IREN Q3 Earnings & Outlook | Latest Bitcoin Stock News | IREN Co CEO Daniel Roberts Q&A | IREN

Sure thing, and thanks, Jay, for having me back. Always a pleasure. Look, it was another strong quarter operationally and financially. Our second consecutive quarter now of profits after tax with a 28% increase quarter on quarter, which is great. I think we're one of the only public miners delivering that profitability, which is great. We achieved record revenue, record EBITDA, $148 million and $183 million respectively. What's really exciting about that is that's all based on an average hash rate during the quarter of about 30 exahash. We have already announced on April 16th that we're at 40 and we're going to 50 in the next few weeks. So there's some natural organic step up coming to those numbers. So we anticipate that earnings momentum to continue into the next quarter and that continued cadence of delivering 50 megawatts of power-dense data centers every month. I think we'd be hard-pressed to find any other data center developer on the planet who's delivering that capacity at that velocity. But look, we can execute infrastructure at scale. We've shown that. But importantly, we're operating it efficiently and profitably.

Yeah. No, thanks, Anthony. And look, the only surprise here, I think, is that we've delivered on it. We've clearly telegraphed it. We've clearly telegraphed our energy costs. We've explained the economies of scale amortizing the level of fixed cost bases as we continue to expand. So again, referencing the average 30 exahash last quarter, the fact that we're hitting 50 exahash in the next few weeks, there should be another benefit coming again. So roughly, energy cost in that 41 today is about 24 and overheads about 17,000. Importantly, that funds all costs associated with everything including AI. So all the AI revenues are for free on top of that. So, we're absorbing 100% of the cost base, indirect, direct of the business in that $41,000 all-in cost. So, that's great. It gives us a strong buffer in the face of network or price volatility. In terms of electricity cost, it was 3.3 cents for the quarter, but I feel like we've really got some momentum in the last probably six to eight weeks. We've got a couple of data analysts who are working around the clock and I'm smiling because it's just fascinating when you dig into the intricacies of how you can optimize your power price. So there's a few anecdotes that we can talk about potentially, but I think we had 2.5 cents in March, 2.9 cents in April. So I'm optimistic that we can continue to at least match that quarterly cost if not beat that, subject to market conditions more broadly. So we're not taking power spot blindly. We're actively managing that curtailment. So look, yeah, it's good and the unit economics are really positive and we're happy with where we're at.

Yeah. Yeah. So what we have successfully negotiated with the utilities in Texas is the need not to hedge — probably reverse language. But historically these utilities have required miners and speculative uses, or what they deem speculative uses of energy, to implement hedges to protect themselves because the risk is power prices go to $2,000 a megawatt hour. People can't curtail, they default on that and ultimately it becomes the utility's problem. So it took us a long time and you might reflect back last year on this, and we successfully demonstrated the robustness of our systems, all the levels of control, both digital software as well as physical controls. We successfully negotiated going to a spot model and what that allows us to do is to hedge most efficiently and effectively because we hedge effectively every 5 minutes. We have this energy market intelligence. We have these data analysts. We have these algorithms and software stack that allows us to trade the market dynamically from a physical sense because we've got a physical hedge with our compute there and ultimately remove all those margins associated with hedging counterparties, etc. So look, it's working.

Yeah, so it was a 33% increase in revenue, the AI cloud service segment, to $3.6 million. Look, frankly, we're pretty focused on the AI colocation opportunities for Horizon 1 and Sweetwater. But there is certainly opportunity to scale the AI cloud business. And ultimately it supports our broader AI ambitions. So it's important to remember that we are one of the few data center developers operating our own GPU cloud and the knowledge and experience we get from being in those customer shoes and actually doing something rather than talking about it is really valuable. So we feel that it really complements the colocation opportunity pipeline. Now whether we scale AI cloud right now is ultimately a question of risk-adjusted returns. So, look, we're negotiating GPU financing at the moment and working with customers in parallel, but we've got to get the risk-return equation right. We'd love to do larger clusters. There's definitely a step up once you hit that 1,200 in terms of the customers and the level of demand. So look, we're working through that, but if we do do something, I'm comfortable that it will be a really good risk-return equation and be a robust investment thesis. We are one of the few companies that can stand up capacity for additional cloud in 6 to 8 weeks. We've got 47 megawatts immediately available at Prince George data center where our existing cluster operates. You know, B200s, even B300s are starting to become available. But obviously the economics need to make a lot of sense as we're displacing Bitcoin capacity to the extent we do this.

We expect it to increase yet again. We're installing latest generation hardware which improves that overall efficiency mix. And we're also installing more compute at Childress where the power cost is cheaper than British Columbia. So, yep, we will see further incremental improvements there.

Absolutely. I mean, that's right. The mining business is highly cash flow generative even at low Bitcoin prices. And you can see just how much 50 exahash is capable of generating at different hash price assumptions on slide 16 of our earnings presentation. And we do overprovision slightly for installed capacity to account for potential DOAs to optimize usage of our data centers in case of longer-term repairs and maintenance needs. So we are confident that we'll be able to operate around that 50 exahash level because there is a buffer in what we ordered. Look, it'd be great. We could scale hash rate rapidly and I think we have it by June, it'll be a 50x factor of growth in two and a half years which I'm not sure anyone else at this scale has been able to achieve. So we're one of the largest lowest cost Bitcoin producers on the planet. We have the power, we have the land, we have the team. And we haven't been reliant on M&A and I guess smaller, less efficient sites for growth. But coming to your question directly, we've made the conscious decision not to expand. We can change that if the market conditions change. We can absolutely change and I'm happy to elaborate on what that might look like. But we're seeing really attractive opportunities on the AI side of the business that simply have the potential to create more shareholder value. So that's where we're focusing.

Yeah, so I think if we step back, like value creation for shareholders is all about the wedge between your cost of funds and how you're allocating those funds. So if your cost of capital for Bitcoin mining is really, really high because the market isn't valuing it, then it doesn't matter how well Bitcoin mining is returning in absolute terms, the value creation wedge isn't necessarily there. And when you look at our market cap and sector-wide today and compare that to replacement cost or new build hardware, there's not an observable material value uplift for shareholders today in building that out. So one of two things can happen. Your cost of capital can drop. So if share prices in the sector run, then that implies cost of capital has fallen. All of a sudden it might be accretive. If you're spending $30 million in exahash to build out capacity and the market starts to value that where we did see it over the last 12-24 months at $100 million plus, then all of a sudden there's that $70 million plus per exahash of value creation to capture for shareholders that doesn't exist today. The other side of the equation is the return that you're getting and effectively how much it's costing you to build it. Sitting here right now today, break-even periods on mining hardware really high relative to where they've been in the past. I mean, you look back over the last seven years that we've been working in this sector, they're probably almost at all-time highs in terms of that, really approaching that, you know, two-year time frame, maybe even a bit longer.

Look, absolutely with any new business you need to resource up. There's only so much we can do as an executive team and me and Will and Kent and the broader team. So look, it's just been this steady build-out of network architects, HPC system engineers, cyber security specialists, automation, software, cloud sales and marketing, you know, specific stuff to InfiniBand, the Nvidia networking technology. So, you just got to work through the process and it's that balance between the cart and the horse and what comes first. As in, we're not putting the cart before the horse, like you've got to create the business and then backfill it. So we're big believers that as founders it's our job to go and create the business and then as we get the business, as we get that validation, ideally as we get revenue coming in, then backfill things that we can backfill, resource up, then go again and make it bigger and better. And we're at a good stage in our journey there and I think you can see that in the way that we've really controlled our corporate overheads over the last 12 to 24 months to create that broader profitability. And this is why we were so comfortable of baking in all of those costs, everything in relation to what I just mentioned, into our Bitcoin numbers just so there's no debate, like we can mine Bitcoin really profitably and have this free option very clearly on building out a big base and platform to launch at the AI infrastructure story.

Yeah, sure. I mean, we're in active negotiations with several multiples of customers. I mean, both in relation to Horizon, but also variations of Horizon, variations of our design, everything from turnkey delivery for their own design through to powered shells. So, there is a broader conversation that encompasses the whole of Childress as well as Sweetwater bundled up in this. But to zero in on Horizon specifically, sitting here today, I'm highly confident we're going to contract that prior to completion in Q4. Like we're at pretty advanced stages of diligence and commercial negotiations with a party. We've got colocation full form agreements being circulated. So, it's just a matter of moving through the process. And, you know, these things may take a little bit longer than ideal, but it's just the nature. I mean, don't forget that 50 megawatt of IT load supports well over a billion dollars of computers. People don't just click their finger and say, 'Oh, can you give me a piece of paper so I can sign it.' Like, there's a bit that goes into it. But we're feeling really good about not just that, but also the additional demand or conversations that that's opened up because what has become really apparent to us in terms of what's been communicated is there's very few sites in North America capable of delivering 20 megawatts plus of liquid-cooled capacity this year. There are, according to several of these prospects, none that have then the expansion capacity that we've potentially got at Childress. And a lot of these players like us, we're not like, the first 50, that's great, like we look forward to ticking that off. I'm not here for 50 megawatt of AI, these guys aren't either. Like it's all about the growth and pursuing this macro exponential trend that's unfolding. So they're looking at growth. They're looking at can you do 200, can you do 250? How quickly can you do that? Could we get options? What might that look like? And we're happy to have all those conversations because ultimately we're going down the same path. We've got 2.9 gigawatts. If it's all AI in two, three years time, then by definition, it's a good result because we're certainly not going to displace Bitcoin mining, unless it's clearly value-accretive.

Yep. There's a bit in there, Anthony, so I'll try and unpack it. So yes, it is a shame if and that we're the only Bitcoin miner that can operate profitably. It's not good for the sector. It's not good for any of us. So like all we can do is hope that others find efficiency in their business models and are able to get to that point. And in parallel we've just got to go on the education pathway because undoubtedly you do get that initial casting of the brush that comes with in this sector and the opportunity to differentiate, distinguish is clearly there. And I think people have been waiting to see the real profits rather than the hopes and dreams and promises. So we've got that. We've had some really great conversations with our largest institutional shareholders over the past 24 hours. I'm comfortable we will educate people and hopefully navigate to a more appropriate cost of capital valuation on our profitable Bitcoin mining business. So far as it concerns Childress, Sweetwater and the roll-out of different AI potential structures there, look, everything's possible but to some extent that's unhelpful, so let me break it down. We've got our own liquid-cooled design which we're implementing at Horizon. That's 50 megawatts. We are able to implement very similar, essentially identical in key aspects, in a retrofit for all the existing buildings at Childress. That has some advantages. It has some disadvantages because you're on a brownfield site and you've just got to navigate through that. Sometimes it's cleaner just to start on a fresh piece of dirt and build from scratch. So there's pros and cons to that, but there's reasons why there might be some time advantages or other nuances specific to customers. So for example, the expansion potential for a larger hyperscaler at Sweetwater is clearer. There's up to 2 gigawatt across Sweetwater 1, Sweetwater 2. The opportunity to contract for a smaller portion of that now but have some sort of pathway to getting up to the larger capacity over time has a lot more attraction. I think given the conversations we've had today on Childress, it has prompted a look at a full redesign of Childress. That's not to suggest that we've had conversations for people wanting 750 megawatts of capacity at Childress, but if you look at the aggregate demand it certainly exceeds that. So the opportunity to work up a multi-tenant colocation business at Childress utilizing that 750 megawatt, happy days, preserve Sweetwater 1, Sweetwater 2, the 2 gigawatts there as a prospective single-site opportunity. But ultimately if it's not single-tenant, sorry, single-site opportunity, but if it's not single-tenant, it's not single-tenant. Like there's an opportunity there, if it doesn't play out, it doesn't play out. I mean equally we've said for many years we've got a development pipeline and we don't advertise it because if you don't get that final piece of paper it's not real. You can talk about all sorts of potential power you want but until you get the contract it's one and zero. So we're going through the process at the moment and working through how much power we're likely to have between kind of that period of Sweetwater through to 2030, starting to work out opportunity cost of megawatts because this actually becomes, and I'm going on a little bit now but I hope it's helpful. It's helpful in the context of assessing different delivery models because you might have a powered shell opportunity where the capex is lower. The notional return might be a bit lower, but so is the risk. You may have a hyperscaler credit against it, the opportunity to put a high level of gear in. The returns still look good, but your absolute dollars monetized per megawatt of power is obviously lower than a turnkey full delivery and ownership of the data center, which is obviously lower again than if you own the GPUs. So there's the opportunity cost of power. Yes, some people might sit there and say, 'Oh, you got 2.9 gigawatts. Surely that's not a concern.' I say, 'Yep, well 18 months ago, no one would have thought signing a grid connection for 1,400 megawatt would have the value it did.' So it's a bit of foresight. So as we work through that development and our later stage development projects, just lining up the sequencing and it does inform partly the strategy that we might take on that front. But ultimately it takes two to tango. We can't... I can sign any contract you put in front of me, but I also need someone else to sign it. Otherwise, we haven't got a deal.

It does. But just to pick on like, we don't, I don't feel we need to deliver Horizon to prove our credibility or to validate who we are. I don't recall the last time, if any, particularly this year, that any conversation with any customer hyperscaler has questioned whether or not we can do this. Like at the end of the day, these are just jigsaw puzzles. And yes, there's a few more pieces in this jigsaw puzzle than there has been for delivering our 50 megawatts per month of air data centers, but they're jigsaw puzzles and you just go through the process of educating them on who we are, what we've done. They look at the execution track record. It's like we don't even go there. It's just straight to, all right, let's do diligence on the design. Let's talk about commercials. Let's work through what it all looks like. So yes, it is far more probable we sign a deal for Horizon prior to signing a differentiated type structure for either Childress or Sweetwater, but the two are not reliant on each other. Both are happening in parallel side by side across multiple different work streams.

Yeah, there's no intention to build data centers there without a client. Look, if Bitcoin rips to three, $400,000 a coin, then let's have a conversation around whether we install a bit more Bitcoin mining. But today, like, and we said this, and this kind of dovetails into the funding conversation. We need $250 million total across the rest of this year on a net basis, or that's what we expect to need to complete the 50x, complete Horizon 1 and complete a lot of site preparation works at Sweetwater. So that includes site grading, internal roads, substations, transformers, stepping down the high voltage power and basically getting it site ready for construction and utilization at a data center level. So we do not expect to start any construction there prior to customer contracts.