$iren Earnings 5.7.26

Hosted by @₿itcoin ₿utcher 🥩 🐑 🐷 · 2026-05-07 · Tags: IREN

TLDR

Participants viewed IREN's Nvidia partnership as a major long-term catalyst that could secure GPU supply, support rapid AI-cloud expansion, and enable billions in future contracted revenue. However, disappointing quarterly AI revenue, delayed Horizon and Microsoft deployments, financing and dilution questions, and poor earnings communication created substantial near-term uncertainty.

Speakers

Notable quotes

Transcript

₿itcoin ₿utcher: What's going on, everyone? It's 9:03. Time to get this started. Let's see if Franz is in the crowd. Let me get him up, Sage. Maybe Small Cap will join us. Let's see if Franz can get up here on stage and then we'll get things going. But interesting after hours. I think it's kind of a upgrade from February, but definitely a lot to sort through here and we're going to try and figure it out together. All right, I think my wingman's on stage. Franz, how are you? Hey, man, I'm good. Pretty good. A bit tired. I woke up at 3 A.m. It's 8 o'clock here now in the morning, so I'm up for five hours, just drinking my second coffee. I'm still processing everything that happened in the last hours. Yeah, I'm kind of with you. I think... Part of my fear going into this call was every day we're doing some sort of due diligence, looking at different job postings and different sites and different geographies throughout the world. And they jammed a lot into a 30 minute presentation. And I think people are still trying to figure it out because the obvious headline is the NVIDIA partnership and that's great and I'm sure that'll be the majority of what we talked about but still earlier in the week had the Mirancis acquisition and then I can't even this is the other thing right now which I would give everyone the benefit of the doubt, but I couldn't even tell you the name of our Spanish subsidiary that we acquired because I can't download the earnings presentation right now. But I do know that we have European exposure now and, you know, 400 some megawatts. But I was listening to the earnings call in traffic driving home on I-75 here in suburban Detroit, but So to kick things off, I mean, my biggest takeaway, maybe that's where we'll start, our biggest reaction, I'll take the easy one, is the partnership with Nvidia. And they have the right to purchase Iron shares, 30 million of them, or excuse me, I'm probably getting that number wrong, up to 2.1 billion at $70 per share, whatever that share count comes out to. I believe it was 30 million, but I think what people right away don't understand how it's different from some of the other strategic partnerships Nvidia's had is one, we didn't sell shares to Nvidia directly right away. And this is going to happen over time as they deliver GPUs. Maybe Franz, you're a little more versed in the vesting schedule. I don't know if they've all how it works, if it's fully vested at the end of the delivery of the 600,000 GPUs that were referenced, but Dan kind of touched on it on the call and I think it's important for people to realize that NVIDIA is working with iron like they're incentivized if iron's at $200 a share in the future, they made $130 per share on each share that they purchased. So there is a strategic alignment between both parties for Ayren to do well. And I think the biggest tell for me is I personally, I'm excited about the long-term trajectory of the company, and we'll talk about some of those strategic moves that were made But in the interim, I think Franz, I think you could speak to it. I certainly will speak to it, but I was expecting a higher AI cloud number than 33 million. I mean, it did double quarter over quarter, so it is progress, but we're, and maybe this was part of the, maybe Mirantis helps us deliver or excuse me, deliver that compute sooner by installing faster. But I I also just like I don't know why any of the analysts lack the balls, for lack of a better word, to just ask the question like what is going on with GPU deliveries with between Lenovo and and Dell. So that's the other thing with all this. If we're now on the same level playing field as Coreweave and Nabius, I would expect our delivery cadence to be improved. So that was kind of my initial thought. But there was just some like good spoken to and again, like super positive about the long term trajectory. But a few times where Dan and Kent kind of were talking over each other mistakenly and like just kind of a strange call where I think there was so much to discuss that I think it's going to take time for people to take another listen at this. And maybe that's why the initial reaction was that pop to 72 because you see a partnership with Nvidia, which may have been confused with a co-location agreement or some sort of larger contract. But to be fair, we did have 60 Megawatts of IT signed to Nvidia for five years at $3.4 billion and it appears to be a better rate, you know, higher approximately at $11 million a megawatt as opposed to the nine that we sold to Microsoft. But it's unclear with the GPU count. Like that wasn't, you know, the analyst, I believe from Goldman asked for that question and Dan didn't provide. So I don't, I'm again, I heard references to Mackenzie and this was all driving, so I didn't see the slide deck. So my apologies. It's more kicking it off and this is where Franz will fill in the gaps. But if it was in Childress and it was the 17,000 GPUs that were earmarked in the 50,000 GPU purchase from March, then, you know, it showed to be a little over $4 GPU hour. But I'll just pause here and Franz, why don't you give us some of your initial reactions or where you want to take this? Well, Let me just first answer your initial question. If I could give a little bit more color on the NVIDIA, let's just call them warrants. I'm not sure if it's technically a warrant, but or an option or a right to invest is what Iren calls it. So NVIDIA has the right to invest 5 million shares of iron, basically, or buy 5 million shares of iron, but it's called an investment right upon deployment of 100,000 GPUs. So every 100,000 GPUs that iron actually deploys. So we're not talking about, you know, purchasing or you know, anything of that sort, it needs to be deployed. That gives Nvidia the right to buy 5 million shares of iron for $70. So those those options are not currently exercisable because, you know, Nvidia will need to fulfill a delivery and iron will need to deploy them. So there will most likely be an OEM in the middle as well. But as Jim stated, this, is obviously mutual beneficial for both parties, but for Iron, this solves a big bottleneck on the procurement side. We know that obviously Iron has had delays with GPUs, and I think that that's also partially an answer for your, the low AI revenue for the quarter. But I'll touch on that later. I think it's just very important to understand that by, you know, going into this partnership agreement, Iron is fixed in for 600,000 GPUs across, you know, over five years, which is 120,000 GPUs a year, which I have basically estimated to be 300 to 350 megawatts of IT load per year. So if Iron can construct 350 megawatts of IT load every year, that means they can deploy 120,000 GPUs every year. That means Nvidia can buy 5 million shares of Iron for $70. But then you've got to ask yourself if Iron is able to finance the build out, of these IT low, let's say 500 megawatt of gross capacity every year, and also is able to deploy these GPUs in there. That means you're already, you know, sure of delivery execution for whoever that may be, right? So the customer is most likely on board, at that time. So it is a very conditional dilution. And the reason why it is attractive for NVIDIA is because obviously a number of 100,000 or 120,000, well, it's vests each tranche has 100,000 GPUs. If Iron can deploy 100,000 GPUs in 2027 under this agreement, that means that we will add a significant amount of ARR because no GPU is sitting idle. So yes, there are some gaps in the whole cycle. You know, who's going to finance what, you know, how are they going to finance those GPUs and those data centers and how quickly can they really build it out? There are obviously some uncertainties around that. But the thing to understand here is that It is very conditional and the conditions under which these GPUs vest are basically saying that we are already good for our delivery and our revenue, right? So when you're talking about the dilution, yes, it will happen, but the shareholders will already be, you know, enjoying a nice uptick in installed capacity at that point, right? And we know that the market really likes backlogs and obviously we will sign some going forward. I will also touch on that later. But it's just very important to understand that this is not just, you know, some kind of investment like other Neo Clouds have enjoyed where basically Nvidia is buying at spot price. This is very different. And it solves a big problem for Iron because we are talking about delayed deliveries of GPUs and it's being reflected, you know, for the most part in Prince George, but also now with McKenzie and Childress, these B300s are going to arrive in the second half of 2026. And now the new NVIDIA contract for 60 megawatts of air-cooled capacity in children is targeted for 2027, right? So these GPUs are all being delivered much later than we initially thought. And it turns out that a big part of this bottleneck is the HPM. And Jim could probably talk a lot more about this. But Iron has basically fixed their HBM problem for the coming five years for 120,000 GPUs a year. And I think that nobody can quantify at this point what that really means, because I've seen people throwing around numbers that Nvidia got a sweetheart deal and that Iron has sold their shares short. But it's actually, there is a secondary market for HBM. There is even like a black market where, Jim gave an example of HP who was trying to buy as much HBM as they can and there just isn't any more because they can't afford it or there isn't even any. So this, the responsibility of the whole HBM, the whole delivery of the GPU is all being shifted towards Nvidia now because if they want to get their rights to invest, they need to deliver the GPUs. So this partnership is really going to be mutual beneficial. And it's also, it's going to be, stimulants for NVIDIA to perform, right? And So, I think the dilution is the least of our worry, because by the time you're at the dilution, the company will already have a significant added installed capacity with the, with the contract probably in hand, and we'll start to see revenues from that moment on. So, I think this is great. It de-risks our procurement and our supply chain massively. I mean, the biggest gripe I personally have with this earnings call was that, I mean, with the earnings in general, the report as well, it's just that the AI revenue for the quarter was pretty much a disaster. And the delivery of Horizon One is, you know, sort of being pushed forward to the third quarter of this year. So a big portion of this can be attributed to supply chain and in particular for the GPUs. And I think that we are now getting a little bit of a breathing room in here with the partner, with Nvidia's partner. Because, you know, if you're not, I mean, I don't know to what degree it is of importance to NVIDIA to, you know, offload these GPUs in a priority to Iron, but it obviously will be beneficial for them. And I'm not entirely sure yet on the obligation part of this. I'm not sure if NVIDIA can just say, okay, we forfeit these options because They have no cash outlay, right? It's just an option to buy in exchange for delivery of GPUs. So, I mean, I don't think if, for example, either of these are like tradable by Nvidia, because it's a bit weird that the delivery would still need to come from Nvidia. So I'm not entirely 100% clear on it yet. I'm trying to structure my train of thoughts, but it looks to me very good in terms of GPU procurement for Iron for the coming five years. And looking back at the last 12 months, GPU deliveries and also looking forward to the coming six months are just a problem for Iron in particular, but for everyone, basically. we're going to have to see if other NeoClouds do this much better. I know there are some very ambitious guidances from other NeoClouds that do see deliveries in, you know, months earlier or quarters earlier than Iron. But so I'll just pause there. Otherwise, it will be one big rant on everything. So I see this part of the earnings as a very positive one. And now I think that we shouldn't really worry about giving Nvidia a sweetheart deal if you consider the constraint around GPUs and HBM. So yeah, that's those are my two thoughts are my two cents on this topic. What was the? Was it Childress or McKenzie that we leased the space out to Nvidia because it's the other component of the deal, the 3.4 billion over five years? I just, like I said, I can't pull up the investor presentation and wasn't able to pull it up in the video in time as I started hosting. But was your understanding that it's Childress or McKenzie Franz? I put a link in the nest where you can click on and then you can download the presentation from the banner. So it's now on the website. Let me just scroll down and see if I can find it. So A $3.4 billion, five-year AI cloud contract with NVIDIA. Let me see if I can find where that is. Well, it is going to be 2027 and it's going to be Childress by. So I say that because I thought it was, I heard Childress and that's where. I don't understand when Goldman asked Dan why he wouldn't just say we earmarked 17,500 GPUs for that 75 megawatt allocation. I mean, it's pretty clear that he has the 480 that are, or I'm sorry, because it's 2027 to your point, Franz, it's included in the 400 that are being delivered next year. But there's a good chance that it's 17,000, but with 60 with the PUE, it might be the size of one horizon. But that's speculation on my end. So I guess I can stop there. But I think it would have just done a lot of good to show the economics of it. And then I was also I'm trying to, again, super bullish long term, but there's just small things that I feel like are layups that we miss on where they asked Anthony how they're going to fund. They went to Anthony for funding and then he didn't really emphasize the fact that the Childress retrofit and maybe it's because they don't know yet. So to be fair to them, but the whole point of converting to air cooled as opposed to demoing and rebuilding greenfield liquid cool is the cost per megawatt to retrofit is likely 3 to 4 million. But even if it was five or six versus 10 to 12, if you're doing it for half the cost in a market where everything's at record pricing, you could show the better rate of return. And that just felt like a missed opportunity. But I guess that kind of covers the high level of that deal. And I think to circle things up with NVIDIA before we have some people up here, just being on a level playing field with the other NeoClouds now and not being not looking. And the fact that Dan could mention so much of our ambitions as shareholders are tied to the next generation of Vera Rubin, and it was good to hear him mention that at least on a call and how that will be incorporated into Sweetwater and then the names escaping me for the Oklahoma site begins with a K, but that was the first time I've heard him reference the name of that. But any thoughts related to that? Or do you want to talk about the European back to your homeland before you moved? to APAC or what's the next topic you want to talk about, Franz? Yeah, I'll leave that contract with Nvidia open for now. That's separate from the strategic partnership for the five gigawatt over five years, in my opinion. So and the right to invest in Iron for 5 million shares per 100,000 deployed GPUs. This is a different thing than the cloud contract. Cloud contract, all I can see for now is it's, you know, roughly $700 million a year in ARR. It's contracted today. So it is added to the contracted ARR in 2026, but it will be deployed in early 2027. So I was just reading through the notes. It is contract to GPUs as of May 6. ARR under contract includes amounts that are not yet revenue generating until the relevant GPUs are delivered, commissioned and in service. There can be no assurance that contract will result in such hours of pricing. Yeah, well, that sounds a bit weird to me, actually. If they have ARR numbers that are under contract, there can be no assures that contracted GPUs will result in search hours or pricing. I mean, what are you contracting if it's not the price, right? So, but I guess that could also mean to the upside, but I can almost not believe that Nvidia would sign a contract where the pricing is a variable upon the time of delivery. That sounds a bit odd to me, but maybe there are clauses in these contracts where they say if the HBM price goes up, then we want to pay less or something. I don't know. I mean, it just doesn't seem to make sense to me that there is something that could really alter the GPU hour price. But Anyway, I was going to talk about something else. Just give me one second so I can, what did I want to talk about next? You just mentioned something. Can you repeat what you just said to me? We haven't talked about Spain yet, the new acquisition. I think that was the other key headline that the previously, I think, You would agree that based off of, I think it was surprising we didn't hear more about Australia. I mean, it was mentioned, but then the fact that we acquired a new company that operates out of Europe and gives us access to the European continent and again, plays into the theme of a global player. I think it was just under 500 megawatts, but they're not fully constructed yet or contracted, but appears to be a talented team that they see an opportunity to get access to that continent and that market. So if you had any thoughts on that, I did bring up or looks like they dropped off, but we'll try and make some people in here too, and they might have specific topics that we can. chime in on, but did you want to touch on that or was there something else you were thinking about? Yeah, I mean, we could go through the earnings presentation first before we get other people in. Maybe that will be more structured, but that's just my opinion. I wanted to touch on the the AI revenue for the quarter. I have tried to get some kind of reply from investor relations. And basically, what they told me is that it was primarily GPU delivery related. So as we all know, they had $400 million of installed, sorry, contracted AI revenue by last earnings. That was like February 5th or something. And for some reason, the AI revenue for the actual quarter came in at $33.6 million, which was kind of disappointing as my own expectation was somewhere between 60 and $75 million. So we managed to get 50% of that, which was very disappointing. So The main reason for this turns out is it's still GPU delivery. So to this point, I think it's very relevant that we have secured this with the NVIDIA, you know, investment rights and having your NVIDIA as your biggest shareholder over the coming five years, scaling in to your company, that will be highly strategic and it will for a big portion, solve this issue that we're seeing now. So the slow ramp up in AI revenue is, again, kind of a letdown. And I think the market didn't like that. But, you know, I think if you zoom in, we are fixing that problem on the go, right? So when we are being confronted with this problem, basically, you can call it a problem, right? late deliveries of GPUs, difficult deliveries, slow AI ramp up. They are fixing it at the same time by getting NVIDIA on board, securing their supply chain for the coming five years. And they have acquired Mirantis, who is going to help them to deploy these GPUs faster, you know, commission them faster and start the revenue coming in faster. So the billing, the the whole process around, you know, selling them and with Nvidia on board and, you know, tight your Nvidia's name tied to your company is also going to get a lot easier to sell. So Mirantis will get an easier time selling irons GPUs. under their own platform with the NVIDIA label that they already had, but we now have on a group level. So I think that's worth mentioning that despite the disappointing AI revenue for the quarter, we are going, you know, we're setting ourselves up to do a lot better exponentially in the, you know, coming a couple of quarters. It will still take a little time, mainly because the Microsoft revenues are delayed. I was expecting them to come in by August, but it turns out that the cluster will only be ready by quarter three. That's a different topic, by the way. We can also touch on that. But just to say that the quarters are not really, you know, hockey sticking up like I predicted. doesn't mean that it's not going to happen, a few partners out from here. We are still on track for that, and I think by getting NVIDIA on board, Mirantis on board, I think we are setting ourselves up to do this a lot better. Yeah, it's certainly the bottleneck in the operation right now, but once things get humming, you can see the opportunity. I think today we just, the results weren't there from the prior quarter, but you see, like Franz said, that they're solving them on the go and that I expect to see some progress on that front. But maybe that was a good segue into the delay at Horizon, the Horizon one, the supercluster, expected to be handed off in Q3 as opposed to Q2 and then two through four expected to most likely be delivered in Q4. Did you want to speak about that now, Franz? You mean the Microsoft revenue? Yep.

Frans Bakker: Yeah.

₿itcoin ₿utcher: While you were speaking about that previously, you said that was a potential new topic. So was that, is this a good time for that? Or do you want me to move through some other highlights now that I have this presentation up and seeing some other things that we can touch on? No, we can touch on that. So, well, you know this because you are in my subscriber group, but we have a pretty good idea what's going on in Horizon One. there are GPUs installed, there are some GPUs burning in, and there are some racks that are empty. So they are still in the process of commissioning the first building of Horizon One, and there are three buildings in Horizon One. So I was not, you know, I was sure they would not, you know, talk about handover or anything like that. You know, it was always going to be we are in the process of commissioning them or something like that. And that's what they said. But to see that the full three buildings are expected to be delivered in Q3 is kind of a disappointment because it means the entire arrest of May and June, they will still be working on those next two buildings. And, you know, this still leaves me to believe that is this possibly another GPU delivery issue, because I know that there were some minor issues with the data centers, like first building of Horizon One was basically the worst, and building two and three are already getting better, and then Horizon Two is getting a lot better, and then Horizon Three and four are even more smooth. And So it is like a refining process, like best practices are being applied from building to building and it's getting better, it's getting more efficient. But it just it feels a bit odd to me that the ramp up is also so slow here, right? Like the shells for Horizon One were already up like somewhere in August last year or something or September, I don't know. It just, I hope that they will show that they can do this a lot better because building fast is really something that they should be, executing on and promoting on. Like that's what they should be posting about. Like this is our strong suit. But yeah, so I guess real world constraints are also a constraint for Iron. despite having that as your your main banner all the time. So I guess the the silver lining here is that they reiterated the delivery of the other three buildings in the in this year. And so, yeah, I guess we can cling on that and just fast forward to quarter one, 2027. It's all going to look a lot better. And that's also what I've been talking about, you know, if you do a billion dollars of AI revenue in the first quarter of 2027, then why are we, trying to catch pennies in front of a steamroller now, right? It is more like short-term options, for example, is really something that, yeah, as you've seen with the price action and the after hours, I mean, it was completely insane again, like from 73 to $56 in a matter of what, 10 minutes or 20 minutes. It's just, this stock is not for the faint of hearted. Let me just put it like that. So yeah, without going too much into that, I think Microsoft will be delivered in 2026. Maybe a little spillover to 2027, but I don't think that really matters. The more important thing to talk about is Horizon 5 and 6. So we could maybe move to that because it fits this topic. Yeah, so that was featured later on for their 2027 plan. And that's a continuation of the Horizon shells that it's unclear if it's related to Microsoft. Feels like that would've been an easy win for them to announce, maybe not disclose terms, but to just tell that they extended their, that they were happy with Horizon one through four. But for some reason, at least they confirmed that it is being constructed and it will be liquid cooled and most likely feature Vera Rubin. I think they connected it to that site, definitely Sweetwater. And then certainly there's the rest of the 250 megawatts gross. So to back up a stack, there's 750 at Childress. Horizon one through four is 300 gross, and then there was another 50 of Childress gross. So 350, and then the remaining 400 is the 150 that's taken up by Horizon five and six, and then the additional 250 that will be as of right now, air cooled capacity for to be built next year. So, but yeah, why don't you jump into horizon five and six and what was your reaction to that Frans? Well, so I already knew this was going to be happening and you knew that as well. I've also spoken about this a couple of times in a recent spaces, but yeah, so They are going to destroy 12 mining buildings that they built last year. So that's the let's just take the negatives first. That's roughly $25 million per building. So we're talking about $300 million of CapEx going to be, you know, steamrolled or however they're going to do it blown up, probably not blown up. They're probably going to raise them down with a with a bulldozer or something. So those are going to move and, sorry, move, they're going to be removed and they're going to put two horizon style data centers there. And if you look at the earnings presentation, it is basically, you know, right next to horizon 1 to 4. And, you know, you can ask yourself, why would they pick these buildings and why would they not, destroy some older buildings? Like why would demolish block one, for example? So I guess there are multiple reasons, but the most obvious one is it is for the same customer, right? I mean, there's a little bit of a gap between Horizon 4 and Horizon 5, but I would assume that the Horizon 4 and sorry, Horizon 5 and 6 will draw from the same primary substation as Horizon 4. So I guess that would also make it more, you know, like, are you really going to share a primary substation with another customer? So I think that this will be an extension or a new contract for Microsoft, for Vera Rubens. I don't know if they specifically said that. I have to admit I didn't really pay a lot of attention to what was said during the earnings call because I was doing three things at the same time after a few hours of sleep that didn't really work. So I have to re-listen and find a transcript somewhere. But so what they said is constructing 100 megawatt IT load of liquid cooled capacity horizon 5 to 6 in the presentation. On the page, it shows a bit weird. I don't know how that makes sense, but it says Childers plus 400 megawatts. It's page number 14, the 2027 expansion. So they expand 400 megawatt, but 100 megawatt of IT load for Horizon 5 to 6 and 250 megawatt of existing air cooled capacity. This implies that they expect a POE of 1.5. So it is, yeah, okay. So it's still the same PUE because they have 300 megawatt for 200 megawatt of IT load in Horizon 1 to 4. Okay, so nothing changes there. So the PUE will stay the same, but they didn't specifically mention the chip type, did they? Did they mention the Vera Rubin 4 Horizon 5 to 6 or did they not get into this? Maybe you know, Butcher. I'm looking at the presentation like you are and the reference to Vera Rubin is for Sweetwater. So it's possible then that. But we've talked about that, I think. And weren't you saying on three and four, Horizon three and four that the. chillers were not on the Horizon three and four. And to back up everyone, the chillers are required for GB300s, but the newer technology, Vera Rubins, don't require the same chilling technology. Correct me if I'm wrong, Franz, but I think that was kind of some circumstantial evidence to suggest three and four may possibly be used for Rubin, but not confirmed yet, but definitely we were on the we had the feeling that given the delivery time in 2027 and now with this partnership, I have a hard time believing that it wouldn't be Vera Rubin, but those are hard to access. But the fact that they mention it in the next slide at Sweetwater leads me to make that assumption right now, but I guess we could confirm that. Yeah, I don't know. It would be better if we have the transcript. I don't know if it's uploaded on Seeking Alpha yet, otherwise I can give it a quick look, but mostly this Seeking Alpha is super late with uploading it. So I will give it a quick look, but I guess it just really depends. You know, the building style in the picture of Sweetwater looks very different than Horizon 5-6. Let me just say that. doesn't really, you know, if this is for Vera Rubin, then you could sort of deduce that maybe Horizon 5-6 are still for GB300, but that doesn't make sense because this was originally designed for VR200. So what you talked about the chillers, well, no, that's just, that's really circumstantial because at the end of the day, Jonathan Cross, our chief innovation officer, denied Vera Rubin needing room temperature water, as opposed to GB 200 needing chilled water, because there was still some kind of at the wreck level or something. I don't know the technical explanation for it, but There will still be the same kind of chillers needed. But what was different here was that Horizon 5 and 6, sorry, Horizon 3 and 4, just as Horizon 2 actually, have no more roof fans. And they are more like solid buildings, like without roof fans and without the side inlets. And the style of the chilling tower or the cooling planned, how you can call it, is entirely different. So Horizon 1 is basically a bit of a, test tube baby and they immediately changed their mind after Horizon 1 to do a difference for Horizon 2. And Horizon 2 style is being applied for Horizon 3 and 4. So it's not really circumstantial evidence for Vera Rubin and they didn't touch on this topic. So I assume it's actually more likely that Horizon 3 and 4 will still be GB 300 because there are no Fiore Rubens this year for Iron most likely, unless Microsoft will change something there. But their arrangement is with Dell. I think the schedule for supplying chips for Horizon 3 and 4 will be, you know, right around this time for Horizon 3. Your 4 months, you need to, they needed to announced this substitution four months before shipment, at least four months. So, I mean, it could still be like an intra quarter announcement that Iron says in a couple of weeks or next month, like Iron and Microsoft have agreed to, you know, substitute GB 300 delivery stated for the end of 2026 for a Vera Rubin delivery by the first half of 2027 or something, or the first quarter. But I am not really getting that idea right now. I think regardless of an NVIDIA partnership, it just seems to me like GPUs are delayed for everyone. So Yeah, we've seen some like showcases at GTC and other events where people showcase NVL 72 VR 200s, but I don't think it's mass production at all. But yeah, if Iron would be able to get that early on in 2027, I think there is a small outlier that they would announce a substitution clause for three and four. But since Horizon 5 and 6 are part of the 2027 expansion. I think it would be, it would make a lot of sense if it would be for Vera Rubens. But since there is no deal yet, there's only IT load. So they are only constructing a data center according to a certain standard, certain metrics. Whatever they're going to plug in there is part of the commercial negotiations. And that's not something that they have talked about now. So It's kind of difficult to pinpoint an ARR number on that. Let me just see. Did they give, no, they haven't given a 2027 ARR guidance, right? I didn't see any. I know that the only reference point we have is I think you could assume the 3.7 that's going to be delivered by the end of this year and then the 680 for the NVIDIA deal, but nothing else has been signed. So conservatively, I can tell you that it's, you know, 4.4 approximately, even though you and I both know it's going to be higher once they contract the first. It's reference later, but Sweetwater specifically. I think it's the 200 critical IT that should be, sorry, I'm pulling this slide up. Yep, 200 megawatts of IT load for liquid cool construction. I mean, you're the, out of the two of us, you're the construction expert, but I gotta think that they could have that online by the end of next year with the experience and the ramp up that they've had. I mean, it just it all depends on horizon one through four and five and six, how those play to like if it's all sequential or if they're going to start splitting that team up, which you would have better insights on. But that I can tell you from the substitution clause research that I did. I mean, I think it's fair to say that those Vera Rubins would generate 14 to $15 million per megawatt annually. So that, you know, that would come out to roughly 3 billion in ARR right there. So it's, it's kind of like we're watching a series, you know, you're on Netflix and it's the episode where it doesn't feel like much is happening, but a lot of the pieces are kind of being laid out right now that set up to something bigger. I mean, certainly NVIDIA is a big deal and we've discussed why it's not immediately dilutive and how that's going to take place over time. And then it implies a build rate of 120,000 GPUs a year. And you made a good point of how it is almost an insurance policy for their supply chain with respect to GPUs. And now they can kind of just do what they do best and go and build. And then ultimately then they have Mirantis by their side now and they have their own people in house that can work with the OEMs and plug in sooner. So a lot of good pieces that we just haven't seen on the financial statements yet. And that's going to take some time, but if you have a long-term view and some vision, you can see what they're working towards right now and it's encouraging. I just, I am, I am very confused of what I'm seeing in this earnings presentation. Maybe it's the, maybe it's me, but I'm seeing numbers that I can't really rhyme. By the way, are you guys hearing this construction going on in the background here, because it's super annoying to me. I hope it's not reflecting through my microphone. I don't hear it, Frans, but is there a specific slide that the numbers don't jive that you want to talk through? Well, I'm just confused what they're talking about. On page 7, they are talking about 150,000 GPUs. and a $3.7 billion ARR, right? So if you then go to the notes, which is page 29, it says an ARR of 3.7 billion represents expected 1.9 billion from the Microsoft contract. Plus estimated $1.8 billion ARR from potential 74,000 GPU deployment across British Columbia and children sites. So then they say it is not fully contracted. There can be no assurance that it will be achieved. So, but if you go to page 5, you see they have $3.1 billion under contract. Of and a part of that is the NVIDIA deployments. So the actual 2026 contracted capacity is lower than $3.1 billion. So it is it's just like I just think they're making it difficult because, you know, on one hand, they are talking about what they have contracted, you know, which goes beyond 2026, and that includes Nvidia. And on the other hand, they're talking about a year end target of contracted ARR. But of that part, they have only contracted $500 million in Prince George and and the Microsoft contract, obviously. So I guess what that what that means if I just do the quick calculation. So $1.9 billion is Microsoft and $500 million is Prince George. So that is 2.4. So And I guess that means. Well, that's sorry to interrupt. That's the that's simply the new 50,000 batch, the Mackenzie and the it was 50,000 at 26,000 a year is 1.3 billion. So I think what they're saying is they were waiting to contract those until delivery, which is until second-half of. the year, but I think I agree with you. Their presentation, the ARR, I mean what you had it right. The two four is the one nine and the 500 of Prince George. So Microsoft and Prince George, I think all it's saying is Mackenzie and Childress, the 50 megawatt. is not yet contracted that they had GPUs bought for, but the 2027 NVIDIA piece is probably related to the other 250 air cool that they reference on the other slide. So I'm sorry to interrupt, but I mean, it makes sense to me, but it also, from an outsider's view, it's kind of, it could have just put two columns and just been like, here's what we guided for 2026. This is the 2027 bucket instead of presenting it cumulative. But that's for another time, I guess. Yeah, but is it safe to assume that the NVIDIA contract for 2027 is no part of the $3.7 billion to be contracted and delivered this year? Because If the, I would not, I cannot really conclude out of these, notes that they are part of that 3.7 number. If you catch my drift, it's just like. I don't think it is. I was agreeing with you. I think they have 1.3 that they have to contract the rest of the year for retrofits. I think that was a 2027 contract based off of how it's labeled. So that also means they still need to buy the GPUs for Nvidia, right? I would agree with that. Okay. I mean, good thing it's Nvidia. I'm not worried about that. I think it was just more that they Part of me thinks that they just wanted to show some sort of deal today, and maybe that was part of their negotiation process where they would give them something and show that they're doing Nvidia's internal workloads. But it's definitely something we can ask Mike Power about or any of the other guys over there. But it also means that the end of year or the exit 2026 exit ARR, the contracted ARR is going to be at least $4.3 billion, right? Yeah, 437 or whatever it is, or I was saying 44 earlier, but if I round up to 700 million plus the 37, that's where I was getting 44. earlier, which is what you're saying. You said 43, but it's 680 a year. So yes. No, it's actually seven. It's stated as $0.7 billion of ARR for the Nvidia number. So you add $700 million of ARR to 3.7, you get to 4.4. Right, I agree with that. I thought you had said 4.3 and I was saying I wasn't sure if it's because it was 3.4 / 5 years, 680 million, which rounds up to 700 million. So we agree with each other. 3.7 +.744. We're good. Yeah, so the question is why can't they contract B3 hundreds that are delivered in 2027 now, but they cannot contract their own B300s that are going to be delivered in the coming six to eight months. You know, their whole reasoning was we want to contract these when the line of sight to the delivery is more close, you know, like to close the gap between the revenue recognition and the signing of the contract. But for some reason with NVIDIA, is this their first non-Microsoft backlog? Because basically they have contracted 2027 capacity way before the GPUs are there. So I think that's maybe is that the first sign of them changing to more backlog, stylish contracts? If it was someone other than Nvidia, I would think it'd be a more relevant question. I mean, it's relevant because it it's a change in their style, but I think you would agree that after NVIDIA just agreed to move them to the front of the line for GPUs, that I think it's reasonable that they might have just locked them in to a certain rate that most likely will be way more user friendly given the constraint of GPUs. And by the way, like NVIDIA controls it anyways. So Like, what were they going to do? Hold out? And then the people that they're ultimately, maybe not indirectly through the OEM, like they were going to charge them higher as it was delivered. I think it was just part of their whole strategic arrangement personally. That's why I don't look into it as much. I think it's consistent with what they've said that Dan doesn't want to sign a piece of paper unless they have the GPUs scheduled for delivery. And it honestly, in some ways it's been frustrating because you see the, I think I saw like 99 billion or something in a tweet for Coreweave, which is sounds sexy, but then if they can't get the GPUs, like are they setting themselves up for failure? So with guidance. So I think until we resolve the delivery cadence and install these things faster like they just they have more work to do I think in the interim and certainly they got to contract the rest of McKinsey and Childress it appears the rest of the year and I'm sure they're already working on that. Yeah I'm sorry I keep hanging in this topic but because I didn't have time to prepare reading this presentation because it was only available recently. I think there is something, you know, to find out in these details because there page 11 is also talking about, you know, the ARR and maybe that can help to clarify it because here you can see the end of 2026. But this is converting capacity to contracts. Yes, okay. So yeah, okay, here we can clearly see it. But in the notes, they say the ARR of 3.7 billion represents expected 1.9 billion average annual revenue under a Microsoft contract. plus estimated 1.5 billion AR from potential 74,000 GPUs deployment across British Columbia and Childress. But in the second line, they say, based on internal company assumptions regarding GPU models, utilization and pricing. So it seems to me that GPU models are a variable. I mean, maybe it's really a minor detail, but I mean, in technical or legal terms, if they were forced to mention this. And for some reason, they, let me see, deployments, Mackenzie Childress. Okay, no, GPU models are also mentioned in different sentences where there is no Microsoft involved. So I guess GPU models are part of their disclaimer as a variable. So just happens to be three times in conjunction with the Microsoft contract. But it's, so what I'm trying to say is there is a possibility of seeing different GPU models, which could alter the ARR assumption. So that's just a minor detail. But okay, so now we are at 4.4 and you're right, I didn't disagree with you. I just made the wrong calculation myself. We could exit 2026 with $4.4 billion of contracted ARR outside of Horizon 5 and 6. So then there is still the question, when will Horizon 5 and 6 be delivered? If it's targeted to be, you know, delivered by, let's say, the second quarter of 2027. So within the first half of 2027, it's very likely that they will sign these 100 megawatt IT load in the coming three months, maybe coming four or five months. So that means you can add that ARR to the exit rate of 2026 as well. So the contracted ARR that Iron will leave 2026 with will be a lot higher than $4.4 billion if they contract Microsoft for Horizon 5 and 6, which is 100 megawatt of IT load, which could most likely be like 36,000 VR200s. I don't know if anyone can put a number on a GPU hour for a VR200, but let's assume it's Five dollars? Is that too generous? Should I go for four? I mean, I had talked about it earlier, Franz, when I did the substitution. It came out to 14 to 15 million a MW, and that was like 4 1/2 to $5. So to your point, 4-4 could look like just under 6 by the end of the year, just with Horizon 5 and 6, and then we haven't included. the 200 of IT load at Sweetwater, which could, if contracted, could be another three. So between the two of those, that's four and a half, you know, so 4.4 turns into 8.9. So there's a lot. And to everyone out there, the crunching of numbers, I know it's not the best pot material, but I guess what Bronz and I are getting at right now is There's a lot that's still down the pipe that even though they're working in the background, like you will see announcements in the next, if not month or two, definitely three to five months where we're going to get a better idea of some of these higher profile clients. Or I think that was another thing, Franz, that I, with Mirantis, it gets back to just having so much jammed into one call where they try to ask Dan during the Q&A, like for a business split, and you have this new acquisition, Mirantis, and if sovereign AI and enterprise AI that clearly command higher margins, then like it just felt like they were throwing him a softball. And Dan didn't, you know, kind of wouldn't commit to something which It just felt like an opportunity for Dan to simply say, yeah, we're excited about Mirantis and instead of selling bare metal to these hypers, we can start doing some higher margin business and we expect to do that at McKenzie and Childress. Like I think that would've been a very sufficient answer that would've played a little more favorably as they're trying to evaluate this and come out with new sell side ratings tomorrow. So and then we we did skip over the Nostrum group. I was able to find that slide. It was 490 megawatts of secured capacity, which is actually maybe you can speak to this, but my gut feeling is it's similar to. Everyone knows that Europe is kind of usually at the cutting edge of the green movement, and we've discussed at length in the past the challenges that Iran's had in expanding their capacity in Canada, where this feels like 490 megawatts is a big chunk to do some damage in the European market. Did you have any other... initial thoughts with respect to that team? Are you surprised that we're going into Europe? Yeah, I'm very surprised about that. But just to finalize the topic of Horizon 5 and 6, if you don't mind, I've calculated it to be roughly $1.4 billion of ARR for 100 megawatts of IT load for Vera Rubens. So that will be based on a very modest like 4.5 ish, 4.6 ish U.S. dollars GPU hour. That means we could exit 2026 with a AR of $5.8 billion. And that is still excluding any Sweetwater pre-contracted capacity and excluding the 250 megawatt of retrofitted capacity in Childress and that is excluding McKenzie. So sorry, excluding kind of flats. So I think just to close this topic, I think Irons has on the one hand they have to deliver on their 2026 ARR, which they have not so far. But more importantly is if we do start to deliver in the coming quarters. And then we start to sign 2027 near by capacity, like first half of 2027. I think we're going to get repriced based on those future earnings that are very near future. So I think that is significant. It's just a bit frustrating that this quarter doesn't really show it yet. in any way or shape or form. So yeah, so going to the acquisition of this Spanish company, I guess. Let me see which page it is. Or do you know the number? And I can. 19. Yeah. Nordstrom Group. Yeah. I have no idea. I just there was someone who said that he has some friend or someone who knows someone who works there or work with them. Or I can see if I can find his quote. But yeah, no, it's interesting. I mean, it's geographical diversification. It's possibly sovereign AI. You know, I don't know. I haven't. I haven't read anything yet about this because I am not really a fan of the EU in general. I think there are legitimate concerns about, you know, certifications and permits and, you know, I guess iron's done their due diligence and, you know, if this power is, you know, grid-connected power and it's energized, then maybe this can be added to their, you know, 2028 plus expansion, as it's mentioned. Other than that, 50 plus person team across development, construction and operations. Well, that's, you know, it's a pretty small team. So I guess that these people will get some iron education and then start to prepare subcontractors for development in 2028. I mean, this is all good and well, but this won't do anything for the stock for the coming 1 1/2 years because, I mean, this is far away. It's nice to add to your pipeline, but the market has never, ever given Iron any value for their pipeline, and it doesn't matter where it is. Australia or Spain. We are not even getting proper credit for our direct AI cloud contract with Nvidia. So I think IR has a lot of explaining to do to, you know, the analysts. And I know they do this. I mean, after the earnings call, it basically their work starts. So they start to call all these guys back-to-back for the whole evening. That's typically how these days go. And so I do expect some revisions from analysts. But, you know, I don't know if this Spain edition is something that will really move the needle here personally. Yeah, I think to touch on, just to circle it back to Mirantis, like if Enterprise is higher margin business and sovereign AI is higher margin business. I thought there was an opportunity to lean into that a little more because iron right now is labeled. historically, as we know, as a Bitcoin miner, but even now, you know, powered land, why aren't you guys doing co-location or you're always going to just be infrastructure as a service and that's going to get commoditized over time where I thought that's one thing that left me wanting a little more. Now, they definitely, at a minimum, they released it separately during the week so we could appreciate it more. But I thought to pull things together. I mean, he emphasized the fact that they'll be able to help with bringing compute online. But I think to me, the story is the blended margin over time and generating more margin per megawatt. And that just felt like it wasn't. And then when the analysts did ask about it, it just wasn't a really left me wanting more personally. But I think you want to bring anyone upstage at Franz, or was there anything else with the presentation that we missed? I think we covered the high level points, but in 2027, we have 3.7 right now. We're going to triple that number probably next year. conservatively, it'll be 8 or 9 billion, but more than likely it'll be 10 to 11, you know, and 3.7 by 3 is 11.1. So kind of crazy how much growth we're going to see here in the next 6 to 12 months. And if you have a long term view, this is a good place to be. Yeah. I am going to share one post in the nest that I just made a comment under the space. This is a quote from someone in only France. I have to scroll back and find his name to give him credit, but it is, I think this is something we should also briefly talk about is, you know, Communication continues to be a recurring issue. There was actually a lot of strong information shared today, but it felt crammed into an extremely tight window. Why? Investors are left with a lot of confusion and unanswered questions. Ironically, they may have released too much positive news at once, which definitely wasn't on my bingo card. Long term, this is positive. but it's frustrating that they consistently seem to make things more difficult than they need to be. They need to improve on telling, selling the story. I wholeheartedly agree with this, and I think that if you add this to the extremely clunky presentation and call they gave, I mean, you know, at the very least, agree with each other who's going to talk or answer a question. And they were like, it was almost like they were on different latencies, which is kind of ironically what they are trying to do with AI Cloud. Like, for example, Dan started to talk and then Kent fell, came in a bit later as if he didn't hear yet that Dan started to talk. You know what I mean? So obviously they were not in the same time zone, but also not on the same like latency kind of thing. So that was Really ridiculous, honestly. I mean, if you pair that with this post, I mean, yes, they will, if anyone from Iron hears this, they think, oh, we can never do it right. yes, you have the most critical investor base here on X, and we will always tell you to improve something. But I mean, come on, man. There was like four or five instances where there were more than two people talking at the same time. And that was amateuristic. Plus the fact they were way too late with the presentation, way too late with the call, and way too late with the earnings upload. But I guess the first one is because of NVIDIA. I mean, that was like a coordinated effort. But, you know, you have Dan Roberts saying, you know, the earnings call will still be happening in 20 minutes. And then everyone is there at the at the right time and then they give this little banner, we will start in 20 minutes. I mean, yes, these are all also real world constraints. I mean, come on, man. But, you know, I just wanted to say that because they did improve on some part. I mean, they switched their website hosting over to AWS and that was definitely an improvement. We didn't crash like the last three times. I mean, You got to be kidding, man. The number one thing that IR doesn't want is this to become a meme stock. And what do they do? They act like one. I mean, yeah, sorry. I just wanted to vent this out because these earnings presentations and calls are always, you know, a bit frustrating. But to the point of this post, I mean, you know, there was actually a little bit too much in there. And then they lacked a lot of details per slide. And, you know, I thought that was they could have handled that better. And now you're going to get a lot of questions. But, you know, we don't get a debriefing here as retail. You know, that's only up to the institutional investors and the sell side analysts who asked terrible questions this time. By the way, last time it was really good questions. But this time I was extremely unimpressed with the questions from the analysts, by the way, especially that no one asked about the horizon delivery or the AI revenue for the quarter. Like it was almost like it was a banned topic or something that they weren't allowed to ask. Why was your AI revenue so low? I mean, that was a bit odd, but I think they're going in the right direction, you know, It can be improved with a little bit more smoothening out of the flow and a bit more communication. But these guys are infrastructure people. I guess they just need a little bit more media training. But it's looking better. And I think we will need a bit of time to digest these earnings. you know, not only us, the market as well, of course, but the presentation is full with a lot of things, but lacking a lot of fine lines as well. So, you know, we have a great community here on X who's going to, you know, dissect the whole thing. I don't know if they've already filed, but I don't think so. Let me see. No, I haven't seen any filing yet. So, yeah, no, I think we covered most of the presentation so far, except we haven't talked about Sweetwater yet. So maybe we should talk about Sweetwater for a little bit. Why don't we vet some people waiting up, if you don't mind, Franz? I'm going to go to Edgar first, and then we'll go with Mafia, then Bitcoin AI guy, and then Sai, and then we'll touch on Sweetwater of Franz, but-- Hello. Edgar, how are you? Hi, great. Yeah, I would like to comment on the Iron Nvidia deal.

Edgar: And I would like to raise some questions and maybe know what you guys think about this. I think the Iron deal with Nvidia was a much needed sacrifice from Iron to lock the GPUs. So if the GPUs weren't the problem, Dan wouldn't sign this deal. And so he's locking the GPUs for the future hyperscaler contracts. And now Iron has everything it needed, the land, the energy, the GPUs, and also the software. But I mean, I'm afraid that this is now the standard that Some is asked for shares from NeoCloud so that they can buy their products. So I don't like to put it that way, but it looks like some sort of a bribe to me. Like there must be some obligations for NVIDIA regarding this deal. And these weren't like disclosed. This was not clear for me that what were the obligations for NVIDIA. Like here, the only way I can think of it is like Iron are paying now premium for these GPUs. So they're paying the GPU price plus they're paying the shares. And I think this should be like factored in their spreadsheet and the potential GPU pricing for the upcoming deals. So what, for example, if they want to like buy AMD GPUs, like what if some hyperscaler comes and who wants to make a deal and wants to deploy AMD chips, would AMD ask iron for share, for example? Is this the standard now? So, and regarding the 60 megawatt IT load, what are, I'm curious to know what are iron revenue here, if someone can?

Frans Bakker: Like shed some light on this.

₿itcoin ₿utcher: The NVIDIA deal you were referencing, the 60 IT load, the five years, 3.4 billion, so 680 million a year, were you referencing something else?

Edgar: Yeah, exactly. Now, everyone's interpreting the deal differently on X. Now, yeah, this is exactly what I meant.

₿itcoin ₿utcher: Yeah, so just again, for anyone who was It's in the earnings presentation. It was 3.4 billion dollars, but that's over 5 years. So 3.4 billion over 5. That's how I get to 680 million a year annually, but it's unclear the GPU count, so it's hard to price out, but my initial. thought was it most likely the 17, it's roughly the size of 75 gross megawatts, which would fill approximately 18,000 B300s. But that's speculation and we can talk to IR and try and get some more detail. But thanks for coming up. Let's go to a little mafia next. Hey, do you hear me guys? Yeah, a little echo, but do your best. And what do you got? Sorry, I'm driving. So I just wanted you to, I wanted to say that I'm very proud to be in this group because we always talk details. And I love that Dan and the team did a little zoom out and they did the... the perfect, I don't know, the perfect chess play because we needed a headline like that we are completely going to the EI world, you know. So I'm very happy. I'm, I suggest everyone to zoom out a little bit and enjoy. Last earnings were completely a mess, and right now I think we are getting repriced with the time. I'm confident that they're going to deliver. I know we have too many questions to ask, but we're going to have a lot of time until the next earnings. And I only have one question to you, friends. It's about the delivery of Horizon One on Q3, what's, I mean, we've seen the pictures, everything looks like, and even Dan posted that everything is like ready to deliver, so I don't get why, to Q3, I don't get it. Yeah, I think that's because of the combination of the infrastructure and the GPUs. because, there was a little bit of an issue with some of the liquid cooling for Horizon One, Building One, and they've started to fix that. So it is a bit of a two-sided story. But I think the, spillover into Q3 is really a combination of all the things, including the Dell installation of the supercluster, because I think Dan specifically mentioned supercluster delivery of the 50 megawatts in Q3. So, you know, it might as well just be like early July that they just connect the last things, you know, who knows, but it's a bit of a letdown. I think I was initially more disappointed in that than it should really be considered because if it's within the delivery schedule of Microsoft, it just means that the revenue will be soft in August, but it's not going to cause any penalties or clauses or cancellations or whatever. So I think it's in the grand scheme of things, I think it's a minor disappointment. So and for the rest, what you said, yes, I agree. This is going to be the end of 2026 and more of a 2027 real re-rating. But I think if they start to deliver across, the coming months, I think we will get a more fair repricing of our to be signed capacity for 2027. So, yeah, zooming out is good, but we like to be granular here. So that's what we do. But I think we completely ignored Edgar's question, Bitcoin Butcher. So I think maybe we should just quickly touch on that. First of all, I think Iron is not allowed to buy or run AMD GPUs in that 5 gigawatt that they have contracted with Nvidia. So I think that is one part of the answer, because you're talking about AMD coming and saying, oh, we want that we want shares. But Iron has a lot more power than 5 gigawatts that I see energized by the end of this decade. So that could possibly become AMD only power, right? I mean, Iron is probably still allowed to contract capacity out of that first 5 gigawatt with other chip providers. And that could be anyone, including AWS or Microsoft or, you know, TPUs from Google. So, but do they, is it like bribing them or, do you, I know what you said. I think the obligations from NVIDIA are completely unknown. So, like I said before, does it mean without the deployment, of these GPUs by Nvidia to Iron, Nvidia just forfeits the investment rights at $70. I mean, if that is the only clause on the side of Nvidia, then, you know, for example, if Iron would be like, for some reason, seeing a massive drawdown like it sometimes does, and the stock is, you know, at $40, could Nvidia say, oh, you know what, we're not going to sell you those 100k GPUs because we can send them to Nibius who will, you know, go up in share price to a degree that benefits us more. I mean, it is just completely unknown if this means that Iron has the right on their end to get these GPUs. I'm not sure if they said this in the earnings call or not. But so if that is the case, if this is also like Iron has the right to get these GPUs and then in return, Nvidia gets the right to invest in Iron or buy the shares at that price, then, you know, I think it is a fair kind of deal because Iron solves their supply chain, their problem with the chips, and that includes the HBM. So I think that Jim stated this better than I can, but basically what Iron is doing by giving them these free options to buy their stock, which could be worth over a billion dollars if you would factor in like the premium for $70 call options on Iron for 2031. That money is basically what Iron is giving away or paying to a certain degree for the HBM in the GPUs. And so they're like paying off their problem, so to say. So Yes, I agree with you that it looks on a like they are, that they need to give stock up to Nvidia to get GPUs. But yes, it's just, it's a bit more nuanced than that, but it's certainly looks that way. But, you know, Iron wants these GPUs because they see a business model in that. So it's also like, you want to play along, you got to, I mean, that's just how this industry works, I guess. I mean, and the first thing you said is, if Dan wouldn't have an issue with getting these GPUs, he would never do this. I totally agree. I think he did this to fix an issue. They want to scale faster across their platform and they just had to give up some equity for that to NVIDIA. Yes. I mean, I mean, just because everyone else has also done it doesn't really mean that it is like unfair for all the neo clouds. I just think it's the cost of doing business, I guess. Just to make sure about what you said.

Edgar: Regarding AMD now, it's now the partnership with NVIDIA, it's an exclusive partnership. Like they cannot deploy AMD on their current data centers.

₿itcoin ₿utcher: I think if I understood Dan correctly on the call, he literally said, or it was a. the CFO, but that is exclusive AMD, sorry, exclusive Nvidia within that five gigawatt. So it is framed that way. But maybe someone can find a quote. I don't have the transcripts in front of me, but I think that was what was said. So that would only pertain to the five gigawatts that is currently contracted, which is obviously including Spain, but not necessarily including Australia, because there is no capacity announced there yet. And anything they would announce there is more than the five gigawatt, obviously. So, you know, I don't know. I guess maybe the geographical location has been pinpointed between Nvidia and iron, but as far as we know, it's only within that five gigawatt. But, you know, We know of a site that is not too far from Sweetwater One that has a potential of more than a gigawatt as well. And that has not been announced yet. So I'm not going to speculate on that. But it's probably towards 2029 that they're probably going to be doing something there. And so that will fall within the five-year period where NVIDIA is working with Iron. And that could will be another chip provider in my opinion. So that that's that that's what I know from from what they said. So but isn't it like a semi concentration risk here?

Edgar: Like they obligated to like only deploy NVIDIA for the for five years because now they only have 5 gigawatt. deployment. So this is what they have right now. So we're locked with NVIDIA right now.

₿itcoin ₿utcher: Yes, but there has been no definitive word from the company during the earnings call that the currently contracted capacity will be D5 gigawatts because it's 5 gigawatts between now and 2031, right? I don't know. I closed a bunch of tabs because my computer was being slow, but I think that means that the capacity is not pinpointed to the currently contracted capacity. It does kind of match, but I don't know if it's specifically said in the call or in the presentation or in the press release that it is D5 gigawatt that Iron currently has contracted. But that doesn't really matter because, like I said, they will get more capacity and could diversify chip supplier. But I think they can already do this within the current grid connected capacity and just future capacity will be added to the. Nvidia part. I think that is a bit granular. I don't think it really matters for the bigger story. I think the company could probably answer it with, yes, we can still do AMD or another brand outside of the five gigawatt. I mean, that would solve the worry about concentration risk, I guess. Let's go to Bitcoin AI guy.

BitcoinAIGuy: Yo, I got what I wanted. which was the guidance, you know, the stock is going to 10x, right? Like that's how early we are, right? So I know Butcher was like, you know, you don't have these numbers, right? How do we get there? Well, what I was waiting on was Dan to say we could, we could construct this many gigawatts worth of data centers by this date. We kind of got that right by 2028, he put up you know, five gigawatts, right? And if we just talk about Sweetwater, the stock can get to 10x just with pretty much, you know, Sweetwater one and two, right? If you look at the pinned post in the nest, Dan, you know, one of our quants, you know, modeled this out. It's very simple, 22 billion ARR at 85% margin, we're looking at 18.7 billion in profits, you know, 500 million shares at 20X PE, that's 750 per share. Okay, well, let's use more dilution at 600 million and you're still looking at $600, $600 stocks in two to three years, right? That's just from Sweetwater, right? Not even looking at the full 5 gigs, right? So that's what I needed, right, to feel really good as a longer term hold. And it doesn't seem like two to three years is a long hold, but when you have an equity that you're holding in size, that's more volatile than, you know, some of the most volatile crypto currencies, right? It's very helpful, right? You know, Iron is like the, you know, they like to pound their chest on the optionality thing, which, you know, at first glance, if you, if you see like, okay, they basically sold their five gigawatts out to Nvidia, you know, that doesn't really sound like iron, right? But I'm still bullish, right? I think the upside is still there and it is still early, right? So if you've got a, you know, $60 cost basis, there's still a path for you to get a 10x. It's not $5 anymore, but if you believe that they can deliver and if you believe that the trade off was good, that uh you know there's now an incentive for Nvidia there's sorry there's now an extra incentive for Nvidia to deliver GPUs right outside of just getting paid for the GPUs it's the upside uh with the iron calls right uh then uh you get another piece that's de-risked so that's that's bullish um But I felt pretty good, just like small little things like, the call seemed rushed. They were talking over each other, whatever. But there was a lot of process, right? The 500 megawatt acquisition in Spain, I was not expecting that. was a complete surprise. And they just, they went by that really quick. I feel like that could have been like a week's worth of news. you know, the frustrating thing was the ARR ramp. I mean, the whole thesis, right, as a platform is you want, similar to Bitcoin mining, right? Fundamentally, why did I invest into iron? And why was it my largest position as a Bitcoin miner? Because of the velocity, right, of the hash rate, of the exit hash ramp, right? And it's the same thing with, with, AI, the velocity of ARR. If this is going to be the machine that produces the most amount of cash flow per unit of time, then this is going to be $1,000 stock in a few years, right? Like, that's what I want. But it's hard to sell that story if they're, you know, dropping these balls in the short term. So you're going to have pushback from the Nubias and Coralweave crowd, right, naturally. But, you know, there's still a 10x upside within two, three years, right? So, yeah, they're, you know, tactically, they're, you know, they drop the ball. It's okay. You know, NBA players drop the ball too sometimes, but over a long period of time, you know, they win, right? Like they're professionals. So that's the bet. And I feel really good. You know, I think the thesis is improving and they're getting better strategically. I mean, you get Nvidia aligned with Iron, I mean, largest company in the world, right? That helps. You have the software ability, you know, that shut up a lot of the you know, the Nibius Bulls that said Iron is just a construction company. Well, now you've got the software layer that can potentially, you know, juice up the margins by an extra 4x. That's good. So I feel really good. You know, it's hard to be bearish, right? Like I'm thinking of some other bear cases, but it's hard. I think Iron, their expertise in energy infrastructure, data center infrastructure is going to pay off. And it is a real edge. And I think with this added incentive with Nvidia, it's another differentiator against not just the big neo-clouds like Nvidia, sorry, Mebius and Corweave, but like the smaller neo-clouds who As a category, they're all raising billions of dollars to buy GPUs from Nvidia, right? But now you have Iron that might have a preferred allocation because of this extra incentive. They're now a strategic partner, right? I think they're taking the right strategic steps that smaller players, right? that don't have the power, that don't have the capital, and don't have the same sway to make those kind of deals with Nvidia, you know, they're going to beat those guys because they're going to have the time to compute advantage over the smaller guys. So yeah, I mean, I think it's a good move.

₿itcoin ₿utcher: Thanks for coming up. Let's go to Sai what's next. Then we'll go Nick, then Neil, then John, then Brian.

Edgar: Hey guys, thank you. And France, I feel sorry, man. I don't know if you're getting enough sleep with all these things, but yeah. So this day is quite a bit of confusion. First of all, I was busy with my schedule, so back and forth, I'm refreshing Twitter to see what's happening. But I got a few questions for you. So the deal with Nvidia, it's if I'm not wrong, is it? This is for 600 K GPUs, right? And I thought it was. is for GB300s or are there Rubens in it? Or they did not mention anything about what kind of GPs we're going to get.

₿itcoin ₿utcher: He might have stepped away for a second, but I don't think there was any distinction in the GPU types.

Edgar: OK. And are we going to use part of that for the current Microsoft deal? Or is it already baked in and we don't have any issue procuring those GPUs? I know that Dan has said that they're receiving the GPUs, but do you see any part of that 600KB used now for Microsoft or it's not?

₿itcoin ₿utcher: And they were already receiving delivery from Dell. So we could, I, I can't say it with certainty, but I would, I personally do not believe we could confirm that with IR, but I think that 76,000 was already in, you know, in transit or on the even if it's delayed, it's still accounted for where this is more the next five years, potentially, you know, a five year cadence at 120,000 GPUs a year that now, as Jim's pointed out, essentially the calls that were calls unit RSUs, whatever you want to call them, the $70 strike for 2031 that kind of blocks in the HBM for the high bandwidth memory for those GPUs from NVIDIA.

Edgar: Okay. And so it appears at least from the revenue we have seen booked so far, like it's Prince George and all of that. The rest of the 75K of the 150 is really having the issue, right? And considering the pace at which we are, do do you think we're gonna we can procure the GPUs if not in time at least here and there plus or minus a month here and there results 75k.

₿itcoin ₿utcher: Yeah I mean it they they're intense to deliver the they didn't revise guidance so I think they believe that this will all be resolved by Q4. And just to be clear, like we're not saying that they're going to generate 3.7 billion by the end of the year as much as the expectation is all of those GPUs associated with the, there's 150,000 GPUs. The understanding is by December 31st that all 150,000 will have already been burned in and generating revenue. So on that last day of the year, it should all be in place so that in 2027, we could expect a full year's revenue of 3.7 billion.

Edgar: Yeah. At this point, I think it's hard to hit that revenue by end of this year. So that's completely out of the window is what I'm guessing. I'll be happy if we have all the GPUs inside the facilities. up and running and probably will have a fresh good start or I mean a good revenue recognition schedule starting next year. OK, so outside of that, do you? So why is do you think that I haven't heard because I'm not following obviously the NBS and all the other other new clouds as keenly as I am following iron. So why is this not an issue for other new clouds? And it's only an Iron's issue that we are not able to get the GPUs in time because they already have the partnerships with NVIDIA in place. And it's not an issue because of that reason.

₿itcoin ₿utcher: I'll take this one, Butcher. Do you have no definitive proof that other Neoclouds are receiving Blackwells at a faster pace than Iron? The reason why other NeoClouds have higher revenue numbers and growing faster in their revenue quarter over quarter is because all of them are still processing all the hoppers and all the early buys of GPUs that they did well before Iron became. So yes, it is true that they are ramping up faster. That's also why you're seeing better numbers presented at Nebius, for example, who has not that much more active power. I think by August we will have more active generating AI, AI revenue generating megawatts online than they have. But, you know, the the reason why we are like delayed on this side by comparison is not because they get, you know, from a moment of order to delivery, they get their GPUs that much faster. It is also because they already had more GPUs. But on the flip side, it is true that Iron now has a more preferred status than they had before. So it might very well be true that, this will resolve in the coming, months leading up to, the end of the year. So I think other Neo clouds may have relatively less of an issue with that. But I think the problem with other Neo Claus is they're going to run out of capacity in the coming 18 months. So I think that maybe Iron loses the 2026 race, but we will certainly have good chances for the 2027 race, so to say. And I understand your sentiments around your worries about getting all that contracted revenue installed and, you know, generating revenue this year. I am a little bit skeptic about Horizon 3 and 4 being really generating revenue in December. Other than that, I think the B3 hundreds should all be there and the retrofit is not that hard. And I think now with Mirantis, I think they will be able to, get them up and running for revenue and billing faster than what they could before. So it we will, we will, I guess we'll find out.

Edgar: Yeah, OK. And I. Yeah, so, okay, so one last question before I give back the stage. So with this currently, this NVIDIA deal, you guys feel that we'll get a preferential treatment much above than Coreweaf and ABS or we think, I mean, we will be given the GPOs after they enabled on what are the fresh Rubin's batch that NVIDIA has to offer them. So did we upgrade the status above those two?

₿itcoin ₿utcher: I think we're on a level playing field. I mean, they have equity. NVIDIA has equity interest in both respective companies. I think where our advantage is potentially as the capacity piece of it, like if they they have to deal with a third party in their co-location provider and their co-location provider has to deliver their capacity on time. So it does potentially streamline the process, but no one's going to know except Jensen how he's allocating it. But I my gut feeling is that it's a level playing field or at least they, they improve their spot in line. Maybe it's instead of a distant third, it's a little closer, but I think it's very encouraging news. And I think to AI guys point, like we saw that they expect in the next year to this year it was 400 some megawatts, 480, whatever the number was, but next year 700 and they'll get up to a thousand a year and the HBM bottleneck will, the foundries will resolve themselves and adjust. And I think right now it's just getting really good at this time to compute thing. Like there's no doubt that it's important, but I think anyone who's an honest shareholder will tell you that we're going through some growing pains right now, but there's a lot of opportunity long term and if you stay for the ride and let this team figure it out that you'll be rewarded down the road. So appreciate it, Sai. Let's go to, we're going to do Nick was next, then John, then Brian, then Jeffrey. Nick, what's up?

Nick: Hey, how are you guys? I really liked the earnings call. I think it was very bullish. In regards to the NVIDIA strategic partnership, It's true, it sounds like an engagement and it is an engagement, but you're engaging with a number one GPU provider. So what NVIDIA is doing is creating an ecosystem and trying to make that ecosystem bigger. And this is why, in a way, I have said before that Ireland could get engaged to Nvidia whenever they want it. And this has demonstrated that. But it is good. It's a win-win for everyone. It's a win-win for Nvidia because it gets to grow its ecosystem, but it's also a win-win for Ireland for obvious reasons. The main one is that they get access to HBM. And so Sky Hynix produces 70% of all Nvidia's HBM. And Nvidia has placed orders for HBM like a long, time ago. So now everyone else, like Google or Amazon, are trying to place orders for HBM. But that capacity is already contracted by Nvidia. So, you know, those guys are not going to be getting what they want for sure. So, you know, there's there's no other way to get the GPUs we need unless we, you know, break a few eggs and so to speak, marry Nvidia, which is super good because you're marrying the prettiest, the prettiest girl in the party, right? Then the other thing with the NVIDIA partnership is that it's a design of that data center that optimizes the Vera Rubin platform. So basically what you're doing is like you are doing the best practices for implementing the Vera Rubin. And so what you're bringing in is you're bringing in an NVIDIA design for your data center, which is great because that's a stack that needs comprised of compute, of networking, of storage, of power, of cooling, and the most important part of it is software. So we're going to be having NVIDIA software in these data centers. And so this is going to clearly maximize the tokens per watt. So again, this is a win-win for NVIDIA and for IREN. And not only that, the deployment of this NVIDIA DSX architecture the flagship deployment of this architecture is going to be Sweetwater. So for sure, media is going to, be doing everything it's in its hands to make Sweetwater a success. And as we all know, Sweetwater is the jewel of the crown. So we really want that to be, you know, to be good. And this, in a way, it's working into that direction. In regards to the Spanish company acquisition, Nostrum, I think it's extremely bullish as well. I am an euro poor in origin, as many of you. And I know a little bit about how things work over there. And first of all, you're increasing your, by 10%, your secure power, which by itself is such a good thing. I mean, increasing 10% when we're talking about, better Rubens, that's huge. And also, I forgot to mention before that these, so Nvidia, they're going to be providing 100,000 better Rubin GPUs for the 5 million shares. And that, but that's the equivalent as 400,000 GB300s of today. So we're talking about a lot of compute here. if we consider that we're using the immediate architecture, the media software, this is totally maximized for token per watt. And so this is going to be big. And going back to the Spanish acquisition, I think it's terribly bullish because in Europe, and we all know how Europe is, they don't want American companies developing data centers. They want to feel that they own, so to speak, their data center. That's the mentality over there. And so governments, they want companies that are, so to speak, considered as not American. So in no way they were going to facilitate any hyperscaler going to Europe to build up the data centers for Europe because they might even see it as a matter of national security. So this opens up a huge door for Iran And not only it opens, it opens it up in reality, but also when you wanted to compare it to other neo-clouds, now we're all playing the same game. So we're not talking anymore about a company with a presence in Texas and a little bit in Canada. We're talking about a company that has opened the door to the European market with which is the EU. So it comprises all of the countries, not just Spain. So you're opening the door for the whole, so to speak, continent. And also, my take is that prices might be good there because things there, especially when the governments are interested in them, Many times they have high subsidies. So I am hoping that when they develop the data centers there, they might get some good pricing as well. The last thing I wanted to talk about was, and the only little thing that I didn't really like about the presentation was when they were asked about the commercial mix. I don't think they were able to offer like a good answer to that. It seemed to me they were flip flopping a little bit and I.

₿itcoin ₿utcher: Nick, I'm sorry to interrupt, but I I totally 100% agree with you on that. Like sovereign AI and enterprise AI are such a layup to say that your blended margin improves and I thought. they missed the, especially with Mirantis. So just like we paid 600 million in stock for a company that enhances our ability to generate. If their average right now after the NVIDIA deal is 11 million a megawatt, what if they can do 12 million a megawatt with Mirantis on the B3 hundreds, that extra 10% or what does that look like? And I just I thought part of the problem, as you had said earlier, is there were so many things going on in that presentation that maybe the. And it, you know, I know they have a tough job and some of it might have got finished last minute this week, which it appears Nvidia did. At least that's how the deal was released. And that was obviously the headline. Maybe it's something they didn't think they could get across the finish line than they did and they tried to include it. But I think what some of our competitors do better is they have rolling communications that kind of tell the story in real time where since November of last year, a lot of times we're waiting until earnings. And I think management and IR have done a better job more recently trying to move things along and with their tweets. So I have to acknowledge that there's been progress, but I think it's still an area for improvement. But I want to go to Brian now, Nick, if that's all right. Thanks for sharing with us. Brian, how are you? Sorry, but I think John was first. I'm sorry to take over, but I want to hear you. Brian, I'm sorry. Can we go to John Austin first? My mistake. John, how are you?

Frans Bakker: Hi, guys.

₿itcoin ₿utcher: Thanks for helping me, friends. Always good to hear from you, mate.

Frans Bakker: I guess two things kind of stood out to me. Number one is inadvertently, you know, they just gave us a pipeline upgrade, obviously the 480, 490, whatever it was coming out of Europe.

₿itcoin ₿utcher: I guess the other one that was.

Frans Bakker: Kind of interesting is, you know, alluded to multi gigawatts coming out of Australia. So you think about that five gigawatts that they've got total?

₿itcoin ₿utcher: I mean, in effect, they've actually just said that it's.

Frans Bakker: You know, six or seven gigawatts by virtue of that because the current numbers add up to five.

Speaker 6: And I guess that kind of goes to concentration risk as well. I mean, I guess how many other. sites are we really talking about an Nvidia you know five gigawatt which might actually only be 60% of the active portfolio here so then you you got an Nvidia.

Frans Bakker: Concentration far lower than you know what what the headline might believe that they've effectively got you know.

₿itcoin ₿utcher: Which is how it reads is they've got the whole portfolio.

Frans Bakker: Tied up I don't think that's necessarily correct probably the other one and we kind of go back to.

Speaker 6: Thinking about what deliveries, I guess, where IRA might sit for Nvidia's delivery cycles.

Frans Bakker: I mean, selfishly speaking, surely Nvidia are going to.

Speaker 6: Prioritize their own Childress GPUs purely for the fact that it's used for internal consumption.

Frans Bakker: So I would imagine that's part of the reason why they've actually done.

Speaker 6: It at Childress is because it gives them the ability to bring forward some GPU deliveries there.

Frans Bakker: And shore up that revenue that might be otherwise slipping because of the Microsoft stuff, albeit I don't read too much into that. I mean, the way that these guys would have, I thought.

Speaker 6: Communicate is, you know, Q3, like you said, Frans, could be, you know.

Frans Bakker: Could be July 1, could be July 10. So, yeah, we're talking about a month here in the overall context of a five-year contract. I guess one other point there is, let's just zoom out of Childress for a second. And maybe this is a precursor to how they're thinking about Sweetwater, for instance, obviously significantly larger scale. But effectively since what, November last year, they have two of the four or four or five, and many you want to classify it as, hyperscalers sitting in the same site with flexible GPU deployments.

Speaker 6: I mean.

Frans Bakker: It's pretty phenomenal, really, to have a site that's got air cooled next to liquid cooled, Microsoft next to it, and ultimately a pretty amazing in the space of, you know, what is, call it eight months. So, I mean, I don't think too much into what we've got at the moment.

Speaker 6: Obviously, there was a few misses in that presentation, a lot in that presentation as well.

₿itcoin ₿utcher: I probably agree with you on some of the messaging.

Speaker 6: But again, I think for the people that have been around for a while, you've seen them kind of go from zero to 100 in Bitcoin mining. You've seen them go from, call it zero to 10 in AI. And I think the 10 to 100 is going to be interesting for us with a whole bunch of.

₿itcoin ₿utcher: New nuggets to come out as we go.

Edgar: So I don't know.

Nick: I mean, overall.

₿itcoin ₿utcher: I think everyone just continues to hold and enjoy the ride, quite frankly.

Speaker 6: At the end of the day, we've got a business that's going to be pumping out 10s of billions of.

Frans Bakker: Dollars worth of revenue over what time frame, if we miss a month or two, who gives a ****?

₿itcoin ₿utcher: Well said, John. Thanks for coming up. Sorry, Franz, I interrupted. No, I just wanted to agree with that. And yeah, thanks for your contribution, John. Good to hear from you. And yeah, refreshing thought on that pipeline for sure. Cool.

Edgar: Thanks, guys.

Nick: Cheers.

₿itcoin ₿utcher: Thank you, John. Brian, thanks for being patient. How are you?

Speaker 6: Yeah, doing well. Thanks as always for putting these on. I agree with John and I remain very bullish on IRIN. I mean, in my humble opinion, you know, equities are anticipatory assets. And so they're going to be priced on the credibility of earnings and execution. And if they want to earn a higher multiple, you know, IRIN needs to deliver on both and tell the story clearly enough for the market to underwrite it. And the asset base is clearly getting more valuable by the quarter. And there's likely, I would argue, a put on the downside, given the power portfolio and the contracted ARR. So if management can't communicate the platform thesis with the rigor the asset base now warrants, the most likely outcome is potentially an M&A bid from a larger operator who can execute across all the fronts and pull forward the value. And so I just continue to believe, and I've made this clear to management, that the IR gap just can no longer be ignored, especially if the company and management want to high grade the shareholder base you know, with the long only type investors they want, lower the volatility and improve their cost of capital. So that's all I wanted to share with you guys. Continue to be a long term large holder. And thanks for what you guys do.

₿itcoin ₿utcher: Well said, Brian. It's I do think tonight, though, I think maybe let me ask you this question before we go to Jeffrey. Do you think tonight getting the seal of approval from Nvidia and having a formal partnership arrangement might bring a little more institutional interest. And also the fact that their Nvidia is allowing Iron to provide compute on their behalf. Does that give you any does that change your perception at all or do you still how would you react to that?

Speaker 6: Yeah, no, I'm I'm a huge supporter and I think that having the the largest or second largest market cap company in the planet, you know, do a partnership and release a press release is, you know, wildly validating. But, you know, consistently not being able to release, you know, earnings, press releases to e-mail distribution lists on time, release, you know, Q's and K's and have consistent issues with messaging, as Franz alluded to earlier, does degrade credibility. I'm sure management is on the phone with institutional shareholders tonight, and we'll see how the market reacts tomorrow. Again, I'm a very long-term shareholder and believer in management and think that they're doing the right things. That being said, you know, if you want to attract the right long-term shareholders that need to see, you know, a stock that doesn't trade at 100 vol, with the beta that they trade at, you know, you got to just high grade all component pieces of the company. And I think our IR continues to be a major problem for them. And that's a key component of managing the street, both the sell side, where we know that sell side analysts are completely mis-modeling the stock. Their price targets are too low. And therefore, the buy side has not caught up yet. That's an opportunity for us, of course, and I've certainly taken advantage of that. I'm sure other people have. But if these guys want to become a core weave at that kind of market cap, they're going to need to improve that function.

₿itcoin ₿utcher: If I may, I think IR is a Not the cause. I think I understand what you're saying, but I think the problem here is that management is keeping a too tight lid on IR. I think that IR is understaffed and, you know, overburdened with workloads. So they are, and they're too overcoverned. So I think that IR needs to be, you know, operating more as an, I won't say independent part of, but they need a bit more freedom to act. And I think that is part of it and they need to be upgraded. I mean, there needs to be more capacity in teams in terms of workforce going to IR. But they have recently made a hire, which I think is sort of like between IR and the CFO side of things. And it could be that there will be improvement there. But seeing this earnings call and presentation and the delivery of everything today, I think that the improvement is yet to be to be shown. But yeah, I agree that there needs to happen somewhere there. And the final point I'll say is I have someone from a buy side in my subscriber group and he He just agrees with this entirely. He sees other companies have their CEOs tour around the country and talk to everyone they can and showcase and demonstrate everything and convince the buy side in person. And then there is Ireland who has their CEOs on the other side of the world and a different approach, so to say. And I think they can improve a lot on this. Yeah, I hope they will take this serious going forward. Jeffrey, you're up next. How are you this evening?

Frans Bakker: Okay, I'm going to kind of go to the other side right now with comments, and I'm going to say that I thought that the earnings report was ******* awful. And I felt that, I mean, these are smart guys. They had a lot of information. I think they provided no color and they gave the bare minimum of information. It's the, by fact that everybody here is sitting here like you have a Rubik's Cube and you're trying to figure out what the ****** going on. And I think personally, like they had some very big events just happen. I mean, one, they energized Sweetwater. Two, they picked up Mirantis, which is a very unusual acquisition. And Mirantis being one of the three independent software vendors in NVIDIA's program, which caters to optimizing their ecosystem and their chip usage, and also that Mirantis sort of embraces that sovereign AI structure. I was expecting something else to follow, and I think they were missing a piece or two of the puzzle that they wanted to present during the earnings report. And so I feel like that they were almost intentionally not going into any embellishments, not providing any extra color or any detail on things because there's a gap. And if they start talking in that direction, it starts kind of letting the, you know, the kind of the worm out of the can too early.

₿itcoin ₿utcher: Can you give an example of what gap you're talking about? Is it financing or?

Frans Bakker: No, I think probably it has to do with, I mean, I mean, I think probably the biggest, one of the things that Mirantis kind of presents as a solution is the sovereign AI model that they can provide. And so I don't know, maybe they had a client. I mean, they kind of have created a picture of themselves as like, you know, here we are, we're kind of a NeoCloud provider. We have access to the video chips. We have access to power. We have sites. And, you know, and now we also have Mirantis that can kind of embellish and kind of elaborate on that whole picture to provide you also a sovereign AI infrastructure. And so if I was a client out there, that would be potentially really attractive to me. And then they were even alluding to the fact that we don't need to advertise like NVIDIA will help us advertise, right? And so maybe there was a part of this that's not completed in the time constraint before the earnings report. where there was some loop or connection, you know, I mean, I mean, it's like random, but like, you know, there was like that thing that was circulating around that BlackRock was supposed to announce that they were going to have announced who they were doing a data center with. And like, you know, if you're someone like BlackRock, do you want to have a sovereign kind of, you know, kind of a cloud AI structure, or do you want to have one that's like shared? I'm the way same way, like maybe a country would want something from a security standpoint. I mean, I don't know.

₿itcoin ₿utcher: Jeffrey, I'll sorry to interrupt. I was going to jump in if you don't mind, like I will get speculative with you. Like we saw the $25 billion investment that's going to occur in Australia by Microsoft. We have a good relationship with Microsoft. You see the government over there trying to get more involved and embrace AI like that feels like a natural fit that you're does that maybe explain the gap that you're suggesting? Maybe not that deal in mind that you had, but you're kind of I think the way I listened to what you said was almost that they're working on things in the background that aren't far enough along that don't have pen to paper and they don't want to tip off. So maybe that's why you're suggesting they kept things vague.

Frans Bakker: Yeah, exactly. It definitely felt like to me, like these guys ran that earning report. They were short. They didn't want to, even when like the people were asking a few questions at the end, I mean, you can always want to ask you a question, use that as a segue to sort of like run and talk about other things. I mean, it's like they wanted to get off the call. I mean, I was surprised. I was like, boom, it was over. I was like, that's it? You know? And there's like all this, this ton of stuff out there with like no elaboration on it. And so it was, I thought it was like very bizarre. And I think you can either kind of blame it on like they don't know how to handle that situation, but I don't think it's just as simple as that. I think they wanted to get off because they have some unfinished pieces and they really didn't want to spend a lot of time talking about it because they don't want to, sort of, you know, spill the can of beans like about what's working and brewing. So I would be in that way, very bullish that maybe in very short term, we're going to hear another big piece of news. So I think that's just the impression I got. I mean, I felt like how that earnings report ran. And then we even talked about that the other night. Like, you know, they've been spending a lot of time on this. That didn't sound like a earnings report that they spent a lot of time putting together to really present. You know, I think it was I think they were hoping that certain ducks were going to line up at that point so that they could, you know, blow it out, blow it out the door at that moment. They were missing a piece. I don't know, maybe I'm wrong, but that's just like, you know, that's that's just what I walked away from from that. And I felt very hollow after that. And I'm I'm kind of very negative on how the market's going to interpret that. And because they left so many holes open, all like, you know, the crazy short people are going to take advantage of this and they're going to start just drilling away at the loose structure, the loose organization, the loose guidance, you know, and direction that there is, you know, and the failure and like the numbers. I mean, you know, you're going to come out and you're going to dump ****** numbers like that. You got to have like a great narrative to follow with it. That's sort of like kind of kind of, you know, kind of hides that and shadows that and makes that seem less significant. And I had a lot of pieces of information and I think that there was a reason they didn't kind of use it, I think, more aggressively in that whole discussion.

₿itcoin ₿utcher: Well, thanks for coming up, Jeffrey. I think where I'm, I was to Fran's subscriber group, I almost got the impression that I know those guys were up late and worked really hard to put this together. And again, I think part of it is specifically the NVIDIA piece appears to have been finished in the last week or two, where it just feels like every time they get to the earnings presentation, that they're so out of gas that by the time they get on the call, there's just, they look tired. And then on top of that, you have a live stream where everyone's on camera, So you can't even, I mean, there's a lot to be said about tonality and inflection of voice, but just looking at someone's body language at that call where it just looks like, and I don't want to overanalyze it, but it might, I sometimes wonder if, is there really value add by them being on camera where I think half of the **** we wouldn't even be talking about if they were just doing a call like most other companies, which again, it's, I mean, so much good stuff in this call. This is a 10, 20 X company, all that good stuff. But part of getting to that 10 or 20 X is telling your story. And I'm in a business where I have to sell to people individually each day. They have to sell to institutions. And right now it just, I'm sure they're on the phone doing their stuff, but sometimes it just feels like their mind is somewhere else like that. They, was it another deal or I can't tell you. I just, but watching that, it was just kind of strange. And I wonder what their, why they insist on doing that still, because there's just been a few times where then people's eyesight aren't on the screen or they're not looking directly into the camera and just weird stuff that is all too visible when you're on camera. But thanks for coming up. And Franz, do you have any other thoughts? I'm kind of pushing 11:30 here on the East Coast. I'm happy to hand it off to you, but I'm probably going to get off here soon. Well, I already said what I wanted to say about the style of the earnings call, et cetera. I won't say it was awful. I think it was a very long earnings presentation. It was more than 10 pages. It was like 10 or 11 pages more than the one from the first quarter. Sorry, the quarter, the November presentation. I couldn't find the February one just now, but I mean, there's just a lot to go over. And another thing was that Kent didn't speak only during the questions. So it was only him and Anthony, Dan and Anthony talking. And yeah, they seem to be beat. Like the first thing I said, why is Dan looking so sad? You know, it was his own birthday and he looked like, you know, he came from a funeral. He looked extremely sad. So maybe he's just very tired, but there are just a couple of obvious things that weren't talked about, like the big miss on revenue, the later delivery of Microsoft, and, the Horizon 5 and 6 without the clients. Hello, maybe that's the thing they are finalizing. Maybe the substitution clause is something they are finalizing. There are a bunch of things that are in the process of being done, right? So, I mean, I am sure that they'd rather be doing what they do than having an earnings call. So, you know, there are a lot of things you could say that could probably influence this and they're probably not natural actors in front of a camera. Like I said, they need media training. I think they're they're working on that. I mean, and it will always probably be a bit more clunky than some other companies, but so be it. I don't think it was awful. I think we are grossly underestimating the strategic advantage of having NVIDIA on board. Yes, it feels a bit of we are joining the other Neo clouds and we are falling in line. I think that Iron wanted to be the outsider, the underdog and the contrarian, and it just didn't work because the power of Jensen and the reach of his GPU deliveries is just too strong. And I think that the customers just want NVIDIA. So you can talk about the concentration risk, but at the same token, they have the best GPUs and, they are miles ahead of competitors. I mean, there are some theories that Vera Rubin is now being rivaled by the newest instance of AMD. But, you know, at the end of the day, it is a whole ecosystem. that you get into. You see Nvidia is now with their factory AI factory reference architecture. They're pulling in vendors all across the infrastructure layer. It's becoming very hard to escape Nvidia. And you know, if Nebius and Coreweave have priority status because they took the bribe for the equity, then Iron gave up call options on their stock in return for, priority status. So they have leveled the playing field. We can now put this behind us and say, oh, Nibius will have the GPUs earlier. And we have acquired a company that has a certain status with Nvidia that Nibius doesn't have. So I think that's also kind of cool. So, I think when the dust settles over the coming week, we will probably, understand this thing a lot better. We will probably be able to fill in a few gaps ourselves. I have a pretty good network of people all over the place that hear a thing or two. So I hope we can model this. But for me, the strongest thing going forward is Yes, we have attracted some dilution over the coming five years, but don't forget it's just 5 million shares per year. I mean, it's not ideal, but we are trading that in exchange for a solid supply of GPUs that we can turn into money generating megawatts. So I just made a comment to someone where I explained the whole thing. I think Iron will be able to sell capacity in the future. Now they know that they are getting the GPUs because now they're aligned with Nvidia is direct and they are, you know, a preferred delivery partner. I'll just, I'm not talking about the 60 megawatts of IT load for Nvidia themselves. I'm talking about the 120,000 GPUs every year that they trade for those warrants. Iron has never really done any backlogs, and the ones that they did do always delayed, you know, like Microsoft is arriving later than we thought. We're seeing these be, sorry, these 23,000 GPUs for Prince George, some of them were bought in the summer of last year. And now we're getting a quarter with $33 million of revenue, you know. I think they want to narrow this gap between ordering the GPUs and contracting them and then generating the revenue. It's all going to get closer now that they have NVIDIA on port. And I think we underestimate that because you can not only can you model your sales around getting the GPUs, you can also model your data center construction around this. Because if you know that in 2028, we're going to get 120,000 of Vera Rubin Ultra. So we have to build 300 megawatt of IT load of Vera Rubin Ultra data centers. And then in 2027, you can start to sell it. and then you can finance your contract and then you pay after the delivery of the GPUs. So you're starting to see, I mean, I saw someone who had some kind of ******** post that this means super much dilution for Iron because Iron will need to pay all the GPUs still. Iron has not had any instance in the last 12 months where they had to pay in advance for a GPU. So These GPU financing flywheel is already ongoing. I mean, we had $3.6 billion of GPU financing for the Microsoft GPUs. And I think they alluded to these 50,000 GPUs going to see GPU financing as well. And I think that that's also why they're issuing equity left and right, because they want to keep this equity and depth ratio. So to the people that are saying that, oh, Iron has welcomed more and more dilution on top of the $6 billion ATM, first of all, it's drawn out over five years. And second of all, you know, we don't have to tap the existing ATM to pay for GPUs that require a deferred payment schedule. That's just not how it works. And if we are really going to judge Iron on their $6 billion ATM. They have issued $380 million, if I'm correctly, since what, February or January? When was it that they announced this? I don't even know, but I put a picture in a nest where you can see my predictions from Wednesday. And if you just quickly go down to #8, I predicted fully funded for 2026 or no more financing needed for current plans. This implies no further ATM usage, plus ATM only used for acquisition and less than $500 million drawn otherwise. Well, I can't watch look into the future, but I think I was pretty on spot with that point in particular because we are fully funded and we have not tapped for more than $500 million outside of acquisition, which some people say can and some people say cannot be used for the acquisition, but they have not generated or, they have not sold more than $380 million in shares from the ATM. So I think there are a lot of bears that want to find a bear case in this, but at the end of the day, you need your GPUs if you want to be an AI cloud. And yes, we have to give up some equity over the coming five years, but we get something in exchange for that. And what we get is what we can sell. So prepayments, not even mentioned yet, but if you add prepayments to that, you're going to see that this just needs a couple more quarters to start rolling. And I think we got a little preview of Iron's real capabilities in terms of supply chain. in terms of construction, delivery and planning. And now we have basically all the ducks in a row. And now it's just up to them to get a few more deals out of this. I think we're going to see a Horizon 5 and 6 deal in the coming couple of months. That would make the most sense to me, honestly, in terms of contracting it. Because they're going to start demolishing these mining buildings soon anyway, Would you really do that without having the customer sign? I mean, that's, there's some, I think that's unlikely. So I think we're going to get a nice upgrade in ARR going out of 2026. And I would personally not be surprised if we will actually exceed 6 or $7 billion ARR contracted by the end of this year. That means not installed yet, but under contract because if you include Horizon 5 and 6, you are already close to $6 billion. And they're going to make plans for, you know, a kind of flats and the air cooled capacity in Childress that hasn't been bought GPUs for yet. But so long story short, yes, we are not looking at a great quarter in terms of P&L. That was a disaster. I agree. But there is a lot of, you know, substance to build on going forward. I think we have a solid base to build on. And the puzzle pieces are coming together for me personally. Yes, it's going to take a little bit longer, but I mean, I think we're in a good spot and I hope the market will see the same. So that's what I will say as my final words. Yeah. Thanks for co-hosting, Franz. One other piece of hopium, because I think people might push back on us that we were too negative. We just we care so much about this company because one, it affects us personally, but also just we know what it's capable of over time. And so I was thinking about this week how people were asking, why would Mirantis take $600 million? You know, is that a red flag? Well, is it a red flag for Iran to take or give up these? you know, options to Nvidia and I think they actually run parallel. It's kind of the same situation. Mirantis needed iron capacity and we need Nvidia GPUs. And now that we don't have to worry about GPUs and Mirantis doesn't have to worry about capacity, think of the productivity just by getting rid of that concern and just the unleashing your workforce and letting them actually go attack and not have any concerns and actually just be able to build. And when they go into a room and are trying to commit to a timeline, they can have a little more confidence than they've had this past year. So I will close it off here and push in midnight in Detroit and I'm going to try and get some sleep here. But I appreciate everyone attending. We throw these because everyone, this is a community and Franz and I appreciate all of you and it's fun to have some people to go along for the ride. So till next time my friends and things are looking up. I think it's just gonna take hopefully Mike Powers. kicking some *** right now and Dan's on a few calls and these guys are communicating the story that we were trying to sift through, but lots to digest from the sell side analysts. But I would expect a few upgrades here in the next day or two and hopefully that flows through to share price. So we'll talk soon. Good night, everyone.