$iren

Hosted by @₿itcoin ₿utcher 🥩 🐑 🐷 · 2026-05-26 · Tags: IREN

TLDR

Participants analyzed IREN’s Horizon One cooling delays, debating whether rear-door heat exchangers were related to temporary chiller trucks or whether construction sequencing and pipe design caused the problem. They also discussed possible Sweetwater construction, an Australian grid connection, prospective NVIDIA and Anthropic relationships, financing and dilution, and expectations for a new commercial deal before the end of August.

Speakers

Notable quotes

Transcript

₿itcoin ₿utcher: Sorry, I got rugged, everyone. Let's get Franz back up on stage and I'll try and retweet this.

Nick: Hey, hi guys. France, we were enjoying your rant. Sorry, got interrupted. So I want to share, hi everyone, an informed theory regarding Dan's post with the motivator picture. The other post is self-explanatory, so I don't have anything to add to that one. But there was a little bit of a confusion with that post and people started, talking about, different things. I think Dan was trying to throw us a teaser about the useful life of the GPUs and as well. try to explain the confusion about the development of Horizon One. First of all, I wanted to shout out to Franz for his deep dive on the on the latest children's picture that he shared on the on the subscribe subscription channel and his honest reporting about it. And so My first question when I saw that post was, why is he sharing the rear of the servers instead of the front of the servers? I wanted to explain here that these Motivare, it's a heat exchanger, so it's not the biggest part of the cooling system of the server. It's just a combined solution with the direct to chip liquid cooling, which we don't see in the picture. And then the back of the server where you can see these heat exchangers, that what they basically do is they they capture 100% of the heat that comes out of the server. And instead of throwing it into the room, is captured by the coolant and then is recycled and the chillers cool that coolant back to whatever the desired temperature and then throw it back to the heat exchanger so that no residual heat is thrown into the room. Why is this, why are they using this combined solution? My theory is that what they're trying to do is to reduce the in-server temperature 10 degrees Celsius less than what the manufacturer in this case, Nvidia, is suggesting. What there's a theory that says that by reducing 10 degrees Celsius, the temperature inside the server, you prolong the life of the GPU by double the time. So that would be their goal. And they try to achieve it in two ways. First, they capture, as we said, this heat before it hits the room. And therefore, they added this equipment, which is the motivator. And here, this is my theory, so I can't prove it. But this would in part explain why they have these trucks in the facility, these cooling trucks. And my theory is that these trucks are not helping the direct to cheap liquid cooling, which would be the pipes that, you know, we all know. But instead, these trucks are helping this additional equipment that maybe it wasn't included in the initial design of the first building or the first buildings of Horizon One. The other way you you achieve this, this, this goal of minus 10 degrees Celsius would be adding more aggressive hull cooling. And these would Franz shared with us pictures of a lot of train compressors, I don't know if they're compressors or what they are, but they are definitely trained air conditioning units, so to speak. And so these whole thing would have the purpose of extending the life of these GPUs from, let's say, four to five years to maybe 8 to 10 years. Maybe 10 would be a little bit of an exaggeration, but definitely I think that eight years would be totally achievable. And of course, with the intent of rotating these GPUs once they get technically obsolete, but totally functional into inference, for instance. So this, I think this theory would explain a lot of these delays and maybe what they are trying to find is this procurement, what they're trying to find is it's some extra chillers that would cool down the coolant that's coming from this motivator. This is just a theory. It's an informed theory that would explain why these delays, why these delays in the chillers in the horizon one part of the of chillers. And this is what I wanted to comment on. Thanks.

Frans Bakker: Guys. Yeah, we should have the God particle come up and tell you why you're so wrong. But I personally also don't agree with you, to be honest. It's just funny to see like 100 thumbs down in 2 minutes. But yeah, no, I think we are zooming too much in on this Motivair picture. I think it was always part of the plan and it's completely unrelated to the chiller trucks. So the chiller trucks are a makeshift solution, temporary solution to cool the GPUs that are supposed to be running for Microsoft in a coming period. I don't know what the deadline is for the first building, but they needed those trucks to, you know, start burning into GPUs and start operating them. But so, the delay is because the liquid cooling plant for Horizon One has some either some, test tube baby issues or some component delivery delays. I think it's mostly a startup first design, you know, constraint where I talked about this before in other spaces, but the way it's explained to me, and I'll just say it one more time, is as if you have a flower garden and you want to plant grass on top of it or something like that. It's like you're going to destroy everything underneath it. basically, if you're not very careful. So that was some kind of analogy that was explained to me. What it comes down to is Iron started building these shells ahead of the contract. They were probably or possibly already talking to Microsoft, but they were convinced that they were going to deliver a data center that was Vera Rubin compatible. And they started with a lot of above ground work, like concrete and shells. And then later they found out that the liquid cooling with the pipes and everything and underground electrical and fiber was going to be, would have been easier to do before they would do some of these above ground structural work. And They obviously weren't able to tear down the shell again and remove the concrete and do the underground work first. But so they had to fix that. And the same thing, they chose a certain cooling plant design where the fans were facing downward. And that turned out not to be a very good solution. So for Horizon 2, they have a different design where the fin fans are on top. So you have the first massive building you see, which is not a data center, but to the south of the data centers, the first very big building you see is the dry cooler structure. And then you get the modular liquid cooling plant. That's That's the smaller building that's right adjacent to the dry cooler structure. And then you get the well, that's the cylindrical like well. And then there are like a bunch of other things like generators and glycol tank and whatever. But the most important thing is why is the liquid cooling plant and a dry cooler structure not ready yet is because the pipes leading from the plants to the data centers and into the data centers, that was where the issue arose, if I'm not mistaken. So that's how I perceived it. So it seems to me that they have They had this temporary solution, which will be phased out soon. So the plant's almost ready to go, if I understand it correctly. But the whole theory about the motif air and I don't think that has any merit. I don't think it's really related to the issue that they had. because Motivare is installed by Dell and it's not a construction by Iron. So there, it just doesn't make any sense that Iron has a structural design delay or a constraint or a, like a, like I said, a test tube baby error, where which is going to be solved somehow by by Dell installing rear door heat exchangers. I mean, I know that's not exactly what you're trying to say, but I don't think these things are related. I'm sure they always had the plan for the rear door heat exchangers. Actually, to add to that, I actually know that they have been talking about rear door heat exchangers in 2024 already. For Iron, the holy grail has always been to run Blackwells in Childress. That's always been the holy grail. And then with a combination of air cooling and liquid cooling, where they would up the liquid cooling in the summer and run air cooling only, it would be like a two-phased kind of solution, where they would have a liquid to air in the non-peak summer and liquid to liquid in peak summer, something like this. And Schneider and heat exchangers have been mentioned to me as early as 2024 already. So this is not, I don't think it's like a novelty or a new development by Iron. I mean, it could be that this technology that they're trying to perform with a hybrid air and liquid solution is somewhat new in standard practices in data centers in, an environment like Texas. But I don't know if that is true or not. I think Horizon One is probably not the Baker math of R&D. I think it's more like figuring it, figuring it out on the go. I mean, obviously, we've seen that picture from Noah Clark on LinkedIn where he was showing all these manifolds lying on the ground with all these, you know, these cables, these liquid cables running into the building. It's just, it doesn't look like it's ready. And okay, they have fixed that now. There are still some of these compressors and generators pumping water from chiller trucks into some of these valves and some of these pipes, but it's starting to get there. And I think by June, I think they will most likely be running on the plant. So I think that's is almost done. So. my two cents on that. I don't think it's about, those rear door heated exchangers, to be honest.

₿itcoin ₿utcher: Nick, anything else you wanted to talk about?

Nick: No, I mean, of course, France has more information than I do. For me, if it was a... an issue with the pipes, it would be exchange, I don't, I mean, I think it should, it would, a better explanation for that would be some procurement issue. But even so, I would think they should have put that order like years before. So I know in iron, that wouldn't be an issue. And then These trucks, are they hooked to the pipes or are they hooked directly to the data center? that would be another question that I would have. Like, it seems to me that these trucks are hooked to the chiller complex, so to speak, instead of directly to the halls of the data center. So, but definitely, I totally get that motivator solution wasn't was there all along. Just I don't understand why the, why the delay with some design, but yeah, well, thanks, Frank. Thanks.

Frans Bakker: Well, we'll probably, we'll never figure out the... the entire truth behind what happened with Horizon One. It's just, one thing to mention is that Iron hasn't even commented on the fact that they completely changed the liquid cooling plant and dry cooler insulation for Horizon Two. You know, they're showing these pictures from the site where you can obviously see that the complete, the design is completely changed. So I think that's also kind of They gloss over details that are immaterial. And, you know, I just I really don't get that. Why should we take everything at face value? Like we're like we're a bunch of nibious investors, right? I mean, we're here to we're former Bitcoin miners, right? We're here to verify and not trust. So I just I don't understand that way of thinking sometimes, but And then it turns out that we're right most of the time. When we're being very granular, we see things and we talk about it. And people that are not granular say, oh, but just it's going to be all right, bro. Don't worry about it. And then it turns out that, you know, there's some issue with liquid cooling in Texas. And then it turns out there's some issue with power in Vineland. But we are still, you know, being too granular and we should just trust whatever the company management are saying. That just doesn't work that way for me. So anyway, I don't want to make this into one big conspiracy space. I guess we can move on to the next topic if you have one, Butcher, or maybe someone wants to come up and ask us a question or?

₿itcoin ₿utcher: You're in the nest post from God Particle. Specifically, my question, Franz, I haven't had a chance to look into it in more detail, but reference to additional grid connection approvals in Australia. Can you comment on anything you've found in research recently to support this?

Frans Bakker: What's the question for me? Sorry, I was just typing something, so.

₿itcoin ₿utcher: So my question was the God particle I threw in his post and he had summarized notes from an analyst. I think it was Needham. And one of their notes was with respect to Australia, where Iran spent a considerable amount of marketing dollars. And it appears there have an imminent Approval for a new grid connection. Do you have anything to add to that or is it are you kind of where you were last at with your research that you had found two potential states and that we kind of were in the same spot still?

Frans Bakker: There is a. There are just too many noises or signals, whatever it is. There are too many rumors coming out of Australia. And this is just another one to add to the list. There was also this article in that Australian newspaper where they, I think the article said that Iron is stated to start building data centers in Australia in 2028. So that's also why I mentioned it earlier when I was talking about 2028 capacity. I guess maybe someone can pull up that article, but it's along the same lines. It's possible that Iron is in final stages. You know, maybe it's like similar with Sweetwater 2, where we found Sweetwater 2 before the company wanted to announce it. And then They did last year, I think it was in February, and they didn't have the grid connection, the connection agreement yet. Maybe Iran is now in the same sort of state where they, in a situation where they have direct line of sight to a connection agreement or a site in Australia, but they don't want to disclose it until they have the firm power. So I think maybe the new strategy is keep it vague when you don't have the connection agreement. And when you do, you can announce the project or something like that. Maybe that's what we're going to see. But honestly, if I look at Ireland's portfolio right now and the plans for next year, I don't think it would make any sense to talk about Australia in 2026. I think they should just move this to next year because it has no value to the market anyway. unless it's paired with the deal, the market will not give a **** about a project in Australia in 2028. So those are my thoughts.

₿itcoin ₿utcher: I invited Masada up stage, or now we switched back to the listener, so I guess that was short-lived. Related to what you just said with Australia, I think I'll try and invite Masad back up, but I think he dropped off. But I would say with Australia specifically, like I just given the amount of money they're spending there, I thought along with the investment being pursued by Microsoft that we might see something sooner than later, but I can understand where you're coming from, Franz.

Frans Bakker: I see Mossad as a speaker, so I don't know if it's a glitch.

₿itcoin ₿utcher: Yeah, I'm showing listener, but Mossad, if you're shown as a speaker on your end, then feel free to speak. Welcome. Looks like he's gone. Dave is requesting to come up. Let's try Dave.

Frans Bakker: I see Phil in the crowd as well. Maybe Phil should come up and talk about how great of a company Nibius is.

₿itcoin ₿utcher: Yeah, we haven't spoken to Phil in a while. Phil, good to see you.

Frans Bakker: Well, anyway, I can give you guys my personal expectations of the coming period. I'm not sure if it's going to happen in the coming week, but because this is going to be a short week, I sold a bunch of $54 puts for June the 5th because this weekend I will not be home. So I didn't want to have options expire on a Friday when I'm out drinking. So I will also not be part of the space most likely. on next Sunday, sorry. But so why did I do that? It's because I've changed my strategy to be a bit more conservative because this stock is just acting like a meme coin at times. But I think the meme coin period is not going to last for much longer. Traditionally, the summer period is Iran's best season, so to say. I think last year we did really well over the summer, especially if you look at back at August and September, it was like our almost only updates. At least that's how it felt like. And I think iron is brewing something. The, you know, the marketing is one thing and raising money is another one. And Dan visiting the US multiple times in a month is a third thing. It just seems to me that it's all coming together. The these Needham conference things are really important. I think we, the market is really waiting for another, reference point where they get like Kent Draper talking to one of those, investment desks or, maybe Dan coming on a Twitter space with us. But I think we still deserve a sort of like a debriefing from earnings where a bit more details will be spilled. I really appreciate their marketing efforts and how they've now mentioned that the 60 megawatt for NVIDIA was gross capacity. But it just seems to me that there is a something brewing. And I have set my focus on 8 to 12 weeks from now by the very latest. So I will go on a record here and it's just a prediction. It's not financial or investment advice, but I think Iron will announce a commercial, you know, deal kind of disclosure. before the end of August. That is currently my take. I can't really pinpoint it to either Horizon 5 to 6 or something for Sweetwater or something for those 50,000 B300s. But if you look at the expected timeline and the way that they've announced that they are now actively engaging with selling their 2027 capacity, I just don't think that it can take much longer. And yeah, there are just a bunch of circumstantial indicators that gives me the strong signal that I won't say it's imminent, but I don't think we're going to be waiting for the end of this year or something. I think it's going to happen over the summer. I think 60 megawatt with NVIDIA was just it was just an appetizer. That's obviously too low of a bar to perform in 2027 for Iron. And since Horizon is now basically on autopilot after the first data center is as good as operational, I think they are going to take it upon them to sign something else. And I think as soon as they will be able to either deliver Horizon One Supercluster or just the first building, I think that will give them enough credibility, confidence. And if they can announce to the market, we've delivered the first tranche to Microsoft, that will be the start sign to sign a new deal for Sweetwater or maybe for Horizon 5 and 6. And this can all happen within the year before the end of August. And then there's also the fact that they have earnings in August, so I don't think they will come up empty-handed. I expect another pretty bad P&L. I think the reason for that is the accelerated the displacement and depreciation of their, well, the one time hit they're going to take for the retrofit and displacement of the miners is going to be in excess of 500 million. And I know how Iron runs their books and they're going to use financial year 26, quarter four, which ends in June as a cleanup quarter. And I think they're going to hit us with a massive $540 million of one time depreciation and displacement of their mining fleet and data centers in Childers and maybe Mackenzie. They already announced this amount in the previous earnings. So I think August is going to show a pretty good quarter over quarter AI revenue growth. It's probably going to be a 3x compared to the last quarter. So we should be approaching 100 or even exceeding $100 million of AI revenue. But it's got, in my opinion, it's going to be another horrible net result. But I think to sort of keep expectations in line, I think they're going to try to come up with a deal with, you know, at or before earnings. So that's another reason why I think we'll see something announced before the end of August, possibly earlier than that, because they will probably have earnings again in the first week of August. So those are just a couple of my thoughts looking forward. So it mostly pertains to Childress. a small chance for Sweetwater. Oh yeah, to add to that, I just checked the most recent satellite image from Sweetwater and there is definitely something going on that first data center area that they have next to the primary substation that is live together with the bulk substation. So Iron is now potentially building next to an energized primary substation. So you can draw your own conclusions based on that. Iron's not unfamiliar with building before they announce a deal. So this could just be immaterial, but I don't think it is immaterial. I think this is another indicator that they are doing something that they haven't shown in any render or earnings presentation yet. So, what it is, I don't know, but it's we'll be. be the first one to find it out. So yeah, many bullish things to look forward to. I see Sean Swanson wants to come up and talk. I'll let you decide on this one, Butcher, if you want to. He's an iron investor, but he's always very bearish. So I'll leave that up to you. So I'll just pause here.

₿itcoin ₿utcher: We're going to go to Jonathan. Jonathan, how are you?

Jonathan: Pretty good, guys. Can you hear me OK?

₿itcoin ₿utcher: Yes, I can.

Jonathan: Perfect. Yeah, I just wanted to maybe piggyback off of what Fran said there in regards to a deal potentially being reached before around August. I just had a kind of like. an angle I kind of had been thinking about in regards to specifically a deal involving Vera Rubens from Nvidia. And if there would be kind of like, since we've signed that deal with Nvidia, if there's kind of an expectation from Jensen's side there in terms of like when we have to get the, when we have to order those and when We like the timeline. I think it's going to be around H2, right? 27. So like, what is the timeline in terms of like the kind of the last sort of the deadline, if you will, of when you'd have to order those? And then that would have to mean that a deal and a prepayment would have to be in place by that date. And so does that affect kind of like does that almost does that almost mean Nvidia is placing a deadline on Iron? on when they have to kind of get a deal done. So I was wondering if anyone had any ideas about that or if that even makes sense. Thanks.

₿itcoin ₿utcher: I think the question was directed to you, Franz, but I'll chime in. I think with the, as far as procurement and GPUs, I think realistically, That would be a great question for Dan if we can get him back on spaces as to what we should set our expectations for. If they're, you know, is it early 27? Is it mid 27? What does that mean? What I would say with respect to Sweetwater, I personally think we're closer at Sweetwater than people are anticipating because of just the headcount. that they're increasing there. And to Franz's point, if you're only going to be showing 100 million of revenue as of August, most likely related to Prince George, you know, if that was a $500 million run rate and you had a full quarter of install or roughly, you know, that would be 125 to 150 million. But we know that as of March 31st that not all of them were fully up and running. I think it was 300 million of ARR. That's kind of how we're getting to 100 million. But that's a long way of saying that. I think you see Sweetwater sooner than later. Personally, given the size of the site and the emphasis on time to power and what we see In the satellite imagery, I I have a hard time believing something that we energized in May goes much longer than August, and I personally think it's in June or July where we see them figure something out with how that relates to Nvidia yet to be seen, but I just think it's. They've been planning on this for way too long to have it sit vacant for more than a few months after energization. That's just my gut feel and intuition along with the work that Franz has done. But, so Jonathan, I think you were saying that, maybe you can correct me, but is it, are you implying that because of the deposits I've seen anywhere from 10 to 50% down for GPUs with NVIDIA. Are you suggesting that that creates a timeline for Iron to essentially procure the Vera Rubens and that accelerates the timeline for Sweetwater? Is that what you were saying or did I misunderstand you?

Jonathan: Yeah, no, I think that's right. I guess I'm just like thinking out loud and I'm thinking does NVIDIA and Jensen, since they seem to be the ones orchestrating this whole thing, right? They're the ones that are always saying, you know, we're sending all these customers for our new clouds. They're the ones that set up, you know, Sweetwater being this flagship site for their DSX architecture. And all this, it just, it all points towards sort of NVIDIA pushing us in towards, towards a deal with someone, perhaps. You know, maybe that's anthropic as some people have speculated, but I just, I guess I'm coming from the angle of, does, are we expected to just order those Vera Rubins before a deal and a prepayment is secured from a customer. Because if not, you would need to have that deal in place probably pretty soon in order for NVIDIA to kind of allocate those toward, for Iron in 27 for Sweetwater, which they want to do because they want to put those verirubins into that flagship site. That's the whole idea.

Frans Bakker: I don't know if it's a real push from Nvidia, but because we don't really know what the real partnership means between Iron and Nvidia, right? We don't really know who is in control. I know it's obviously, it's obvious to expect it's Nvidia, but would Iron really, you know, offer themselves to be Jensen's ***** just because they have 5 GW of grid secured power. I don't see that. I think Dan is, you know, very happy to play with Nvidia, to play along with this partnership, but he's not going to bend his knee completely. I mean, it just, it wouldn't make sense because they had the optionality to do something else with the power as well. I mean, they could have completely went against anything that Nvidia stands for by becoming something completely different. Instead, they are becoming the next poster child of Nvidia, right? So, I mean, this is going a lot further than what other neoclouds have done with Nvidia. I mean, we are going to construct the flagship site for NVIDIA DSX reference architecture. So AI factories is the buzzword that they want to brainwash people with. If we are going to be the blueprint for the first AI factory for Vera Rubens, I don't think there is going to be a commercial pressure from Jensen. I think if if if Jensen knows how what what the value is of Sweetwater in terms of monetization potential and they are you know giving iron their logo and brand name to attach to this geographical location and capacity I I think that it will sell itself I honestly think that Iron's biggest problem right now is going over all the term sheets that they have. If not just being in final stages with Anthropic, you know, that could also be the case. I don't think that, like I just said, Iron's, it appears that Iron's starting to build now in Sweetwater. I only have the, very low quality picture. But I guess for entertainment's sake, I will post a picture here in the nest so you guys can see that I'm not talking ****. I actually have a proof of something that is being constructed. Just give me one second. Let's see if this works. So I'll just post it in the nest now. This is part of the Sweetwater site that was not part of the render that they gave in the earnings presentation. But this is the first primary substation you can see on the left. And then this square used to be only white color. So now instead of just white, it has these black lines over it. And I know this is a low resolution picture, but they have stalled out these components of some kind all across the site in rows, which obviously immediately resemble data center rows. So This is not in line with the rendering that we saw in the earnings presentation, which looks more like one big hall, like a data hall. I will also post that picture in a nest real quick. So I just do this on the go, if you guys don't mind. And you can see that this doesn't look like what they what they shown in the presentation. So whatever they are doing in Sweetwater is ahead of any deal because Iron has not announced any deal in Sweetwater, but they are starting to show signs of some kind of construction there, which is not the type of building that they talked about for 2027 for the DSX flagship. So I don't think Nvidia is pushing Iron to do anything. I think Iron has plans of their own and they don't need Nvidia to tell them what to do. I think it's very likely that they are just, I don't know, doing like liquid cooling infrastructure in rows that fit up, that fit the, you know, the plants or whatever kind of water cooling solution they're going to implement. That's one theory of what they're doing there. Another theory is that they are actually going to build the air-cooled data centers on this part of the site, which could be for a customer that needs power really quickly. And if there is one company in the whole world that you could think about that needs power really like yesterday and a lot of it, so much that they are willing to pay a massive premium to Elon Musk for hoppers, then you can consider if you see these two pictures where the picture from the presentation is about NVDA's DSX reference architecture for 2027 200 megawatt of IT load of liquid cooled capacity in Sweetwater. And the top picture is another part of of Sweetwater that is literally grass in the rendering that we see in the presentation. So they left that piece of land entirely empty in the picture as if it doesn't exist. Well, it does exist and there is a substation there and there is also something happening there. So I don't know, this is not really answering your question. I know I'm sort of digressing, but I just wanted to say that I don't think that it makes any sense if Iron is going to deliver 200 megawatt of IT load in Sweetwater next year for NVIDIA's AI reference design. I mean, is this going to happen and they will get a customer through NVIDIA's channels or through their own? Because obviously they don't have a problem with getting customer interest, they're just extremely picky to sign one. And that's my take.

Jonathan: Yeah. Can I just add one thing, guys?

Frans Bakker: Yeah, go for.

Jonathan: It. Yeah. And thanks for the response, Frans. And I should have actually added one more thing, because when you said you don't expect Nvidia to put commercial pressure on iron. I agree with that too. I should have specified that there's two sides to the commercial pressure though. It's not just Nvidia putting commercial pressure on iron to make a deal, but it would also be... Nvidia putting commercial pressure on, say, someone like Anthropic to make a deal with Iron because it's a two-sided coin for Nvidia. They have vested interests on both sides. For example, Jensen said, right, just recently, he's got big plans for Anthropic and he's trying to help them. And he's also talking about how he's referring all these customers to, you know, various NeoClouds. and stuff. And so I'm thinking more along the lines of commercial pressure, yes, but on maybe more so on anthropic to actually make a deal with iron and vice versa too. But that's, that's kind of where my mind was at. So I just wondered if you had any thoughts on that aspect?

Speaker 5: Yeah, I mean, of course, it's, it's also, It goes both ways. But when it pertains to the Nvidia and Iron relationship or partnership, I'm not entirely sure yet how it really works. It's in Iron's best interest to get their share price above $70. So the investment rise are in the money and that would give Nvidia additional, you know, direct motivation, financial motivation to deliver GPUs to them. I guess that's one way to look at it. Another way to look at it is just that Nvidia and Iron have entered into an agreement for that last five years where it is in their best interest of both these two to deliver state-of-the-art AI data centers with GPUs. And by having this partnership and investment rights tied to it, this is going to be a win-win for both of them. Obviously, NVIDIA will win more on a net margin basis, but it is mutual beneficial. And Anthropic as a third party in this, you know, picture is obviously very happy to use NVIDIA GPUs. and pay for that. And I think they will solve it somehow together. And I don't know if it's going to be a joint venture or some other kind of structure, but I don't think it fits in Iron's balance sheet to get into a gigawatts AI factory without a joint venture. Like I just think that is taking on too much of a capacity in a, in a single contract or a phased contract. But I wouldn't rule out that this rendering is a part one of a one gigawatt AI factory in Sweetwater. And if you see the picture, the green line is 200 megawatts. So that would mean the other side is another 200. So they have 400 a MW of IT load across, these two buildings. I think it would, I think Anthropic is a natural fit and I think with Nvidia now is a sort of a party in the middle. I guess it will just work itself out in some way, shape or form. I don't know yet how, but yeah. That's how I look at it now. I can't give you a definitive answer because I just don't know. But they've raised a lot of money, right? More than they really need this year. So what are they planning? You tell me.

₿itcoin ₿utcher: Anything else, Jonathan, before we move on to Dave?

Jonathan: Yeah, no, thanks, Grant, for that. I think that's good. I just maybe one other thing. And that is the last thing I'm going to say is that did you guys want to maybe touch on the the latest XAI deal with Anthropic and how that might possibly change the market dynamics of like how much compute is worth and like is it is it kind of skyrocketing or was that more of the fact that XAI could offer that compute sort of like right now in 2026 and there is a premium to be paid on that? I know There's been some threads where people have debated that. So I just wondered if France or you Bitcoin, you had any opinion on that?

₿itcoin ₿utcher: It's kind of unclear to me. The initial reaction was we're all going to be rich and they're paying through their eyes to Elon. I need more information to come to that conclusion. The famous time to power, or excuse me, time to compute comes to mind that Colossus is built, but I think it's primarily H200s. I think the documentation also suggested potentially Colossus 2 that they would have access to, but I'm not, well read enough on it, but I think personally it's kind of the timing of it. I'm suspicious of just because Elon happens to be about to IPO SpaceX, which I'm pretty sure XAI is incorporated into now via merger or. it's one combined legal entity is my understanding and will be part of the upcoming IPO. So I think Elon has a way of playing markets pretty effectively to increase his valuation. And I think this was more so like I'm looking at it more from that perspective with the upcoming IPO because increasing the, I think Anthropic's ARR, the latest stat I saw, now it's 40 some billion as opposed to the 30, like 44, I think's the most recent number I saw. So there's no, everyone knows that they need compute and they need it yesterday. I'm sure they're willing to pay a premium, but the initial number that I calculated, it was like over six or $7 an hour for 220,000 GPUs and it was bare metal and it just didn't really like make sense on the surface. So that's why it's I kind of hesitate to come to some conclusive opinion on it. But I think from XAI's standpoint, it's kind of convenient that they're disclosing that right before their IPO in the next month or two. I mean, I think they would have to via their disclosures, but it would be a pretty strategic move on Elon's part to attach his name to Anthropic any way possible right now. So, Franz, the question while you were gone was if you had any commentary on the bare metal arrangement between XAI and Anthropic. And to bring you up to speed, I just said, I don't think I have enough information personally, because I read the tweets of 220,000 GPUs, and if you apply the revenue number, it came out to something like six or $7 an hour when I initially did it. But for bare metal, it doesn't really make much sense to me that even though they need the compute right now because their current ARR is in excess of 40 billion. Now, I have a hard time believing they would pay such a high rate for that. And maybe it had to do with the additional capacity at Colossus 2, and it's unclear to me. So I have some more homework to do on that, but if you had any comments on it, Before we go to Dave, feel free to chime in.

Frans Bakker: Yeah, I have no thoughts on it as well. I thought the initial number wasn't correct for the capacity. So I think if you add some capacity of Colossus 2, you get a lower GPU hour rate. But at the same time, you know, maybe they did really manage to negotiate like a spot pricing for, because it has a 90 days, cancellation period or something, a termination period. So I guess this is not a long-term duration contract. Even though it is a long-duration contract, if it has a 90-day clause, then I guess it will be somewhere between a on-demand and a, you know, a reserve capacity contract price. So I don't know. I have no real thoughts on it. I guess they paid for compute now, which is more expensive. And then the optionality of canceling in 90 days, I think this all adds to the price. Entropic has a massive problem, a luxury problem that is, but it's not a luxury problem if they can't solve it quickly and they have to pay up for that. So, you know, I think it's a, it's an Anomaly, I don't think it's like a standard for the industry or something.

₿itcoin ₿utcher: Let's go to Dave. Dave, how are you this evening? It looks like Dave is an active. Sean, Sean, what's up?

Sean: Hey, what's going on in the airport? So if you hear some background, you kind of touched on it a little bit about the Colossus. compute deal with Anthropic and didn't know if obviously there's an Elon premium attached to it. It's pretty expensive for the compute. Majority of what I've seen is just that's because Gronk just obviously is not performing like all the other LLMs. But I didn't know if that made you like more excited or what your thoughts were on either, you know, how the market is pricing power, or if you still believe that Anthropic still has additional money, and if not, like, what would be an ideal partner for you in your eyes for Iran? Because I did open up another QX levered long position on Iran. I sound bearish, but I keep putting all my money in it. So I'm just I think it's extremely undervalued. I'm just waiting for the execution to come through. Just want to see what your opinion was on like an ideal partnership, what that would look like, like shorter term in terms of years or or that. And also your your thoughts on the fact that these Players like Amazon and Google are actually starting to build their own data centers. And if Ireland doesn't come up with a deal, you know, that was kind of like the biggest fear when they were starting. I was like, well, why won't these hyperscalers do it? And now that they're actually starting the process, how are you guys looking about?

₿itcoin ₿utcher: Okay, your last question, sorry, Franz, the last question is kind of I don't agree with the premise like Amazon and Google have always had data centers, which I think is what Franz was going to comment on. But yeah, I I just don't agree with that, Sean. It's they have to go. They've used colocation partners. Amazon signed with Cipher for Barbara Lake for. air cooled and liquid cooled capacity and certainly Google and FluidStack or FluidStack via Google have signed with Cipher as well, as well as TerraWolf. I think it's a multi-pronged solution if you're a hyperscaler, like you're trying to find as much readily available power right now. I think long term, ideally, they would bring it all in-house so that they can build it to their specs and not deal with third parties. So maybe there's some truth to that, but it's not some new strategy. Like they've used Equinix and DLR in the past for legacy data centers. So that piece, I don't necessarily agree with the framing of the question. As far as Colossus goes, if Grok's not taking off, that's not to say that AI is some failure all of a sudden that that's a Grok problem. Like clearly the growth numbers.

Sean: That's what I was referencing that was specific to Grok, not like other LLMs. I'm just saying like that's why they have compute like.

₿itcoin ₿utcher: Sure, they have access capacity that maybe he was expecting a larger growth rate and he's sitting on an idle asset and he's about to go to market via an IPO with SpaceX and he has an asset that's not producing revenue, it would be pretty foolish to not take advantage of what he's built and how much time and effort they've put into it. So, but that doesn't change. I mean, to back things up, you asked the ideal partner for Iron. There's a post that I could add to the nest or you can look up later, but Mirantis is partnering with Saturn Cloud, and in that post, it references a user experience that can circumvent using a hyperscaler. So to me, Iron has had these conversations with hypers, and the hypers are kind of, we're Microsoft, or we're Google, or we're whoever, and We don't pay that price and blah, blah, blah. And Dan's probably just like, you know what, **** this. Like, we'll make our own money and do it our own way. And I'm sure that was part of the inspiration behind the Mirantis acquisition. And I think part of it too is I think Nvidia was very influential in this. And Mark had commented that we're not giving credit to Nvidia and Iron's relationship in the future. I agree with what Mark's saying, but I would also say to Mark, like I said pretty early on after reading that post, that it's clear that Nvidia is hedging against the hyperscalers. Most notably, Google and Amazon are coming to market with TPUs and Trainium and selling them to the marketplace respectively. So they're starting to get on, you know, walk into Nvidia's territory and interfere with their GPU sales. So to the extent that you have Nvidia investing in the equity of Coreweave, Nevius, and now Iron, I think that sets up for a world where it's Nvidia and the Neo Clouds versus the hyperscalers. And Iron specifically helps Nvidia the most. because they have the capacity without dealing with other co-location providers, where once they hit their stride with the DSX architecture, they can start producing those token factories very quickly and start selling more GPUs, ideally. So I think that kind of hits on most of your questions. But the ideal partner, though, for Iron right now is an enterprise that needs some orchestration help for higher margins, who's not going to ***** about pricing like a hyperscaler. I mean, that's ultimately what we're trying to do is maximize the revenue per megawatt. And I think the hypers are just upset that some Aussies that might talk a little funnier than they're used to beat them to the punch with all this power that they amassed six, seven years ago. And they're not willing to pay market rate right now because they feel like their reputation merits some sort of discount when they're planning sucked and they're they need to pay up now. So but yeah, Dan's kind of like either pay or don't, but we're going to get our price and. his actions consistent with that right now. But Bronze probably has something to add.

Sean: And then just like on the term of that, just wanted to make sure like I got that out. So for the term, do you, would you, in your ideal world, would it be like a month to month payment like anthropic with SpaceX or would he still get like a multi mirror dual would be better or no difference?

₿itcoin ₿utcher: There's actually Execution risk. I think what you're referring to is spot rates on demand versus contracted deals that are have terms. I personally err on the side of it's a really large undertaking that they can't get GPU financing without contracts. They haven't. No, no one in their right mind is going to, unless they have a signed deal, agree to lend them that money or it's not going to be for the same interest rate as if, you know, if they're just buying on spec and saying, wow, people are willing to pay 50% to 100% more for on demand. But that also leaves them subject to even though right now there's no idle GPUs, and in theory in the future, if that on-demand wasn't being used, they could have a problem. So I personally think from a business standpoint, their job is to de-risk their investments, and the best way to do so with the infrastructure is to lock in deals that just like they did with the Microsoft deal. Everyone can ***** about the price of it, but at the end of five years, they have a data center that's free and clear and they have GB 300s that are also free and clear. So whether they choose to do those on demand or they could even do colo on that site, even though I don't find that likely, the point is they have the choice. So I would say contracts. probably two to three years given this pricing right now, they could pay down their GPUs and maybe they extend it to four or five so that they could pay for the data center itself. But I lean towards contracts because I think that's easier to sell to an institution than, hey, we're going to speculate on the market pricing, which there's no guarantee in the next few years. I would feel better as a shareholder knowing that they de-risk their infrastructure before they get more aggressive. I mean, they've hinted at the ability to blend their margin upward by having on-demand pricing in the future, but I don't, that might be 10 to 20% of their portfolio. I don't think it would ever be more than 30 or 40%. That's just me speculating.

Frans Bakker: I wanted to just comment on the last part of Sean's first question. I think you shouldn't look at it like that. You have to consider that there will be a shortfall of data centers regardless of who's building it and how many they will get online. Even if all the hyperscalers start building their own data centers, there will still be a shortfall of multiple gigawatts by 2029, 2030. So I think It is in Ireland's best interest if all the hyperscalers build as much data centers themselves to expand the market of AI compute to grow adoption. And ultimately, this will increase competition for their existing and future clients, which will make those clients buy more compute to compete. So this is not a, how to say, This is not a business where, there is a limit to how much compute we want online or how much data centers we want built. We want as much as possible across the board because it will, grow this AI market And that's ultimately what we want. We don't want this to be a three or five year cycle. It has to be, you know, a real life changing event where ultimately the ultimate goal is ASI, right? So, I mean, I don't think you have to worry about hyperscalers building data centers, which means Iron has no customers. I mean, that is basically at this point in time with all the information that we have is honestly is this bad take. So just to point on that. And everything else you said, who is the perfect customer for Iron? I think the perfect customer for Iron is any customer. that is willing to pay down and is willing to commit for longer durations and is willing to scale and able to scale with Iron across their portfolio. So these are not just only frontier labs or just enterprises or just hyper scales. It's all of them at the same time across multiple sites. We will have a multi-tenancy site. We have single tenancy data center sites. It's going to be a mix of all of it. And I think the optionality that Iron now has with Mirantis included is going to be the biggest upside in our story. So I will say I would rather like Iron to sign Anthropic now, then another deal with Fireworks AI, of course, because that will be a big headline. It will do really well for the stock. It's obviously a very creditworthy counterparty if they keep raising money the way they do. So, but I don't want them to stop there. I want them to do it all across their sites and all the kind of customer segments that we are tapping into. including, but not limited to, Anthropic and other hyperscalers. So I still somehow have hopes that Iron will get Google on board as well. That sounds far-fetched with all we know right now, but I know that by the end of last year, Google was certainly interested in Iron's capacity. and I don't know if anything changed. We know that they have been sort of, they've been making a partnership with Blackstone, right? And it looks like it's going around Nvidia, so it doesn't sound, it doesn't seem like Iron would want to be dealing with that, but I think it would be really good for Iron to show that they are still chip agnostic and they're not only going to work with Nvidia going forward.

Sean: So I am really first mover advantage the same way that like, you know, how ChatGPT came on and like a lot of people, even though like VOD is, you know, superior in many ways, they still go back to like the original source. Was that the same way with the The NeoClouds, or is it like if someone didn't like the service of Nebius or TerraWealth, they could just switch over to Iron and it would be easy, or is it like a super tough thing to unwind?

Frans Bakker: This is, you know, the stickiness of these of compute and platforms. That is basically the... the thing that Nibius investors are usually very, strong about that once you work with Nibius, it's too expensive or too difficult or both of those to move to another provider. I don't know if that is really true because by the same token, there has there have been a lot of, there are also AI cloud providers and Frontier Labs that are claiming that they can just stream their data from data center to data center and it doesn't really matter. They are cloud provider agnostic. And if you look at Together AI, Fireworks AI, those guys all use like, I don't know, 20 different providers of computes. Basically, they go with anyone who has computes. So there is no stickiness whatsoever because they are not captured into an ecosystem or a platform where they cannot migrate from. I think, the kind of customers that Iron currently works with are very fluid. And I think that now with Mirantis, obviously they're going to try to become more sticky. But, first mover advantage for AI cloud users, I don't think that is a real thing because you cannot sell more than the capacity that you have, right? So let's say Nebius has a Uber or whoever they have, but if they run out of megawatts, you know, where will where will Lightspeed go, right? I mean, that's also one of Nibius' customers, right? If they need more megawatts, but they are locked up in the ecosystem of Nibius, what's going to happen? Are they going to do a part somewhere else, or are they going to migrate to the cloud provider with the most capacity? So I think this is something that conceptually people should start to think about. I have metaphorically talked about this before as a car with gas. Nibius has the faster car, but iron has more gas. So I think we are going to catch up with the car that's going to slow down next year. and our car will keep on going. it's very simplified, but this is not just about share price. this is really, you should think about customers that come talk with Iron and say, look, I need 100 MW this year. And then Iron say, yeah, well, sorry, I don't have it, but I have 200 MW next year. And then they'll say, yeah, but okay, I can take 200 next year, but I really need 400 in 2038. And if you have these kind of conversations, would you rather be Iron who has a grid connected capacity of 5 gigawatts to roll out across multiple sites in the coming five years? Or would you rather be Nebius who has just paid for 250 megawatt of bloom fuel cells to fix an issue with ship engines in a community in a residential area. So, this is just a whole different game here. And I think the first movers have locked in smaller capacities, smaller clients, or bigger big clients that need more capacity in the coming years. In the coming three to five years, it's all going to be about big amounts of power. massive capacities, big sites. And I know that, which player I want to bet on, I want to bet on the company that has the power and not the ones that are trying to scramble it left and right with makeshift solutions and fuel cells. So I don't want to put fuel again on this feud or fight or a fire that we have between these companies. But it just makes a lot of sense if you think about it. If you want to secure your business model, you are modeling out your capacity needs and your compute needs for the coming years, not just for next year, right? If Aaron's already talking about what they're going to build in 2028, You can really consider that at Anthropic right now, do you really think they're just concerned about 2026 and 2027? They will not get into this issue again. They're going to over buy capacity. You know, they are going to sign deals for more than they think they need because they will probably have some clauses in there that they could potentially shift out of or get out of. But they're not going to under allocate their growth plans. I mean, if you see how the market is rewarding them for growth, they're going to make sure that they're going to get, you know, grow into this valuation. So I think having the most fuel in your tank is going to be the differentiator. So There is just no way that, and I haven't even talked about Jensen's Paradox, Jevon's Paradox yet. There's just no way that this is going to be a situation where demand runs out or capacity is somehow, you know, available in the year itself. I think everything is going to be you know, get bigger, capacity is going to expand, contracts are going to get bigger. And yeah, I think the late mover advantage is much more valuable than the first mover's advantage in terms of securing GPUs and signing contracts. But when it comes to securing power, I think first mover advantage will be the absolute differentiator that's going to last until the far into the next decade.

Sean: Are acquisitions of Iran totally stupid to even think about by any of these like mega hyperscalers?

Frans Bakker: Well, You may have seen this comment from the CEO of Data One where he said Nibis is going to go to $300 a share price and they're going to buy Iron. I think if people would have any time on their hands, they should read up on Class B shares that Iron has, well, that Dan Roberts and Will Roberts have. To take over Iron is pretty difficult without their consent. So that counts for hyperscalers as well. I'm sure that they had offers, you know, I'm sure Nvidia wants to buy Iron, but you know, I don't think it's going to happen anytime soon.

₿itcoin ₿utcher: Let's go to Bryce McNally money. Sean, I just I'm going to wrap this up and after Bryce, if that's all right, because I'm on the East Coast here in

Sean: one portion of midnight for your time.

₿itcoin ₿utcher: All right. Thanks, Sean. Bryce, how are you?

Sean: Hey, buddy. Good. How are you doing?

₿itcoin ₿utcher: Doing well. What's on your mind?

Sean: A couple of things, buddy. Number one, I saw that. video of your butcher shop. It looks incredible. So I just wanted to say that's great. I wanted to ask a couple kind of rapid action questions here. You guys can answer however you want. Childress, with Microsoft, do you think we haven't seen a big expansion there, more so on the iron side or the Microsoft side? That's to anyone.

Frans Bakker: Sorry, can you repeat the question?

Sean: Yeah, so with Childress, with Iron, in terms of the Microsoft deal, have we not seen a large scale increase or expansion of that contract? Do you think more so from Iron's perspective or from Microsoft? And do you think that will be a single client site?

Frans Bakker: No, it's not a single client site because NVIDIA is going to be there as well. That's the first other client we already know is going to be in Childress, but I think the reason why Horizon 5 and 6 hasn't been signed yet is because they are probably waiting for the delivery of that first building or maybe of the first cluster. So I talked about this a little bit earlier in the space where I think that Iron said they're going to deliver the super cluster of 50 megawatts of Horizon 1 in Q3, which would be by the end of September. But I think it's going to It's more, I think they should be able to bring it online in August by the latest. So that could be the start sign of them signing Horizon 5 and 6 with Microsoft as well. But then there is also the other thing is they are going to demolish block four and five of the mining buildings, and that's going to start already next month. So, Would you really take down your mining buildings if you don't have a signature of Microsoft yet, right? So who is the reason of no signature? It's probably a bit of both. I don't think it's like Microsoft holding off, because if Microsoft would be holding off, Iron would not announce, you guys, we're going to build Horizon 5 or 6, right? Because they still need to convince Microsoft. So I think it's rather iron who's waiting for better dual economics by first delivering Horizon One and then signing rather than Microsoft holding off. I think Microsoft will just say like, oh, come on, let's sign now before the GPU prices go up further.

Sean: Sure, yeah. Okay, that's fair. And then I got a couple more quick ones for you. The ATM, where do you think they're at on the 6 billion and what price do you hope they're selling at or above?

₿itcoin ₿utcher: I think as of earnings, as of March 31st, that was just over 1 billion. And maybe they, through the increased volume, Post-convertible, I could see them selling an additional billion more recently. I would hope that it's, I think part of it is the exercise price on the convertible was 55, so I could see that they were probably between 50 and $60 selling quite a bit of shares to the market because there was institutional demand. and post announcing Nvidia, I think there was, they knew that they would have the volume to support it. So I would guess that anywhere from 500 million to an additional billion dollars just off the top of my head. But yeah, obviously we'd like it to be higher, but I would say the convertible actually the timing, it was ****** that it was the Monday after the deal was announced, but then we went as low as 46 this past week before recovering. So for us to lock in the exercise price, you know, there's a 32 or 33% premium on top of the 55. So it ends up being in the 70s. It kind of sucks that there's the additional volatility in the profit and loss noise that comes from the markup and markdown of the securities on the income statement. And then Franz, unrelated Franz had mentioned the impairment charges related to mining. So there's still a lot of noise in their financials. And I personally look forward to the day when This just gets to a company that, you know, I said at the end of next year could be generating not necessarily you'll see 12 billion in revenue, but they will have GPUs installed for 2028. That would produce 12 billion in revenue. And ideally at that point, that's a company that's producing the internal cash flow to fuel that flywheel that we've been talking about for so long that doesn't exist yet, but also we'll have infrastructure in place where we can lever up as opposed to going back to convertibles. I mean, there'll still be convertibles in the ATM. I just would like to rely on it less in the future.

Frans Bakker: I think we should also consider the fact that they just raised $3 billion of debt. And they only announced a billion dollars of equity. So I think they're going to raise equity to counter the debt that they are taking on because they are extremely conservative in their balance sheet. And I think they are still looking to add more debt for GPUs that they're going to receive and front running things for Sweetwater and Horizon 5 to 6. So I think they're going to just Just consider the $6 billion ATM tapped in 2026. Just get it out of your system, just accept it as a reality. Don't look at a share price, just look at a share count and then look at the market cap and then you can consider what you think is fair. But they are building something that is going to provide for shareholder returns. in the coming 5 to 10 years, right? This is not something that is going to pay itself back in 2026. We are front running pain through convertible notes, volatility and dilution, but it's not going to be a permanent dilution like, you know, some of these accounts like Serenity are trying to depict that this is a never ending dilution nightmare where the share price is never going to go up. I mean, The reason why we are in the 50s and not in the 70s is not because of dilution. It's because this stock is in control by market makers and hedge funds. And that has nothing to do with Dan Roberts stepping the ATM. Yes, they are feeding on that a little bit, on those sentiments. But the people that are responsible for the volatility of the stock, you should look in the mirror because we are all part of this. We are the degenerates that are buying 5,000 contracts for one date in the future, right? The market makers see the options tape and they're just going to drain us completely. You know, we are all responsible for this. And, you know, there are not that many degenerates in Nibius that are doing these options the way that people are doing on iron, right? So I think, yeah, I mean, sorry if anyone felt attacked by this, but I am just as much responsible for this too, right? I've been losing a lot of money with selling puts that I thought that we will never go as low as that by next week, right? And what do you know, we went twice as low as that, right? It's just, I think dilution is one thing, but it is not our biggest enemy here. Part of I think the biggest enemy we have is our own expectations that we should manage better.

₿itcoin ₿utcher: Hey, Bryce, before your next question, what we're not also including is the 50,000 GPUs that they had purchased in March, they never completed their financing for, and that was another event we were anticipating. And as a reminder, that was 50,000 GPUs at 70,000 apiece. That's three and a half billion dollars. So if you consider like if you took those 50,000 GPUs and said we think we could generate $4 an hour, I mean, their guidance is 1.3 billion. Maybe we should do that to be conservative. If we'll we'll do that for the sake of argument, but 1.3 billion that generates conservatively 1 billion in cash a year. But as it as it pertains to a prepayment, let's say you got a five year contract on those and you got one year prepaid, so roughly a billion dollars. they probably can get prepayments between 500 and 1 billion on that. Let's just say 1 billion for easier math. And then that leaves you have out of three and a half billion, 1 billion was prepaid. That's still two and a half that you have to finance. Maybe you're only financing 2 billion of it, though. So potentially that would be $500 million that they might use the ATM on as well. So there's a lot of Until all of this stuff is plugged in and producing cash flow and scale, we just have to hope for better days and believe in the plan. And some of that's going to be that ATM. So yeah, it's not ideal, but when you realize what they're building towards, I think it's worth it. And over time, the value will be self-evident. But what other questions did you have? Bryce, are you still there? I think we lost Bryce. Bryce, thanks for coming up. We enjoy your program with Anthony. Franz, any closing thoughts? Otherwise, I'm going to wrap this up because we're pushing midnight on the East Coast and I got to work tomorrow.

Frans Bakker: No, I mean, I hope we'll have a good week. I think we may be positively surprised at some point in the near future. There is something that I know they are working on that I cannot talk about because it's part of someone else's NDA. So I expect something to be announced in the coming two weeks, which is not a deal, but something that could be, good for the market. So I hope everyone enjoyed this space. I would really be happy if we could just get above 60 this week. The sentiment seems good. I hope the macro will cooperate. if this war can finally, be put to an end of some source, that would be really good for the market and for the world. So let's see. Thanks for hosting. Next week I won't be here. So maybe we can catch up on this space later this week that Small Cap Sniper usually hosts. So thanks for hosting and have a good night, everyone.

₿itcoin ₿utcher: All right, thanks. And yeah, Bryce, I'll give you if you had one, any other question for Franz and I before I sign off last call, for some reason you got knocked off stage there for a sec.

Sean: Yeah, I don't mean to keep everyone late. I was just wondering, do you think there's any value in a colo deal at this point for Iron or do you think they're past that point? Is there any benefit in doing a colo for diversification of revenue or do you think CSP is the way to go full value stack and they've proven they can do it?

₿itcoin ₿utcher: I think bare metal is the new colocation, how they plan to de-risk it. My thing is it gets back to if they can do a quote unquote ****** bare metal deal that pays for the GPUs and the infrastructure. They would rather have the option at year six to do what they want as opposed to lock up their asset for 10 to 20 years, depending on the hyper. So I personally rule Colo out at this point in time.

Speaker 5: I will give one final comment on co-location is I think no traditional co-location. The only thing I could see happening is a version of co-location that includes Iron somehow retaining or gaining ownership of the GPUs after the initial contract. And that would really only work with Google for TPUs or NVIDIA for GPUs. So that would say that NVIDIA would refinance the GPUs and keep the ownership until the end of the contract where Iron has the data centers and sort of gets the GPUs for the service they provided or something like that. But this is very short version. I think there is a way that this could work. And I think that it's not completely impossible that they are trying to figure out some kind of agreement for Sweetwater where Iron would not need to pay for all the GPUs in advance or after shipment. Maybe it will be like a semi-hybrid co-location kind of agreement with Nvidia being involved. But that's just a thought. It's not real based on anything else. So that's all for me and co-location.

Sean: Gotcha. You guys are awesome. Thanks so much. We'll see you next week.

₿itcoin ₿utcher: Yeah, we'll check you out this week, you and Anthony on the show. And thanks for coming by, Bryce. And yeah, I'm with friends. I would like to see 60 become the new floor, but we also didn't think we'd see 46 this past week. So I'm prepared for the worst, but hopeful for the best here. And appreciate everyone coming out. apologize for the tech issues, whether it's me or X or whatever. You guys still had to experience it, so I'll take ownership of that. But we will be back next week without friends and I promise not to ruin the Sunday night space without your friends, but I'm sure some brave souls will join me on stage and we'll keep things busy here, but we'll see if they give us something more to talk about this week or if it's going to be quiet. But to your point, I would like to put the Iran conflict behind us and just simply focus on not what our president is about to say or what he's not going to do and just look at companies for what they are and this AI macro trend. So thanks again, everyone. We'll see you maybe at small cap. I don't know if he's hosting this week on Wednesday, but otherwise I should be hosting next Sunday. Have a great week, everyone.