$iren Spaces - Memorial Monday Day Special - 5.26.25

Hosted by @₿itcoin ₿utcher 🥩 🐑 🐷 · 2026-05-24 · Tags: IREN

TLDR

The hosts assessed IREN's construction progress, financing, marketing overhaul, AI-cloud strategy, and potential revenue growth through 2031. They remained broadly bullish but acknowledged execution risks, disappointing recent AI revenue, uncertainty around GPU pricing, and the need to accelerate liquid-cooled data-center construction.

Speakers

Notable quotes

Transcript

Frans Bakker: Hey, Butcher, were you talking so far? It's been quiet on my end.

₿itcoin ₿utcher: I was just playing the music in the background, but I guess we'll get started because you unmuted the background music, but we can get this going.

Frans Bakker: That's really odd because I didn't hear any background music. I don't know, maybe it was just you or if other people heard it. Well, it doesn't really matter. Anyway, we started it. So let's go.

₿itcoin ₿utcher: Yeah, we're back, everyone. It's good to be back. And we took a little break and a lot's happened in the past two weeks. So I'm with my co-host, Franz. I don't know why people are putting their thumbs down, maybe because we took the break last week. I hope they can hear us. Maybe a thumbs up if you can hear us, that would help. But yeah, there's so much to talk about. So why don't I just kick it off to my co-host friends and some of his opening thoughts. There's a few different directions we can take this tonight and we'll just kind of see where we end up. But I don't have a specific agenda. plenty of headlines that we can go through and it might not be organized. We might be a little rusty after a week off, but we're going to find a way. So why don't you kick it off, Frans?

Frans Bakker: Yeah, so I didn't prepare much for today. I've been very focused on our analysis of Childress. Iren has been working really hard to try to catch up with Horizon 1 to 4 as a total project for Microsoft. And it's really showing that they are working on multiple fronts at the same time. You know, if you've been following the community in the last couple of years, you know that Iron likes to work staggered. So they build concrete slabs first, and then they start to build shells in a sort of sequential order where they can work on multiple buildings at the same time. So I've been really focused on Childress, not so much on the financial models. I know you've been working on those, Butcher, and I guess you're going to talk about that. I'm not sure if you actually published them already, but I remember you were saying you were going to build like a ARR model for 2031. But yes, okay, I see. I see your post. And that's now also I'll leave that to you. But so yeah, operationally and based on the construction progress that I've seen, it looks like first of all, you know, they are increasing their pace there are a lot more people on site on average. And so I think they have sort of figured out or solved the people problem, if that was even a problem in the 1st place. But what I do think, what I'm seeing so far is, a little bit of a lack of certain components slash material materials on site that I would wish that would be there by now, sort of. I guess that is a high-level take on how I see things. But, just to, the short version is, I think they will deliver Horizon as a project to Microsoft within the deadline that they have for Microsoft. I'm not 100% convinced it will be in calendar year twenty-six, but I think Iron has a, an internal deadline for Microsoft that is beyond that, or at least with a grace period or something. So, that's the first thing that I'll just launch here is that I think they will, be within that deadline for Microsoft. Maybe not within our own expected deadline, like retail community, we want it to be ready as soon as possible. It's more back-end weighted. And I'm personally, like, let's say it's 65, 70% convinced today we'll finish it in this calendar year and, like, 35, 30% that it will trickle into Q1 2027. So without going too much further into it, that's the first thing from the top of my mind to talk about. And maybe you can throw up another subject before I just make this into one big operational space again.

₿itcoin ₿utcher: Yeah, so apologies everyone for the lack of prep. There's been so much going on with my own personal business, but looking at just the press releases in the past two weeks, we can go a few different directions, Franz. We had the convertible note. that close that originally was issued for $2.6 billion and got upsized to $3 billion. That's certainly important from a financing perspective. There was the acquisition of Awaken. I'm sure you have thoughts on that. Also, I'll post it in the Nest, but Dan's tweet, and now Dan went from not saying anything to saying everything. on X, which is I'm sure welcomed by the vast majority of the community, including myself. So any of those you want to dig into first or I've also posted Jonah Loopton, who's a shareholder, had asked a question earlier in the day about forward guidance. Going into 2031 and I didn't build this out to 2031, but I wanted to give him a teaser tonight and I threw what I posted in the nest together in the last 15, 20 minutes before the starting the show just. got home from a family thing and but I wanted to have something that we could talk about potentially just because I wasn't sure if you had anything to discuss. So just as a backup plan, so I'm happy to go through that. So I'll leave it to you, Franz, anything that you want to dig into first or I'm happy to discuss what I put together.

Frans Bakker: Yeah, I can give my thoughts on those two topics. So first of all, convertible note was great. I think, it was above our expectations. We had some, there was that Bloomberg terminal sort of leak where the conditions that were, talked about were not that great and it turned out to be better than that and more money than that. So I think that was a pretty good deal. Yeah, it's long term. I'm not exactly very happy with convertible notes. And, I'm not really concerned about the dilution. I'm more concerned about, the volatility that seems to be incorporated into the stock because of it. And we have a bunch of other convertible notes that have really wreaked some havoc in the last months, especially if you remember that rolling of those converts. You know, I I am not a big fan of that. I just think it's completely manipulated. Even at a $20 billion market cap, it's just a handful of trading desks that can completely control the stock at a whim. And I think that is, you know, it's not something that I'm happy with. And these new convertible notes will undoubtedly add to volatility over the long term. So there's a good thing about it and a bad thing about it, but obviously the financial result is good. So I guess we'll take it and let's hope they will have some, good purpose for the money. there has been a couple of convertible notes in the past where there where nothing happened afterwards. And there's been one time that they announced a Microsoft deal afterwards. I don't really think there is like a pattern you can draw. So I'm not like waiting for an imminent deal or something just because of a convertible note. If anything, it's more likely that there's nothing after a convertible note if you based on the statistics. But so that's that. They have plenty of money. available right now. So at least we don't have to worry about them being strapped for cash and needing to tap the ATM at these levels, which they have done. But yeah, there are some explanations for that, which we can talk about later. But so, and then yeah, the whole awakened acquisition. I just want to make it clear one more time. It's not a real acquisition. It wasn't acqui-hire, as it's called. So the people awaken just, flip to iron. They literally just signed a contract that we will now work for you and we will take, this kind of salary for the workers, like the general staff guys, and the management will take like compensation, like RSUs or something. And of course, that one guy, I forgot his name, I think it's Chris or something, got a CMO. So, you know, he's now in the C-suite. So I guess the deal was good enough for him to, you know, don't, you know, there was no like a purchase price or anything involved. So it's not like a Mirantis or Nordstrom Group or something. So I think it was a bit overblown by the retail community again, that why are they buying a company? Well, that's not literally what happened. So, that was interesting. And I think we can see what they're doing with this marketing media, with the collaboration with Awaken. Obviously, they've already been working with them for like more than six months. But what we are seeing now seems to be a very different style. And I think it's just a whole, it's a whole new iron almost, with Mirantis on top of it and now Awaken also, in the ranks of iron. We're just seeing a lot more frequent posting. The type of posts are different, they're better articulated. And I like it. I think this is a bunch of things like previously I've been asking Iron like, hey, why don't you guys talk about this? Why don't you share a bit more information about your progress of horizon construction? Like, why don't you, instead of no more monthly operational reports, okay, we get that, but why not just give like a construction update sometimes or something? And It just seems to me that they are really making, moves now to improve their frequency of posting and sharing updates. And, if you see the Mirantis X account, they're like posting sometimes two times in the same day. I think there's going to be a bit of overlap there as well. maybe Mirantis has like a very good social media team and they'll say like, okay, we will help you guys to get a bit more engagement. And if you compare Iron's X post or X accounts engagement with some of the peers, it's extremely obvious that Iron has got a lot more engagement already as it is now. So Yeah. And then there is, of course, a whole big marketing budget with the sphere advertisements. There is also a bit of context to that. There, you know, there's one retail take that says why would they need to advertise if the power is so, you know, scarce and they don't need a sales team because they have Nvidia. So why do they need marketing like that? Why are they blowing $1,000,000 on a sphere advertisement, for example? This is just a very short-sighted take because Iron's not targeting, you know, customers only with this thing. It is a broader, you know, brand awareness campaign where they're trying to, you know, sort of get on par with the NVIDIA and Dells of this world in terms of impact and, they just don't want to be trading them with marketing outages. And I think that it is mostly directed at a mix of local communities. if you think about it, there's a lot of opposition for data centers right now, and it's not only in the United States. So they want to build a global brand that is recognized as sustainable, as, concerned about, considerate about the communities that they operate in. So I think we're going to see, a campaign that is focused on that. That's one facet of it. And then, of course, you have to consider if they advertise on trams, who owns the trams, right? So you have the people that ride in it and you have the people that own it. So there's, it's like targeted at multiple layers of the audience. And it includes like local governments, federal governments, regulators, and of course, local communities. And then of course, they also want to like impress enterprises and, create brand awareness that they are the global leader of AI cloud. So it is like a mix of purposes and it's not something that you can just, put your finger on and say, okay, it's specifically designed for this only. It is like, just stepping up their marketing game. And I think that we're seeing this now with social media as well. I think Iron is just reinventing themselves with, partially by acquisitions and hires. And then, it's just a whole new move that they are doing. And, they're leaving Bitcoin behind and they're going fully into, the next big thing in AI cloud. And now with Nvidia, on board. I think marketing and brand awareness just gets a whole different, reaches a whole different level. Some of you may know how Nvidia is like completely in control when it involves their patents and their intellectual properties with regards to their own logo, for example. You need to be like, ask permission or something to use their logo in a presentation or something. So the brand needs to be, more valuable as a brand on itself. So I think that that's what they're sort of trying to build. So I would just say that don't focus too much on it in the short term. Like it's not really, it's not like impacting the budget or something. They're not tapping the ATM to pay for a sphere ad, like those kind of takes are just a bit silly in my opinion. It's just, you know, getting on par with their potential partners in the DSX space going forward, right? And that's just one of the examples. It's a lot bigger than we currently can think. I'm still trying to wrap my head around it as well, like what are they really doing? To give you one example, this is sort of unrelated, but I'm just putting my tinfoil hat on just for a second. We have a new satellite image from Childress and it literally shows that there is some kind of ramp going under the new office that they built. So it feels to me like they're sort of like, you know, What are they going to do? Is this some kind of secret Batmobile garage where they're going to like design all these futuristic things that they can drive straight under the building and out of it? You know, it's because they have so many parking spots, right? They wouldn't need an underground parking just for parking. So I'm trying to figure out what this is about. Like it feels to me like this is for, maybe it's for VIPs or like, I don't know, like people that fly in with an airplane go into limousines and can drive straight under the new office building or something. you start to think about government people or I don't know, like agencies or super VIP people. So why am I saying this is because I don't know how big it is. they made their own post where they said the next big thing turns out to be pretty big. we don't know what they're cooking up. We don't know who they're talking to. If they already have Dell and NVIDIA now lined up for the DSX flagship in Sweetwater, you know, who knows what other parties are interested in this much power. I mean, this could be bigger than we can currently grasp. So I think leveling up the media and marketing team is just part of one part of this. I think it just, I think it's going to be very valuable looking in a few years. And that's just my two cents on the recent developments. If I forgot anything, there's probably more like Dan interview on CNBC or where did he show up? There's, yeah, what's this Charles Payne? That was kind of cool, right? Where he mentioned there are cult following. So, yeah, I'll just pause there. Otherwise, it's going to be a complete rant.

₿itcoin ₿utcher: Yeah, Charles Payne was a good opportunity to get Dan on stage. And I think his presence I don't you don't want to read too far into things, but I do feel better. I mean, if I'm going to if we were going to complain about how they presented on the earnings call, then I can at least acknowledge that him on stage with pain that he did well. And then the newer posts, I'm not sure that he necessarily wrote it, but I think it whether it's someone within Mirantis or certainly Awaken, leaving their mark on the business early on. So I do, I mean, when I think about where this business was last year, it was only a few months before they were trying to announce reaching the 50 exa hash. I think that was by June or July. I don't remember off the top of my head, but now this is a business that we're talking about potentially at least pursuing a hyperscaler, you know, that's the end goal, being a trillion dollar market cap. And Mike openly talks about that. And whether they get there or not, I'm not sure, but I like our chances given what we're doing and the guys that are behind it. So I'm just trying to enjoy the ride. I mean, we still analyze, I still put together stuff to try and bring people along for the ride and have them maybe see how I think of things or how you think of things, Franz. But a lot of this just gets back to, do you trust the guys in charge? And there was a lot of disappointment earlier in the year, but I think the tide is turning and it's starting to make a little more sense what they were doing in the background. And it's encouraging. I'll take a second here, Franz, before we get back to Dan's tweet. We can briefly run through what I had posted, if that's all right with you, and give you a little breather. But I had a-- so Jonah had put out a post earlier today asking members of the IRN community to come out with an estimated 2031 where would annual recurring revenue be at? And I thought, I gave him an estimate that it would be in excess of $40 billion. And then you get the Jim Chanos of the world. Someone, I forgot the name of the gentleman who had a post, but he came out with 65 billion plus. And then Jim responded to that about how are we gonna fund it? So I think for purposes of not necessarily appeasing the bears, but sticking to what Dan has guided, and then maybe we can build off of that. I went back to the earnings presentation and there was a slide specifically that was detailing how the company planned to build out the portfolio in the next few months here to end the year, but also what they plan to do in 2027. So I thought I would run everyone through that very briefly just to show. And this would actually, once I talk through this, I'll certainly ask for some feedback from Franz, but also I think where it gets, where this is kind of related to his work earlier on Childress and Horizon is kind of what is a fair expectation going forward on how quickly we can construct data centers because my revenue projections are only as good as our operations are and building those data centers. So let's take a second here. I have in the post the slide that I was referencing so that you guys know that there's some truth to what management is saying and what I'm saying. I added a little of my own spice on it to try and bring it to life. I think what I was a little disappointed with when Dan and the guys put this presentation together is the fact that they would commit to having 1.2 gigawatts constructed by the end of next year, but they wouldn't do an estimated guidance related to it. I think that would have activated people's imagination a little more, but that's where I come in and that's where I'm going to take a first stab at it. And it's far from perfect, but I know it's directionally correct and I can, I didn't have a chance to outline my assumptions, but I can explain what I put together here quickly. And this is supposed to be a very coarse draft that, you know, if you want something sexier, go see God Particle or Agrippa. But my job here is to get the conversation started and maybe those guys can run with it. And certainly I will take another stab at this later and kind of see where we can What does 2028 look like? What does 2029 look like? And I haven't had a chance to connect with Franz, and certainly I would want to hear his input on construction timelines. So I just thought, though, the company guided 2026 and 2027, what they plan to build, and I want to show how we got to those numbers and where I think revenue can be at the end of next year, which I think is close to $12 billion. So at the top, If you don't see it, it's in the nest. Otherwise I have it posted in my profile, but the we're all familiar with the $3.7 billion number and all I've done here is I've tried to separate everything by site. So you have Prince George. Childress, Mackenzie, and again Childress, and I've separated Childress into the Microsoft bucket and then the 50 megawatt bucket that's related to the B300s. So I just show how those components, Prince George being 500 million, 1.9 billion for Childress. And then the recent 50,000 GPUs split amongst McKenzie and Childress add up to 1.3 billion. So that's how you get to 3.7 billion that they're targeting by the end of the year. What I would say is I think those numbers are full of **** after seeing what they signed Nvidia for. Maybe Prince George is closer and certainly Childress is locked in. in stone here, but Mackenzie, if you added up the 1.3 billion and divided it over 50,000 GPUs, that's 26,000 a year, which at an annual GPU hour is less than $3. But we have reason to believe that we just signed Nvidia next year for just under $4 at around 377. So I personally think that 1.3 is underestimated, and I've had other posts related to that. So that 1.3 probably will look like 1.5 to 1.7. So that's an extra 200 to $400 million. So we might exceed $4 billion by the end of the year just from the uptick in pricing. But right now, management isn't guiding for that. And then you can see the gross megawatts, the 50 at Prince George, the 300 at Childress, the 80 at Mackenzie and the 50 add up to the 480 that are shown on the slide. So where the analysis comes in is Dan in that slide commits to 1210 megawatts for 2027. That's cumulative, meaning the difference of 1210 and 480 is 730 megawatts that they plan to build out in 2027. And again, in 2027, I've broken that out and that's 30 for Canal Flats, 60 for the Nvidia deal at Childress. The rest of Childress is 340 because the 340 and the 60 adds up to 400 and then the prior year you had 350. That's how you get to 750 cumulative for Childress. And then lastly, Sweetwater, 300 megawatts gross guided and I come up with estimated critical IT numbers. The estimated ARRs, how I get there is I basically benchmark it against the Nvidia deal and kind of assume that there's a hometown discount. And I think it's conservative to assume that B300 pricing for the Childress and McKenzie buckets this year and going forward, the childress, the remaining 340 in 2027. I think it's fair to say that Nvidia got a very friendly deal given the partnership. And I think it's fair to say that we'll be charging more than $4 an hour for the remaining unless someone takes the remaining site. If you had a hyperscaler or Anthropic or someone, maybe it would be sub $4, but if it's exclusively enterprise and you have Mirantis, you know, adding their value to the transaction via whether it's Kubernetes or any other value add that they can provide that will enhance the revenue per megawatt. So that's kind of a very high level overview. If you add all that up, that's how I there's approximately 8 billion added to the three seven. So I come just shy of 12 billion for the end of next year. But when you say that out loud, you know, we're pushing for at the end of next year, we could potentially triple by the end of 2027. And it's honestly not that difficult assuming that they figure out the retrofit piece fairly quickly in Texas. And I have reason to believe that with the pains of horizon one kind of close to behind them, Horizon two through four and maybe this is where Franz can chime in and it gets back to all of this is achievable if we can build on time. And I don't have concerns about GPU deliveries any further after partnering with Nvidia. So kind of rambled a little. Franz, if you have any questions or feedback, I'd love to hear what you have to say. Otherwise, if anyone has a question, feel free to raise your hand and I can do the best I can to address it.

Frans Bakker: I don't have a particular detailed response right now or a question. I just think a high level, I think Iron has been guiding mixed between conservative and existing customer contracts that may or may not roll off. during the period. I think it seems to me that there is another way to look at their end of year ARR internally. And that's if we go by the quad particles theory of the 50,000 B300s have not been contracted yet as of our most recent data point, which was the earnings report and the filings that came with it. It seems to him that Iron has a certain way of reporting contracted revenue that is, you know, ahead of, it's, I'm not sure if it was the RPO or or another way to describe it in their fightings, but he has basically figured out that Iron has not contracted yet any of those P300s that they purchased, so 50,000 of which roughly 17.2,000 will go to Childress and the rest will go to Mackenzie. So you could also argue that Well, perhaps they are, you know, in trying to make a deal for a big part of those or all of them. I don't know. I mean, and if that is a, you know, a prestige kind of project where they rather take a lower rate for, you know, a high credit worth credit worthy counterparty, then maybe it's not really sandbagging, right? Or conservative. Maybe it's just based on actual range of negotiations or something. I mean, I, I also don't exactly subscribe to the, the take that this guy, Lazarus, whatever his name is, had where he said that the rates of B200s are deteriorating in Prince George. I know that you have made a conclusion as well that it seems that we are paying more for the B300s but guiding lower on a GPU hour basis. But it seems to me that it's probably somewhere in the middle of all this. I don't want to go with the ultra pool case that we should just assume that the NVIDIA contract is now the basis for every further to be contracted GPU. It's, I guess it would make sense that we should distill or try to approach the managed services that Iron's providing to Nvidia on top of the bare metal, right? It's like Kubernetes and whatever they add on top of that is, you know, impacting the price. So Can they, maybe that's why they didn't pre-contract any of those B300s yet, because now with Mirantis, they will be able to, offer a different menu to the potential prospective clients. So I guess it's going to be somewhere in the middle there. And should they have upped their guidance because they acquired Mirantis and they already had that information by the last earnings? Yeah, potentially. But I guess what they're also, they are very aware of that they underperformed, right? I mean, you can say whatever you want, but $33 million of AI revenue was just a massive letdown. regardless of the nuance or context around it. was just wasn't exactly great. So, and because we didn't get an explanation, we know that it was mostly the GPU deliveries, but if they are not going to talk about these issues, like not a single analyst question was about why did your revenue suck, right? And that's kind of odd. We can keep looking forward and kicking the can down the road, but someone needs to ask the the painful questions. And if they don't do it, then we'll do it. But yes, so I think Iron is guiding conservatively based on everything I've said before this. But I think they may surprise us with one or multiple contracts with regards to those 50,000 B300s, including Mirantis added value. So yeah, I guess your, if you have these things incorporated in your expectations, like a sensitivity between high-end Nvidia, low-end Iron's guidance, I think it will be probably somewhere in the middle. That's my take on it.

₿itcoin ₿utcher: Oh, sorry, just unmuted. I would add to that that they had the slide in the presentation. Granted, it was, I mean, it was as of three weeks ago that they had the 3.1 contracted, which was the one, the 2.4 for this year from Childress and Prince George. along with the 700 roughly next year for NVIDIA. So what I think happened, two things related to what you just said, Franz. First, we all know that 33 million was disappointing and not what we collectively expected. So if you're Dan and you know that you didn't hit the mark and most of the market or you're you know, if people are expecting 60 to 80 million realistically and you didn't plug in, kind of looks like a dumb move on the outside if they're raising guidance when they didn't meet initial guidance. So I could understand in retrospect why they kept things where they're at, but I do expect for them to surprise on the upside going forward. My second piece related to that is if they knew as of March that they were going to acquire Mirantis, which I think it's fair to say that that transaction took at least a few months to complete, it made no sense for them. To your point, if you had a new value add proposition, new menu items, as you put it, then why would you undersell yourself and get a lower rate when you could potentially solicit more value from your new acquisition. So I think in hindsight being 2020, you want to call it copium or whatever. Like I think it makes sense though that they didn't commit when GPUs are as valuable and as scarce as they are right now and nothing sits idly. If you're trying to maximize your value per megawatt and your value per GPU, then waiting those few months until it was closer to delivery and until you completed the transaction that not only you have more in-house capabilities, but you also have a new Rolodex of 1,500 enterprise clients that are most likely going to execute at a higher rate than a hyperscaler, then it kind of makes sense that they Weighted and I do expect I I I think we kind of. I get your point that there's managed services and maybe that an potential clients wouldn't. Necessarily need as much as NVIDIA potentially, but I also think you could see my point where if. If anyone had more negotiation leverage than a hyperscaler, it would be Nvidia. So if they're dealing with enterprises, they might not provide the same managed services, but it's also most likely going to be a similar size deal or slightly larger with a company without as much influence as Jensen. So I I'm comfortable saying that maybe it's not the 377, but I. I got to think that 350 might be the floor going forward. And if it's not, there's got to be a really good reason whether it's we're doing business with Anthropic or something that we just have to trust management with. But that's kind of where I was coming from. Let me ask you this question. And there were a few people in the crowd. Nick was asking specifically, and maybe you can he might have missed your earlier commentary on Childress, but With the updates at Childress, you were saying you were pretty comfortable that we would meet the deadlines for the Microsoft deal. My question to you is, if you can refresh people who might have missed it, how you came to that conclusion. And then the follow-up question would be, I just built out that analysis based off of what they guided, where this year they expect to have 480 megawatts completed next year, they expect to have an incremental 730, but a lot of that is air cooled retrofit capacity. What's a reasonable number in your mind of liquid cool capacity that they can bring on annually in 2028 and beyond? So those are a few questions and want to hear where where your head's at.

Frans Bakker: The first question was, you asked me why did they miss their last quarter's guidance?

₿itcoin ₿utcher: No, Nick was asking, I think Nick showed up late. Nick was asking for related to your subscriber tweets on Childress, what led you to the conclusion that you're comfortable with them delivering on time to Microsoft? And then my follow up question was, A lot of this revenue scaling has to do with their ability to build. And I think the bear case would be they've clearly fallen behind their initial expect or maybe our expectations and we live in a bubble. That's a whole other conversation, but they haven't built nearly as quickly as they did for Bitcoin mining because it's a way more complicated process with liquid cooling. But what gives you confidence that in the future, like they have new experience, you expect Horizon two through four to go more quickly. And certainly they've learned from that and they've already implemented construction improvements at Sweetwater. I'm trying to, from an analysis standpoint, if I'm going to say that they can build in the next four years 3 gigawatts and do 750 gross per year. Is that reasonable or am I full of ****? So that's what I'm trying to for everyone here. It's like at what point do they kind of find their stride and get into a rhythm like they did with Bitcoin mining where they just started popping out data centers every month, 50 megawatts at a time. Can they get to a similar cadence for liquid cooling DCs with what they've learned so far?

Frans Bakker: I think Nick should just request to speak. I think he wants to share a theory. So I'm going to ignore that first part. I'll just focus on what I know best, and that's how many data centers can they construct in a year. I think it, well, yeah, so your concern is that in the 2027 guidance, there's a lot of retrofit capacity. So how much liquid-cooled data centers can they really build per year? So they are going to build out 100 megawatt of IT load of liquid cool data centers in Childers and 200 megawatt of IT load in Sweetwater. So that's 300. Is that their cap? Is that their max? I don't think so. I think it's important to understand that 2028 unlocks a lot of capacity. And even more so in 2029. So if they want to keep this year-over-year growth of capacity up, they will have to scale wide. That means they will have to become operational and start building on more than two sites at the same time. Canada, I'll just ignore because I think they will conclude Canada completely next year. I think, there won't be much building going on there. It's mostly retrofits, maybe, a couple of smaller like buildings that are not, I don't think they're going to be building data centers like they did in Prince George. They have a permit for two more smaller, sorry, for one more smaller building. But I don't, I'm not sure if they're going to do that if they don't get more power, which seems to be like a bottleneck. So if we focus on taxes, we know, sorry, on North America, no, on the United States, that we know they have Oklahoma, where they are going to start paying taxes in 2027. And I think they're going to start building in 2027 as well. to attempt to have at least 50 megawatts of operational data centers in 2028. I'm not sure if that is a realistic goal or if it's going to be more than that. But conservatively, I think they will be able to pull 50 megawatts out of Oklahoma. Bouquet is 100 megawatt. And then In Childers, they don't have any more power. Childers should basically be concluded in 2027. So that just leaves Sweetwater 1 and Sweetwater 2. Sweetwater 2 is a 2028 thing as well, but then we'll also see like site preps and civil works in 2027. So how many data centers can they really build across these three sites? in 2028. I think if they are going to build 200 megawatt in Sweetwater next year is basically going to be the horizon one of Sweetwater. So I think that that building that they have rendered in that presentation is going to be what horizon one was for Childress. So That's why I think they're aiming low and they're just going to take their time to deliver a DSX AI factory for 200 megawatt of IT load together with, you know, NVIDIA next year. And 2028 will be a whole different story. So if they can somehow professionalize perfection. I just made a new made-up a new word perfection. This process of delivering a 200 megawatt IT load DSX AI factory with NVIDIA next year, then they can probably do this a lot faster in 2028 with a bigger team. We know how big the site is. We know how much they have been preparing this site. You of all people know that very well because of all the satellite footage we have. But it just, it must be in excess of, let's say, 300 MW IT load. So 2028, in my opinion, could see 50 to 100 from Oklahoma, 300 to 400 from Sweetwater, and maybe 100 megawatt from, 50 to 100 from Sweetwater too. So base case or bear, I don't really want to give a bear case. Let me just say base case is a 400 megawatt and a pool case would be 600 megawatt. for 2028. I did say 1000 before in the chat, but I will just take that back because that's unrealistic. So because there is no retrofits in 2028, I think in the United States, they will be able to construct between 400 and 600 megawatt of liquid cooled data centers in terms of IT load. If there's anything in Australia or are there, if there are other sites that they start to build in, build on, then it could possibly be a bit more. But I think that if you see year over year, this will be a massive achievement, obviously. So I think the peak, the peak of liquid cool data center construction cadence is not 2028 yet. is close, but it will probably be 2029 or 2030 where they can approach my absolute bull case of 1000, you know, a gigawatt of IT load per year. But that will require at least like 5 sites in multiple countries. So yeah, that's that's that's my expectation for 2028.

₿itcoin ₿utcher: Thanks. I as you say that I think they've gone through growing pains, which we've all acknowledged, but it gets back to the, I'm revisiting my prior view where I was, I was really early on considering colocation or other means to activate this power, the time to compute and signing something earlier because I almost thought with the behind the meter coming online that we would lose this energy arbitrage and that we had to strike, you know, next year in 2028. But as I say it out loud, it's kind of like some of these projects haven't even been approved yet and they probably won't be built till 2028, 2029 anyways. Not to mention, I think another topic we can touch on briefly, but we've been going back and forth with the CEO of DataOne related to the Vineland site for Nebius, not because we're trying to, no one here is trying to FUD Nebius, I don't know how to pronounce it, but It's clear, though, that they couldn't file proper permits for their site, didn't get approved, and had to pivot like any good business would to Bloom Energy to get their own, you know, to replace the power supply and use fusion technology to power their data center, but they have to pay whether it's Data One or Nevius, there's still $2.6 billion extra that they didn't account for initially that's going to be added. And it's kind of funny that within the same week, they published new updated pricing, and certainly it's relative to the demand environment, but you can't help but ask out loud. Did that come as a result of knowing that they had a new $2.6 billion that whether they had to pay it directly or they have to pay a higher fee if that influenced their decision making? All of this is to say that the grid connections, I don't think can be we can't appreciate them enough. And like now you're starting to see some of the **** hit the fan with some of the competition. And while they may deliver that site on time, it is clear that it will be less profitable. It's it's just math like and anyone who denies that is just being intellectually dishonest. Maybe they do have a software layer that's superior and maybe that makes up for it, but they pay colo fees and now they're paying for fusion power. And we don't have to deal with any of that ****. And that gives me comfort as a shareholder. But I think where our edge and how we win this neo-cloud war, Franz, is can they get back to where they were as a Bitcoin miner, where they started kicking everyone's *** because they built better data centers faster and plugged in miners faster. And it's kind of the same game all over again. Who can plug in GPUs faster and take advantage of this pricing environment? Because there's delays throughout the industry. We claim to have the best procurement team. But we had our own struggles with GPUs, which I think we address now with NVIDIA and now with Mirantis. I think their in-house capabilities help us with our managed services, certainly enhanced margin, but also just simply plug in faster. So that's kind of the, again, is it Hopium or are we gonna see it play out? That's kind of why I'm still bullish here. I've always been bullish, but why I'm more bullish is we're finally addressing some of those things and maybe they were and just now they're being communicated a little more directly or indirectly to us. But if you have any response to that, certainly go for it. Otherwise we can go to Nick and Mark's requesting to come up and I'm going to invite him on stage.

Frans Bakker: Well. You know, I was one of the people talking with Mr. Data1 in the comments, and I admit I was a bit direct and maybe a bit blunt, but honestly, I appreciate him trying to make the time and effort to talk to the folks on here. He is, you know, despite being, you know, talked to pretty roughly strongly. So I'll just start with that. But, the thing that I was trying to uncover is who is at fault here. He was really posting that, oh, we're so proud that we are holding these ship engines. And it seems to me that it's his choice and his project. But Why is Nibius the one having the contract with Bloom Energy if it is the power source that is taken care of by Data One, right? It's Data One's site. It's a co-location agreement. The thing with co-location is that, and he was Charles was being straightforward with this. We're passing through the cost to our customers. So I don't understand why, who's, you know, why is Nibius paying for Bloom Energy's OPEX? Like they're paying like a flat fee for a 10-year agreement of fixed power. And it's up to Bloom Energy to deliver that. So if Bloom Energy has some kind of issue, they need to fix it because Nibis is paying for the firm power, so to say, of 250 megawatt. So I mean, that's at least how I read it. So I just don't understand why is it Data One's responsibility to handle these Bergen ship engines, but now it is somehow Nibius responsibility to take care of the Bloom Energy contract and delivery. It's just somewhere here there is a data one and Nibius are intertwined in a to a degree that I don't really understand it. It doesn't seem to me like a traditional colo agreement. That's why I was trying to push him to be transparent on, first of all, who is to blame here. And second of all, who is paying to fix it. Is it only Nebius who is just going to pay $2.6 billion over 10 years to fix a mistake that's been made by Data One? It seems that doesn't make any sense. So what if it was Nebius all along who made the mistake of these ship engines, right? Maybe they consulted with Data One and said, can you operate these things there? And this guy said, yeah, sure, we can. It's not so hard. We can use these step-up transformers and blah, blah, blah. And then later they found out that it was hard to get the permits or something. You know, it just doesn't seem to me that this is all Data One's fault because If it would be Data One at fault here, then Data One would be the one to get the contract with Bloom Energy to fix their mistake with, you know, with Nibius, right? But now it's Nibius who is going into this contract. So that's something that I tried to uncover and on the way, it just, you know, digressed into a different kind of conversation and ultimately people started to call this guy a bunch of words, which was kind of disappointing because, I was actually trying to keep it like sort of civilized to get him to talk about, what was really going on there. But he's not willing to throw his customer under the bus. And to be honest, the way he's talking, it seems to me like he's a guy with no problems. So I think that he really is not the one to blame for this mistake. And I know that Nibius Bulls that will hear this are going to say, of course, here's the iron guys trying to, you know, gaslight Nibius again. Well, it's obviously a mistake of both of these parties and it's not a real mistake. But the truth is there is going to be additional costs over this contract term and even beyond that. So you can ask yourself, after five years, the Microsoft contract runs off, Nibius doesn't just end up with chips that are depreciated. They also end up with a contract that they have to pay for. I think people forgot that Microsoft revenue stops after five years, but the expenses towards Bloom Energy keep going for another five years. And I know it's a monthly event. It's not very impacting the business in five years. Yeah, I'm sure. But they have literally gone into a contract that extends beyond their, you know, beyond their income contracts. So That's something that you can also consider how great of a deal that is. Apparently, they have plans to keep working with Data One beyond the first five years. Otherwise, why would they contract Bloom Energy for 10 years? So I'm sure that there are people who know more about this, but it all seems to me extremely opaque. And I want to know what's really going on there. And I guess at some point we will be able to find out what was the truth behind all this. Luckily, we have a bunch of people that are very well connected in Vineland. And of course, we can get the SEC filings over time. But yeah, so I'm not going to get into the unit economics now. I mean, you know my thoughts on that. I know that their margin is going to be absolutely horrible coming out of Vineland. There's just no way around that. Nibius signed a contract with Microsoft for a certain GPU hour quantity rate for a certain amount of GPUs that we don't know. We just know it's a fixed amount over a fixed period of time for a fixed amount of megawatts that we don't exactly know. Is it 300 or 350 MW gross? I don't know. But the truth is, they have added expenses after they signed this contract. So these $2.6 billion over 10 years, if you take five years of those and you add it to the expenses of the contract or you subtract it from the headline, it's not going to look that much, it's not going to get any better, it's just going to get worse. So, and then of course, the case from the Nibius investment.