And the CapEx guidance that we're providing for both the mining hardware as well as the data centers is approximately $300 million. And as you can see on the top right quadrant of that slide, that comprises approximately $190 million for 10 exahash of the latest generation Bitmain S21 Pro machines, and approximately $110 million for 150 megawatts of data centers, which is broadly in-line with the previous guidance we provided. That sort of benchmarked data center CapEx at about $750,000 per megawatt. Now turning to the funding strategy. We anticipate funding the step up to 30 exahash through a combination of existing cash resources, operating cash flows and other sources of capital. As we've previously mentioned, we have $322 million of cash, and that is of 30 April, 2024. The business is generating very strong operating cash flows, which we are reinvesting into growing the business. And Belinda is going to touch on this in more detail in the financial statements. But you will note that we reported $48 million of positive operating cash flow year-to-date, this financial year. And we are also continuing to explore other funding opportunities right across the capital structure. So as you're aware we've got our existing ATM which has a capacity of $137 million remaining. We've also just filed a new $500 million shelf in ATM. And in parallel we're also exploring opportunities with debt capital including potential funding for data centers and equipment financing for GPUs which Dan is going to touch on a little bit later in this presentation. And then finally, in terms of returns, and that's just talking to the shaded box on the left of this page, we just wanted to take the opportunity to provide a little bit more detail around how we frame up our capital raising and investment decisions. So fundamentally, the use of the ATM program, for example is entirely at our discretion and we just view this as an important instrument that helps us fund growth that makes sense strategically and also financially in terms of returns to shareholders. We've run some analysis, some benchmarking analysis on how the market is valuing Bitcoin miners today and we note a high degree of value being placed on installed exahash capacity. And in particular for the larger scale miners specifically Marathon, CleanSpark and Riot, each of these operators with hash rates above 10 exahash. The market is valuing their capacity at an average of around $135 million for installed exahash. And as you can see on the right hand side of the page, we are building and delivering incremental hash rate at $30 million CapEx for 1 exahash and that is inclusive of the mining hardware and the data centers. So putting all of that together we see raising capital to become one of the largest and most efficient publicly listed miners this year as making strategic sense and also driving long-term value creation for our shareholders. Stepping to the next slide, we won't spend too much time on this. There's a lot of detail in our disclosures, but we've entered into a number of purchase and option agreements with Bitmain, which underpin our growth to 30 exahash this year and providing a pathway to 40 exahash in the first half of calendar 2025. Just wanted to emphasize that these new and amended agreements relate to a combined 35 exahash of the S21 Pro Miners, which as Dan mentioned, have an efficiency of 15 joules per terahash. These agreements have been struck at [$18.90] (ph) per terahash. And as you can see on the table, as we start to deploy that hardware, you can see that meaningful improvement in overall fleet efficiency to 16 joules per terahash at 30 exahash and then to 15 joules per terahash at 40 exahash. And in terms of just the structuring, these agreements were very deliberately structured in this manner with delivery dates that align to the data center rollout at Childress over the rest of 2024 and 2025. Having options in the structure provide a degree of flexibility around growth as compared to, for example, a firm purchase, as well as an element of downside protection to the extent market conditions change. And as we've just demonstrated, as new technology emerges and new and more efficient mining hardware emerges, we were able to, for example, amend the terms of our T21 option into the latest generation S21 Pros without impacting the deployment timeline and just flagging that this may not have been possible if we'd made a firm purchase order for those same machines. Turning to the next slide, this just maps out the landscape and a couple of perspectives to share here. The chart shows Bitcoin mined in April and the table underneath just shows market cap as of last Friday and installed hashrate across the sector. So as Dan mentioned we were the fastest growing miner in 2023, increasing our hashrate last year from 1.5 exahash to 5.6 exahash at the end of 2023, whilst maintaining market leading levels of efficiency. We are continuing that trajectory this year, having grown from 5.6 exahash at the start of the year to 10 exahash today. And obviously, there's a lot more growth to come in 2024, as we grow to 30 exahash this year. And moving into 2025, fully expect to maintain that momentum as well with our pathway to 40 exahash in the first half. I'm going to hand this back to Dan now to talk to the AI Cloud Service business.

[technical difficulty] Please stand by your conference will begin momentarily.

So, with the technical difficulties, I will step in and talk to the next couple of slides while Dan, hey Dan I think we've got you back there.

Excellent I'm not sure what's going on technical issues we'll have a chat afterwards. Alright, so into the AI update. Thanks, Lincoln for [rounding out] (ph) the Bitcoin mining. Obviously, very exciting. So most of you aware that when we set out building this business, we built data centers. So we never built shipping containers, sea-cans, we never subscribed to the abandoned warehouse Model of mining Bitcoin. We built multifunction data centers from day one, capable of supporting different types of computing power. We signed an MOU with Dell Computing many, many years ago, about four years ago, four and a bit, to bring some of their customers and hardware out to our first site at the time. The market wasn't quite ready. We put it on -- focused on Bitcoin mining. And then with the roundup in AI last year, we dusted off the old strategy, went and ordered some NVIDIA H100s, have installed them in the same data centers, i.e. we unplugged the ASICs and we plugged in the GPUs, the same infrastructure, the same data centers, and we launched an AI cloud service. It's a highly complementary business line. We don't see it as competing with Bitcoin mining, we see it as additional and incremental to Bitcoin mining. And the key reason for that is the power consumption. As you can see in the chart on that right hand side, every $100 million of GPU expenditure is only using around 1% of our data center capacity that we'll have online this year. So we could go and spend $1 billion on GPUs tomorrow, run rate earnings from that would be somewhere in the order of $400 million plus based on market data and it would use around 10% of our data center capacity. So again, we don't see it as competing with Bitcoin mining, we see it as complementary and additional. Now in terms of cost of capital and optimizing it, we see this as a really important opportunity or inflection point for the business. It's no secret that Bitcoin miners have struggled to raise debt type financing. It's a volatile in commodity, it's been a volatile business -- but all of a sudden with the introduction of an additional earnings line and importantly an additional view on the collateral within the platform conversations with lenders are looking far more prospective in terms of -- terms of potential debt. To elaborate on that more, visualise yourself as a lender, you go up to your credit committee and they say you're lending to a Bitcoin mining business. What happens if Bitcoin goes to zero? Now most of us, if not all of us on this call, understand the probability of Bitcoin going to zero is very low. But that's not the point. Those types of questions get asked. So all of a sudden what's the secondary value of the infrastructure. Having now proven that our data centers can be used for alternate use cases, all of a sudden there's a collateral value in the data centers. The infrastructure, the land portfolio, the power connections as we know, which really opens up a very different lens for these credit funds and prospective debt providers. So this is something we're quite excited about pursuing. We're in a number of conversations at the moment. If the terms are right, we will pursue some corporate level debt financing, but equally given the accretion of continuing to use existing capital sources, as Lincoln went through a couple of slides earlier, we certainly don't feel the pressure to do this, so we'll make the right decision over time. The other interesting aspect to scaling the AI cloud service business, so far as capital is concerned, is there appears to be substantial demand for GPU financing structure. We saw CoreWeave announced a $2.3 billion GPU financing structure led by Blackstone. We then saw, only a couple of weeks ago, Lambda, another cloud service provider, secure $500 million in GPU financing. Again, we've now proven out our cloud service. We've had exceptional customer feedback from our customers. And importantly, we've got the PowerPoints. We've got the data centers. And the same cannot be sit for other cloud providers, certainly not all of them, where essentially they sit in the middle of a data center and the end customer, and they own the computers. And I think part of the feedback we're receiving and trying to dissect why it's been so positive, I think is down to largely two things. One is our vertical integrated platform, where we're not sitting in the middle of the data center and an end customer. So when the end customer wants to optimize some of their machines, they want to update some software, firmware, they want to do something on the ground in the data center, if you're simply a cloud service provider, all you can do is lodge a support ticket with the end data center and wait for them to come back to you. You're governed under [certain terms] (ph) of your Service Level Agreement. Now contrast that to IREN where the customer picks up the phone to us, our CTO, our customer service and all of a sudden they can do absolutely anything live time on site. We've got 24x7 people on site and we own and control everything all the way from the soil, the concrete foundations, the steel structures, the ventilation, the network cabling, all of the infrastructure, the software, the firmware layer. So it's very easy in relative terms for us to tweak, make adjustments and optimize the service offering for these end customers live time. The second element having received again really positive feedback around the performance of our clusters and you can see the quote on your screen around performance of one metric being 3 times bigger than any other hardware setup one of our customers has used and we've been trying to dissect why is this the case. We've set it all up properly and our theory something to be tested is that it's due to the rack density that we're operating under. So to step back traditional data centers, 80% of traditional data centers operate at 5 kilowatts or less rack density according to the Uptime Institute report. Nvidia GPUs require 40 to 45 kilowatt rack density based on their reference architecture. Now, if you're installing a rack in different rooms because you can't manage the power, you can't manage the heat in the ventilation, all of a sudden you've got these servers spread out across quite a distance within a data center. Now Jensen, the NVIDIA CEO, has been out there talking about things like the data center is the new computer and when you zoom out you think about over the last 10 to 20 years, a lot of the innovation in compute at a chip level has been minimizing and shrinking the gate width on the transistors to enable the signal to travel back and forth faster and more effectively. Extrapolate that out to the data center where all of a sudden these GPUs are now talking to each other to crunch training models to run these AI models and it stands to reason that if they're all spread out and there's a lot of latency between the GPUs go back and forth then there might be a lower performance as compared to someone like us who is operating 70 kilowatt rack density. So all of our servers are in one spot, it's all tight. So in theory the latency is very low, the signal is communicating very quickly. So it is possible that is leading to the exceptional service reviews that we are getting but we're testing that further to try and validate. In terms of the rest of the strategy on our AI cloud service we're currently testing the on-demand market. So we've proven the model with bilateral contracts and now we're going down the pathway of testing an on-demand service, which is really exciting because we believe we understand that there are a number of smaller users and a number of bigger users that don't want to sign contractual commitments medium or long-term for compute. It does deliver a much higher price for us versus a contracted relationship. Yes, it's at a higher risk. But for us using some of the servers to do this, to build relationships, to learn we're in the process of developing software actually to allow true burst up and burst down, such that essentially prospective customers can come onto a website, click a drop-down box and start utilizing our cloud service. So this type of capability will take our cloud service to another level and it's something that we're quite excited about. So in terms of the outlook for this technically we've proven it up, commercially it's looking really good with the likes of Poolside, the on-demand market and we're now in the process of working through the optimal way to scale this up further through capital, maybe it's GPU financing, maybe it's other sources. Now, on to a little bit about our power and land portfolio. So I mentioned at the outset that Will and I, when we set this business up, we had a view that over the next 10 years to 20 years the world was going to have a very large growing, exponentially growing demand for compute and power from renewable energy sources to power that compute. So developing power and land is something that doesn't cost a lot of money but it costs a lot of time. It takes years identifying sites, securing the sites, putting in grid connection studies, building out grid connections. It's something that historically has taken three, four, five years and now with the onset of a tremendous amount of demand from Bitcoin miners, hyperscalers, data center providers, we are being told that these timelines are getting pushed out five, six, seven years. So what does this mean? The opportunity for us has always been to organically grow into this capacity, but we're now engaged in conversations with various stakeholders that continue to triangulate and validate that this data and power crunch is real. Morgan Stanley, Goldman Sachs have all released reports in the last month. In fact the Morgan Stanley report went into some detailed quantitative analysis around what the value of having power and land was and they came up with a number of $5 to $12 per watt. We've got 3 billion watts so that implies a tremendous amount of value in the portfolio. Is it worth that? Is it worth something different? We have no idea but certainly this is the time in the market to start finding out. So we're undertaking a process to explore various structures, everything from prospective sales of some of our development sites to joint ventures over our development sites where we could build, own and operate some sort of shell, provide a co-location service, provide a cloud service. We're talking to a number of the technology companies, we're talking to a number of end investors, the large banks, and it's something that we're excited to pursue over the coming months. Ultimately again we don't feel any pressure to execute on anything specific here. We're in a great position to continue growing organically and utilize this power for our own purposes as a going concern. So on that note I'll pass over to Linc to continue with some financial summary before Belinda then takes over from him. Thank you.

Thanks Dan. This page is actually exactly the same one from our previous half yearly update. We've just refreshed some of the economics based on current market conditions. And as you can see, Bitcoin mining margins and returns on a post-halving basis remain strong. And as we scale the business to 30 exahash at current Bitcoin prices and network hashrate. We've calculated illustrative electricity cost per Bitcoin of $17,000 and annualized hardware profit of $408 million driving sub two-year paybacks. And those economics there were calculated on a Bitcoin price of about $63,000 a couple of days ago. Turning to the right hand column the AI cloud services business similar economics to what we've discussed previously with very high margins at the gross margin level above 95% driving approximately two-year paybacks on the GPU hardware as well. So just to reiterate we continue to see very healthy margins and attractive payback periods across both lines of business, which underpins why we're continuing to invest in growing both lines of business. Stepping through to the next page before I hand over to Belinda. This is the first financial reporting period where we've recorded revenue for our AI cloud services business. And you'll see that come through on the face of the P&L. And as you can see from February to today, a rapid ramp up in revenue which coincides with the onboarding of Poolside, as a customer in February. You note that they also announced an upsize and extension to their contract to 504 GPUs, which was effective in mid-April. So we should start to see some of that increased revenues from the upsized contracts start flowing through to the revenue line in May and in the coming months. And on the right hand side of the chart, we've just highlighted there the $14 million to $18 million revenue opportunity that's associated with the current GPU fleet that's already operating. I'll hand over now to Belinda to take us through the rest of the financial section.

Thank you, Lincoln and Dan. So good morning to those in Sydney and good morning to those in North America. Thank you for joining us today for our first quarterly update. To start with, I wanted to highlight our positive cash flow from operations of $48 million to the nine-month period ended 31st of March 2024. This includes 129 of receipts resulting from the 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. Turning to the consolidated statement of profit and loss, during the three months ended 31st of March 2024, we reported a positive net profit before tax of $12 million and net profit after tax of $8.6 million. As we do not hold Bitcoin on our balance sheet, this result has been achieved without any reliance on Bitcoin mark-to-market revaluation gains, includes non-cash expenses of $17 million. Moving on to adjusted EBITDA for the quarter, the EBITDA increased from $13.9 million to $21.8 million. This is due to the Bitcoin mining revenue increasing from $42 million to $53.4 million. As the average operating hashrate increased from 5.6 exahash to 6.4 exahash, resulting in 1,003 Bitcoin mined, at an average realized price of $53,200 being a 45% increase in price during the quarter. Our average net electricity cost per Bitcoin mined increased from $14,100 to $19,300 primarily due to lower number of Bitcoin mined during the quarter as a result of increased global mining difficulty. Quarter-on-quarter, our other costs remained relatively flat. Looking at the adjusted EBITDA for the nine months ended 31st of March 2024, we had a record adjusted EBITDA of $42.5 million. This is an increase of 32% year-on-year and very pleasing to see as I talked about earlier the nine months cash flow being $48 million. So directly converting into that cash flow from operations. Bitcoin mining revenue for this period increased from $41.3 million to $129.9 million as the average operating hashrate increased from 2 exahash to 5.8 exahash and we mined 3,371 Bitcoin at an average price of $38,500 being an 87% increase in price. Average net electricity costs during this period increased from $9,900 to $15,400 primarily due to the increased global mining difficulty. And our other costs increased from $32 million to $36 million with the Childress site operational in April, 2023. Moving on to our consolidated cash flow. Net cash and cash equivalents increased by $191 million for the nine months ended March 31, 2024. The increase in net cash flow from the operations operating activities was $48 million due to the increase in the average operating hashrate coupled with the higher average realized Bitcoin price. The increase in net cash used in investing activities was $188 million due to the expansion at Childress, as well as our purchase of 816 NVIDIA H100 GPUs. Increase in net cash from financing activities of $331 million primarily due to net proceeds received from sales sold under the ATM and ELOC facilities. Lastly, moving on to the balance sheet, we had a cash, closing cash balance of $260 million at the end of March, no debt facilities and as I've mentioned, strong operating cash flows during that period. The cash position further has strengthened to $322 million at 30 April 2024. During the period of the quarter we raised [$294 million] (ph) from the sale of approximately 56 million shares and post April 1, we raised a further $45 million from the sale of 8.2 million shares. So we have a strong balance sheet with total assets of $724 million, which will provide us flexibility to fund and grow into the future. I'll now hand back to the operator to take any Q&A for our session today.

Thank you. [Operator Instructions] Our first question comes from Lucas Pipes with B. Riley. You may proceed.

Thank you very much, operator, and good afternoon, morning, everyone. Ben, my first question is on the incremental exahash and the optionality there that you outlined earlier on the call. Could you kind of walk us through where this would be deployed and then also just in terms of the potential timing around all of that. Thank you so much.

No problems. Thanks Lucas. So in terms of deployment, -- the goal that we will upgrading our year-end data [technical difficulty].

Sorry, Dan.

Dan might have dropped off there. Lucas, I'll take that question. So the majority of the capacity is being deployed into Childress, and as we've previously disclosed, that is scaling to 350 megawatts this year. So the vast majority of the new exahash growth is going to come on at Childress throughout the remainder of this year. And in terms of hitting that 40 exahash in the first half of next year that is also going to be deployed at Childress. We did mention fleet upgrade as well. So some of the capacity will be used to upgrade the existing fleet of S19j Pro machines, which are primarily located in our British Columbia data centers. So it'll be a combination of both, but primarily at Childress, where it's a single-site expansion currently operating at 100 megawatts and scaling to 350 megawatts this calendar year.

Lincoln, thank you so much for the additional color there. I appreciate it. Best of luck. I'll turn it over.

Thanks, Lucas.

Thank you. One moment for your questions. Our next question comes from Joseph Vafi with Canaccord Genuity. You may proceed.

Hey, everyone. Thanks for the update. Good morning to you. A few questions. Number one, on the 10 exahash option with Bitmain that you have, I'm just wondering if you could kind of go through what would be some of the reasons you might exercise that. Obviously, if Bitcoin is a lot higher, that may be one, but just kind of balanced against maybe some of your other initiatives that you've got going on. And I know there is a lot of moving parts but just trying to get inside your head on that and then I'll have a follow-up. Thanks.

Thanks Joe. So just to clarify that's the 10 exahash option to go from 30 to 40 exahash?

Right exactly Dan.

Yeah perfect. Look I expect we'll exercise it. The presentation is clear in terms of the metrics when we can mine Bitcoin at around that indicative cost of $17,000 a coin with Bitcoin where it is today around [65] (ph). That's a pretty healthy gross margin and a pretty good exposure going into what we believe is the next Bitcoin cycle. And all we need to do is to continue to build out the same data centers at our existing site. So for us, we're planning to execute it and continue to grow through the cycle. And when you see those market metrics that Linc spoke to, where miners above 10 exahash capacity are being valued at $135 million per exahash, and we can continue building out an exahash for $30 million, I mean, the value creation to shareholders is absolutely crystal clear. And we're only going to grow when it's accretive. As you would have seen in the last six weeks, I think we used the ATM three days. We've sold a grand total of 5 million shares. Now, yes we've lodged a shelf to replace the existing one. But if we use it, it's to generate that type of value creation for our shareholders. So we anticipate continuing to grow and to ultimately exercise that 10 exahash expansion from 30 to 40 early next year.

Great, thanks for that update. And then just maybe one quick follow-up here on the three gigawatts of development sites that you have. We could maybe drill down on that a little bit. I know, I think Dan, you said that you may be entertaining some deals and that we might hear about that soon. Just want to understand, I mean, with things moving so fast, does it make sense to structure a deal now or perhaps wait another nine months to a year when perhaps there's even more scarcity in the market relative to power and development. Thanks a lot.

It's a great question Joe. The short answer is I have no idea. All you can do is work through the process and see what options are presented in front of you and make a decision on that basis. And I can make the case both ways for waiting. You know, if we've got this level of scarcity now in respect of an asset, i.e. power and land that can take five years to eight years to develop, then the scarcity factor may indeed increase over the next six months, 12, 18 months. But equally, I'm a believer in bird in the hand. If you get a compelling offer that can demonstrate considerable value to shareholders, particularly recognising out market capitalisation now as compared to some of these numbers that have been thrown around in terms of the value of the asset base we have, then we're compelled to explore that and we will. What I also should say is selling off or doing a joint venture on one of these sites is going to have realistically a very limited impact on our organic growth. You might recall we've always said we've got over a thousand megawatts of additional development sites where we don't disclose additional detail. That was the case before we announced our 1400 megawatt site late last year and it was still the case after we announced that 1400 megawatt site last year. So over a 1000 megawatts under development could indicate any number above a 1000 megawatts. So I feel like we're long an asset that we can look to monetize without necessarily compromising our organic growth plans.

Sure. Fair enough. It's nice to have options. Thanks a lot, Dan.

Thank you. One moment for questions. Our next question comes from Mike Colonnese with H.C. Wainwright. You may proceed.

Hey, good morning, guys. Congrats on the quarter and great to see you delivering ahead of schedule again with your build out here at Childress. So you guys have taken a measured approach to rolling out the AI Cloud services business, which has certainly shown early successful pull-side with the upsize deal there. And now that you have this proven model in place and service in place, how have conversations evolved with potential customers for this business and how should we think about growth of the business based on the current pipeline and some of the funding opportunities you spoke to there?

Yeah, thanks Mike. Look, it's an evolving space. Clearly we've received really good reviews. We've received really good demand for the cloud service. We see the mainstream narrative around the world being short GPUs and short data center capacity. And we seem to be living that day-to-day. But the focus for us is continuing to develop this service and expand its capability, which is why we've taken a group or a cluster of the last units we bought and developing this on-demand service where the pricing is much higher and we believe that there might be quite a deep market that's attractive where you can provide that true on-demand burst up, burst down service. We're continuing to engage in conversations with everything from smaller customers to larger customers on multi-year contracts and all of that goes hand in hand with the financing [for us] (ph). So looking at different structures around how we finance it, there's a number of different type of structures that are being suggested around GPU financing. So it's just working through all that and looking at the best way just to match it up and continuing to grow the service. I must say over the last couple of weeks this has been absolutely on nailing down this 30 exahash trajectory, finalizing the transaction with Bitmain and giving ourselves a very clear path to industry leadership on the Bitcoin mining and now we can really focus back on the AI cloud service. We're engaged with a number of those large customers, financing providers and yeah, let's see what happens over the next few months.

Great. Appreciate the caller and just a follow-up from me. What did power costs come in at during the quarter at Childress and how are you thinking about pricing in the market and uptime versus curtailment levels at the facility as we look out through the balance of the calendar year.

I'm happy to take that one, Mike. So it is touched on in our presentation, but $0.033 a kilowatt hour, sort of average price at Childress since, sorry, this financial year. And that's just based on our experience. Obviously it's a volatile market. We trade energy in that market, and that's for purely economic reasons. So $0.033 per kilowatt hour at Childress, which is a very attractive rate of energy costs. From a curtailment perspective and downtime uptime to your earlier point, we're always making life decisions around whether we take the power to mine Bitcoin and we sell it back to the grid. And we've proven up an ability to operate the data centers through relatively extreme conditions up to 111 Fahrenheit last year in the summer without skipping a beat. So these facilities are operating near 100% uptime with any curtailment or downtime basically purely for economic reasons. And we expect that to be the case, as we continue to scale the data to [consider] (ph) capacity there.

Great. Thank you for taking my questions.

Thank you. One moment for questions. Our next question comes from Joe Flynn with Compass Point Research & Trading. You may begin.

Hi, guys. Thanks for the question. I was hoping you could provide more color, potentially on the two-year payback on H100s. And ultimately, if you expect data center design specs to change materially as we move to further generations of chips, such as the Blackwell. And yeah, just any color there would be helpful.

Thanks, Joe. The two-year payback on chips is a rough guide. Reality is when you offer on-demand and shorter-term contracts, you're receiving a price where that payback will be shorter than 24 months. Equally, if you sign a longer-term contract, then the payback might be a bit higher than 24 months. Market pricing for GPUs is pretty transparent. We can see how many dollars per GPU hour. NVIDIA H100s cost around $40,000. So really it's back solving into that to get that rough two year payback on the chips. In terms of data center configuration, look it's something to watch but what's becoming really apparent is very few people can manage the 70 kilowatt rack density that we're operating at. We're seeing traditional data centers have to go down this liquid to chip and liquid cooling path where they can move the problem outside the data center. They physically cannot deal with the heat and the ventilation and the airflow in these multi-story legacy data centers. So they're having to go down the path already of retrofitting and improvising to deal with that rack density. And the reference architecture for NVIDIA chips is only that 40 to 45. So it's really hard to see how our data centers operating at 70 kilowatt plus rack density fit the purpose and somewhat future proof as these new chips are released.

Great, that's all for me.

Thank you. One moment for questions. Our next question comes from Paul Golding with Macquarie. You may proceed.

Thanks so much. Just a quick question on the supply chain for GPUs. Just wondering, Dan, if you could comment on availability. Is the main constraint really just capital at this point? In other words, if you have the power capacity available based on your energizing Childress and the subsequent data center, if you had it available, are you confident you'd be able to get the GPUs enhanced through your suppliers. And then as a follow-on to that, just briefly, and looking at the table showing the 95% plus gross margin, is there a scenario that we should keep in mind where it might make more sense to unplug more ASICs and plug in GPUs, incremental GPUs. Thanks so much.

No, a pleasure Paul. So on the last one, absolutely there are scenarios where you might unplug some ASICs and as I went through in the presentation, you're almost not mutually exclusive where you can continue to manage both of these businesses in parallel, because the GPUs simply cost so much. But they're so expensive per unit of power consumed. So if we secured a financing and wanted to go out hypothetically and buy $1 billion worth of NVIDIA H100s tomorrow, 25,000 units, that would displace a bit less than 10% of our overall data center portfolio. So we're still able to be a very large scale Bitcoin mining business in addition to having all this optionality and additionality on the AI side and in terms of timing look it's a moving feast and week to week it does move around but you know guidance at the moment we can probably stand up new clusters within two months of go to work.

Great thanks so much and then just around availability it sounds like it's more so a question of capital but just wanted to confirm that the GPU marketplace would allow you to make the strategic purchases as needed.

I believe so. Current guidance that we're receiving from our suppliers is such that we could stand up new clusters within that two month timeframe maybe three or four if it was a very large size.

Understood. Thank you so much.

Thank you. One moment for questions. Our next question comes from Josh Siegler with Cantor. You may proceed.

Yeah, hi guys. Thanks for taking my questions. First of all, I was wondering if you could comment briefly on if you've started exploring and doing a little bit more work on the co-location aspect of the HBC business and kind of what that would entail to actually get set up. Thanks.

Thanks Josh. Yeah this is something we've been looking at and could form part of this kind of JV type structure to monetize additional land and power that we've got in the portfolio and it really goes to who's the end customer, what's the length of the contract, how do you finance that and what's the opportunity cost around using that capacity ourselves. So do we want to lease out the data center capacity and access to that power and land which we understand is scarce in the current market or do we want to own and operate our own compute unit whether it's Bitcoin mining ASICs or whether that's NVIDIA GPUs and monetize that capacity directly. So there's an opportunity cost, there's options around that. It's something that we're continuing to explore. But equally, like I keep saying in the presentation, we don't feel any pressure to engage and do a deal over a portion of this capacity when we can continue to monetize it out ourselves. So we'll just keep looking at the options that present themselves.

Yeah, understood. Appreciate the color there. I guess, you know, secondly, I'd like to take a second and just ask a little bit about what you're seeing in the market post-halving, if you're seeing different dynamics and you can kind of talk to how your unit economics are holding up, especially given your new growth plans. Thanks.

Yeah, look, it's interesting. My view publicly before the [parving] (ph) was that you wouldn't see any downward adjustment in the hashrate. I was proven slightly wrong, there was. And maybe there will be a little bit more to come, just given where the hash price is now and the squeeze that that is having on some higher cost miners. But ultimately, when you believe that Bitcoin is going to remain robust, you assume that the hashrate will continue to remain where it is and potentially trickle higher over the coming period. Private miners, you see a lot of perspective processes where people are looking to sell. We've seen some of the public miners informally or formally talk about putting themselves up for sale and for us, it's really hard to justify when we've got the ability to grow organically, when we can grow so creatively and with those metrics outlined in the presentation where by the end of this year in 7.5 months you know indicative cost per Bitcoin mined of $17,000 post-halving and generating that value for shareholders where you're spending $30 million in exahash on CapEx and the markets value in that at $135 million, it just seems very clear decisions for us.

Yeah, understood. I appreciate the caller Dan. Thanks for taking my question.

Thanks, Josh.

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

I think that's just over the hour now, and that's probably all we have time for this morning. Thank you for the questions. I think just to reiterate, 30 exahash this year in 2024, pathway to 40 exahash in the first half of 2025. We're getting into very accretive growth, with fastest growing miner. Last year we fully intend to continue that trajectory. And we look forward to meeting the shareholders after this result. Thank you very much for your time, everybody.

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