$iren RAISE Your Expectations 7.12.26

Hosted by @Bitcoin Butcher · 2026-07-12 · Tags: IREN

TLDR

Participants argued that IREN is evolving into a vertically integrated AI infrastructure and compute provider positioned to benefit from rising token demand, open-source models, data sovereignty, and scarce powered data-center capacity. They were broadly bullish on its NVIDIA relationship, construction pipeline, customer demand, and ability to pass higher costs through, while criticizing weak investor communication, governance, dilution, financing uncertainty, and execution delays.

Speakers

Notable quotes

Transcript

₿itcoin ₿utcher: Two minutes after 9 o'clock here on the East Coast. Welcome everyone. Franz, are you with me? Good morning, I am. Well, so I got my co-host Franz and good to be back after a one-week hiatus and I appreciate everyone and their understanding and I won't dig into what's behind us now, but let's look forward. So Franz, if it's cool with you, I had a few initial thoughts on what's going on in the market this week, and then we can kind of dig into a few topics that I sent to you earlier. Sound good? All right, so it's Sunday night here in the US. And tomorrow we'll have the start of another trade week. Things I'm looking at this week as an active market participant, I posted a few of them today to kind of get people starting to mirror kind of what I'm looking at. And certainly anyone's free to come up and bring up other things they're looking at. But most specifically, You have a lot of earnings this week. Xcap isn't with us tonight, but he's looking specifically at the bank earnings because those bank earnings usually have good commentary on the macro economy and whether the country itself is headed into a more, you know, economic growth environment or a slowdown and potential recession. So it'll be good feedback from those banks such as, I believe, Chase being one of them. So someone like Jamie Dimon commenting on the macro is always one new data point that can be taken into consideration. I'm specifically looking at, on Tuesday, you have the Consumer Price Index, CPI, being reported, and that, I believe, in the post I had earlier today is expected to come in at either 3.7% or 3.8%. And I think the key driver of that, even though food and energy prices aren't captured in the CPI, the fact is that the price of oil is still the most important commodity price in the world, as it's a key input in a lot of those goods and services that are measured by CPI. Previously, it was as high as, I believe, $120 a barrel or definitely 110 a barrel. And now it's settled in, despite all the rhetoric back and forth between the administration and the government of Iran, it's still trading roughly above $70 a barrel. So from peak, let's say it was 110, that 40 is still 30, 40% down. So I would expect to see some relief and the inflation rate. That's not to say that things have gotten cheaper, but if the rate at which they're increasing instead of it being over 4% is slowing down and cooling down to 3.7%, yes, pricing is still increasing. But again, this is such a confidence and narrative driven market where if it appears that the growth rate is deaccelerating and slowing down, that that is more support to not implement rate hikes later in the year, which would impact economic growth. The other one that I pointed out, I incorrectly identified the ticker as TSMC. It's TSM Taiwan Semiconductor, as well as ASML. Those are two early reporters and then I also usually look at applied digital even though it's colocation as opposed to a neo Cloud but gives you a sense of where the AI Market is headed um most specifically though TSM is they produce 100% of the GPUs that Nvidia eventually sells so having an idea of the producer of NVIDIA chips will give you a good idea. You know, if they're doing well, it's fair to say that NVIDIA is doing well. And similarly, the machinery orders to ASML that eventually are used by TSM, you know, that specialized equipment is necessary to build those chips. So it's all just one big puzzle and you're connecting the pieces and you're trying to figure out those interconnective relationships and you start to, over the past few quarters, you kind of see the patterns as they arise. So those are just, I gave you 3 examples, Apply Digital, TSM and ASML are usually I look at those earnings calls and listen to their management to get a better feel for The future of AI and whether we're still in the right place, but I'm willing to bet in the way I'm positioned is that they will support continuation of the trend that is AI. I've posted in the nest a few different articles and. Certainly we'll hover back and forth, but I think the best way to frame it is I watched a podcast, the all in podcast today. Those guys do a pretty good job of speaking to AI, given their attack backgrounds and kind of laying things up. So just really briefly and I encourage you guys to watch the episode. It's in the nest, but some of the key takeaways that I took from it were the following. And this is just in sequential order, so forgive me, but Anthropic, we've been talking about, I believe at the end of 2024, maybe it was 2025, their ARR, their annual recurring revenue was a run rate of $1 billion. Well, it's projected that by the end of this year, 2026, that that will be $100 billion, which is just astonishing, the growth rate of that company. They're looking to IPO within the next six to nine months along with OpenAI and should attract a multi-trillion dollar valuation, especially coming off the success of the SpaceX IPO. Key thing that I wanted to touch on from that All-In podcast was Chamath was speaking with his CTO and there's a clip circulating where token costs has gone up for their company. I don't know which portfolio company it is, but it's doubling every 45 days. Yet they're only seeing incremental productivity gains of roughly 5%. So they're trying to look at that more closely. And I think this is a really good primer for what we're going to discuss tonight with Iron. And I had posted another article of how Iron potentially sits at the intersection of these solutions. And then I also have an article that was written by Satya Nadella, the CEO of Microsoft, as well as Xcap had a nice thesis that he published regarding Nvidia, Nemotron, Palantir, and how those that collaboration with Iron, what that might look like and how open source models will be in competition with these frontier models. So continuing with the theme from Chamath, the productivity gains that don't appear to be relevant right now, but that's kind of the conversation that you're going to see with the token usage increasing at all of these respective companies. how do you justify exploding token growth costs and what can you do to manage that? So I had spoken to Anthropic and OpenAI IPO-ing in the next six to five months, or excuse me, six to nine months. And then some of the other notes here. Again, the idea that foreign governments, Chamath was on a committee with Marc Benioff and a few other high-level tech executives. And all of these foreign governments have the same goal. Rather than using closed-ended models such as OpenAI or Anthropic, there's going to be more of a push towards open-source solutions, even if they're only 90% to 95% as effective. The idea is, and this is what Nadella was speaking to in his article, is there's a opportunity or a, I should say a risk that if I have company X and I seek out the help of OpenAI or Anthropic, they offer me some cheap tokens to be able to use their models. The problem is in some sense, that cheap token is an incentive to have my company in theory use their model. And by doing so, their engineers have access to my proprietary data and they can use that. And potentially, if the market is large enough, use that proprietary data against me in the future and come up with potential you know, app killer or whatever you wanted to speak to. So it's it's starting to reveal itself at a very high level. And you're seeing more of the pushback from the Alex Karps of the world, Tia Nadella, Jensen, and they all are economic actors in their own right, and they're self-interested, and they want to see their own usage go up. So it makes sense to speak to the counterpoints of their competitors in, you know, with open AI and anthropic. But I think the what's actually going to play out here is more there's a gentleman, Andy, I forgot his last name, but he's the CTO at DoorDash. And you're going to see more of these hybrid approaches, I think, at the enterprise level, where the hardest tasks are dedicated to those frontier models. And then they're going to use open source models and private forked for the to decrease their token costs on the easier, more repetitive tasks. So those were just some of my initial thoughts. Franz, what do you see in the market this week? Do you have any comments on what I brought up? But I thought those were some good talking points to kick off this conversation.

Frans Bakker: Yeah, so I have also listened to the All In Podcast and Yeah, it was really some some very relevant points for, you know, the neo clouds in particular. Of course, I saw some things that I thought were bullish for iron. Other than that, I have not really been looking at the broader market this these last few days. So I could comment on the on the podcast, but I have not really been focused on, you know, the broader economic markets this weekend. But yeah, so I think what Shamas basically was saying is that they are not really seeing a short-term return on their spend, on their token spend. But regardless, they're not gonna back down. And I think that this is another way of looking at how these companies out there are utilizing their token spend budget. And I think the big, it's a very, there's just one simple approach. Either you are using AI and tokens or you're not. And as soon as you are, you will find out that you will need more of it. And that will outpace your return in short order. But You know, so I think that is very interesting to find out that at some point, you know, efficiency and ROI is going to become more relevant again. And at that point, you will see that, you know, these companies want to spread out their token use and their tasks across multiple models, notably, give the bulk of the simple workload to the open weights and open source models and the high quality tasks to the proprietary frontier LLMs. And I think the example that he gave was you don't want to outsource the work of a top tier software engineer to an open weight model that you know, it could like run for like 90 minutes and then run into some kind of error and then you would be, you know, you would lose all those tokens and that time. Even though those tokens would still be relatively cheaper perhaps than the proprietary model, but if you would run them in a, you know, if you would use the top Claude model, for example, you would most likely not run into any bug or you know, so the point being here is the very extremely expensive tasks are still better outsourced, you know, the very important tasks are still better outsourced to the proprietary models. But I think what he was arguing is that the bulk of the more simpler tasks and workload is gonna shift towards cheaper tokens and that those are obviously more likely to come from the open source, open weight area. So yeah, so that's basically what I got from it. I mean, I listened to the podcast on a plane, so it's not like I was super focused, but yeah, I can definitely comment on some things they said. Other than that, it's mostly been, you know, I've been very, I've been analyzing all these race keynotes and panels and of course I've talked to people that were there on the ground. So I could probably talk more on the event and how that relates to Iron and maybe some Iron internal.

₿itcoin ₿utcher: Things. Hey Franz, sorry to interrupt. I'm getting multiple comments from people that I don't know if it's your mic or maybe you need to rejoin, but we're having trouble hearing you. I can hear you well on my end, but if you want to take a second to see if you can adjust your mic or rejoin, we can jump into raise and speak to it a little. But just for whatever reason, I'm not sure if it has to do with your laptop or if you're mobile or what's going on there. But just I've had five or six people in the comments ask about your volume. So if you can take a look at that, I would appreciate it.

Frans Bakker: I think I had a driver update from GeForce, so I'll look into that. You just keep going.

₿itcoin ₿utcher: OK, so one other while Franz is taking a look. I don't know if Mark's in here tonight, but Mark has a pretty good understanding of the design architecture, but the key theme that I see going on here is I have a tweet that. I don't want to read someone's work word for word, but it's Gavin Baker and I think what's going on here. is ultimately there's X amount of dollars being spent on tokens. And right now, the majority of the economic benefit is going to the Frontier Labs. So Baker's point is, and we've touched on this, is if you're Nvidia and you have Nemotron and you have someone like Palantir on your side that can create a software application on top of it, And if you have the infrastructure of Iron, now you have a full stack solution that does two things. Most importantly, it protects the enterprise from giving its proprietary data a way to OpenAI and Anthropic. But the second piece is, this plays into Jevon's paradox, is if you're able to lower the token cost which is the whole point of the DGX hardware and the DSX architecture that's being used at Sweetwater. That is the thesis. And that's the reason why, despite the frustrations with some of the scaling efforts that we've had with ARR and the delays with Horizon, they're playing for such a big prize that has such a reward at the end of the tunnel that to not have any exposure to this theme just doesn't make sense to me, especially when Iron is trading at a $15 billion market cap right now. And that's not investment advice, but that's just, you have a call option on Nvidia being able to execute architecture and Nemotron that have been designed by the best engineers in the world. And it's only able to be get carried out via Sweetwater. So I've been a vocal critic, but I'm also, I just, I try and keep it objective for everyone because I think there's at times where we're accused of being non-objective and not pointing out the flaws of the company. So Franz and I just try and keep it balanced, but make no mistake, like they're playing for such a bigger prize. And it's not to be dismissive of the colo players. Like someone like Patrick Flurry was unbelievable this week, like watched a great episode of McNally Money with Anthony Power and just simply gave a masterclass on colo economics and what they're trying to achieve as an organization at Wolf or someone like Tyler Pate, like there's a ton of great thought leadership in this industry. And my point is like it doesn't like there's a there you can choose multiple flavors and take different paths. But with iron specifically as a reminder, all of this comes back to iron like when Satya Nadella. And then the guys from all in are talking about not only data sovereignty, but the cost of token production. This parallels everything Iron accomplished in Bitcoin mining. And I'll be the first to admit that there have been challenges and it hasn't gone perfectly. But there's a very good chance that this week or next week, we hear an announcement on delivery of Horizon One and the subsequent completion of horizon two through four at a way faster build rate. I mean, that's momentum that any bearer right now that's short, they're just chasing momentum to the downside. And at some point, those shorts have to buy this. And right now, I feel comfortable. I can't predict the bottom, but if I'm looking at the asset value and the revenue opportunity that's ahead of us, now is not the time. And my brief departure and re-entry was more around something governance and not the actual business opportunity in itself. So I want people to understand that I never lost faith in what IRIN was trying to accomplish. And in this past week, all I hear is open source this, open source that, and this isn't going away. And Iron's gonna be at the epicenter of it, and they will have the largest token factory in the world at Sweetwater. And that doesn't even talk about the accomplishments that they're gonna have with Microsoft at Horizon one through four. And there's a good chance that Horizon five and six gets signed in the next few months. And that extension would reflect two things. One, that Microsoft is happy with their work, but two, that the demand is still there, and then most likely at higher economics. So kind of a mini-ran or deviation from... But I think just like sometimes you have to take a step back and ask yourself why I'm in this. And I had to do that last weekend, but I know what Iron's playing for and I believe what they're playing for right now. And it is the token factory. And it is my thesis that Iron, with the help of NVIDIA and with the help of DSX Architecture and DGX Hardware, and with the help of someone like Michael Dell, will create the most efficient token factory in the world. that will produce the most intelligence in the world, and that will trickle to enterprises in the form of revenue. That intelligence will lead to more productivity, which will lead to more profitable businesses, and that will self-reinforce itself. So Franz, anything to add to that or did you want to dive in to raise? I hope your mic is situated now.

Frans Bakker: How am I sounding now? I haven't amplified my input, so this should be better.

₿itcoin ₿utcher: I think it's better, yeah.

Frans Bakker: Okay, otherwise I'll just, there's one more thing I can do, but then I'll have to switch to a phone. So yeah, so I think, you know, the topic you're covering is mostly, you know, the thing that was talked about at race as well. So it's one giant overlap, obviously, because Iron has been talking a lot at race and it has everything to do with what you just said. So yeah, I think the ultimate goal of Iron is to cut out as much middleman as they can. I think that is just the general concept of the, I think I asked Mike Alfred a while back, you know, what is the what is the colonization on Mars for Iron? as in a reference to how SpaceX wants to, ultimately start a civilization on Mars or Elon Musk, at least. So, what is the end goal for Iron? And I think a big, a very good way to look at it, for the time being is that Iron wants to obviously monetize their power portfolio, but while doing so, they want to, deliver as much as possible to the end customer. And on the way there, they want to establish meaningful relationships with these end customers to be able to help them grow. So in a way, it's like nurturing your end users help them scale and grow so they will become the mini hyperscalers of the future, basically, or at least investment grade customers, the investment grade customers of the future. So, you know, it sounds like a lot of buzzwords when Aaron's talking about, yeah, we're going to target enterprises now, but there are a lot of tears in these companies that are now starting to want to operate GPUs or at least start to rent out GPUs for their own compute use. But Iron is very selective in picking their customers. And I know that everyone has been talking about when is the next hyperscaler customer going to arrive? When are they going to sign another Microsoft deal or a Meta or a Google? But I think it is very important to understand that it's the customer mix that is actually the reason why Iron is so extremely late in signing. And late, it doesn't mean a late revenue recognition. It just means that the announcement to the market is going to be last minute or none at all, depending on how big the customer is, obviously. But, you know, I think that you should look at it like this. And this is basically coming straight from Iron C-suite at Raise. They will always like to have some part of their portfolio dedicated to hyperscalers. So they will always be part of the customer mix. But going forward, you will see increasingly more, you know, enterprises and smaller AI native companies and frontier companies. But the point is that the hyperscalers in particular All they are doing is locking away capacity in the future, future capacity, and then they turn around and they lease it back to the enterprises. So a hyperscaler is in many ways for a cloud provider like Iron or a compute provider, no matter what you want to call it, a hyperscaler is nothing more than a middleman. And it sounds kind of weird, but it is actually how it's viewed internally. from Iron. So I know that last year we were anticipating who's Iron's customer and then Oracle came up and then everyone was like, no, Oracle is a middleman, but technically all these companies are middlemen. I mean, Microsoft is just going to run these Horizon compute hours for Azure and guess what? That's just going to be used by enterprises at some point. So You know, I think that the reason Iron went with Microsoft is because Microsoft doesn't have their own LLM. And, you know, at least they are really, you know, supporting or promoting adoption in enterprise AI use. But I don't think they're going to feed them indefinitely with their massive power portfolio because they are just competing with themselves in the long run. And a big reason why Iron is very proud of their Microsoft deal is because for a company in hyper-growth mode like Iron with a relatively small balance list, it is very impressive to have 95 or 96% of the CapEx for the Microsoft deal financed entirely. So They are, this is actually something they're very proud of. So, there are some analysts on X that are saying, it's a negative NPV. It's not, it's a loss making project. You know, it's not going to add meaningfully to the income or at all. You know, there's people that said it's capital destruction. there are all kinds of metrics you can apply to it and it is all terrible. Aaron is just looking at it like we have a AAA hyperscaler customer where we have 95 or 96% of the CapEx financed through, you know, low interest or high, you know, like investment grade debt. So To them, it's not viewed as that. It's an anchor tenant that helps them learn better and faster how to develop a top-tier liquid-cooled data center for the next iteration of GPU, which is the Vera Rubin. Because ultimately, yes, it's going to be used for GB300s, but During this construction phase and this whole design phase of the last, well, now it's like 12 months, we're still have six months to go or five months to go. This is 1 massive R&D project at the same time. So Aaron is learning how to optimally build Vera Rubin data centers. And they're going to start doing that in Childress for Horizon 5 and 6. It's going to be for Vera Rubin. And they are applying all the best practices of the first data centers, the first horizons. But at the same time, they can use all these methods and things they learned in Sweetwater as well. And this is exactly what's going to happen. So, you know, the market is very, you know, doesn't believe that Iron can actually build these data centers People are pointing at the revenue. Where is the revenue? If they are so, you know, if we have been drawing these parallels with Bitcoin mining and how they completely dominated the Bitcoin mining industry in less than two years. And now people are saying, well, that's very different from AI data centers. And as it looks now, Iron can't do that. So the parallels can no longer be drawn and it's so different. And they are compare, you know, they're comparing Iron now to data center providers that have built data centers before for AI, but not to this, not at this scale or not at this complexity level. So just to, you know, if you compare Iron to Nebius or to Coreweave, you know, These companies have not built a greenfield 200 megawatt IT data center at 200 kilowatt rack density. So, and a lot of the companies that are private also haven't done that. There is actually not a lot of precedent in, you know, standing up these data centers at scale at all. So, There is very little material for a very good apples to apples comparison. And I think what we're going to find out in the coming 12 to 18 months is that there's actually, you know, real winners and people that are just barely getting by. And I mean, this is what we've been predicting for the last, like, basically 12 months since Iron has been very actively getting into AI data center construction. I think this is going to be a differentiator for Iron. And, you know, I'm mostly covering the construction side of things and, you know, the whole operational thing. But my take from Raise is that Iron is really excelling across these both both of these layers. So I made a post about this a few days ago where I mentioned the three layers where layer one is your power and your land and your data centers, and layer two is your your compute, your GPUs, your networking, and then layer three is your software. So your added value on top of the of the bare metal compute. And I think Iron is is really finding ways to excel in layer one. And layer 2 is something that they have really figured out now. And this is going to reinforce the layer 3. So the main thing to look at going forward, and I know everyone has said, power is no longer a bottleneck. It's all about HBM or about GPU availability. But you're going to find out that Nvidia will allocate GPUs to the companies that can build the data centers in the time frame when these GPUs are released. So in the last six months, when everyone was losing their mind over when are the GPUs getting delivered, what about the memory, Iron has been perfectioning their construction teams and their tech to build faster than they have so far. And all the best practices are going to be applied to build extremely fast going forward. So they're going to increase their capacity that are going to stand up year over year for the coming 5 to 10 years. It's just going to keep increasing. And they are actually, by doing so, they are going to force Nvidia's hand to provide them GPUs. I know they also have this agreement with Nvidia for 600,000 GPUs for the investment rights. But Nvidia has said multiple times at Raze, and also we heard this from the people that were there that talked to, you know, the Iron executives or to other like Mirantis people, Nvidia is not just going to follow like a priority schedule based on past performance or like you have bought so many GPUs from me in the past, you're going to get this many GPUs next year because that's not going to do anything for Nvidia. Nvidia wants to see these GPUs being installed and ran. So the more data centers you can stand up the high, the more pressure you basically have towards Nvidia to get the GPUs. So I think the bottleneck is honestly still power and powered data centers. And I think that the differentiator is how fast can you stand up all these data centers. So, you know, that's the bet that we place. We place our bets on Iron being faster and in a better position to do so, not having to run into constraints with power sources like energy sources or having to divert to behind the meter. I mean, there's, do you, know, you should ask yourself, is there any company out there that's running computers other than Iron that is running solely on grid power? I'll just, I can just pause there because you're going to find out that it's extremely rare. I mean, there may be like maybe Grok or something is a company that you could say like, they're bought by Nvidia now, I guess. They had like, what, like 50 megawatts globally. Like it's like a very small portfolio and they use co-location only from these, you know, these former tier 4, data center providers. There are a bunch of these like smaller players that are like in the metropolitan areas for the extremely low latency data centers, but those are like tiny. I'm talking about like the AI data center providers of this AI build out cycle. So You can look around, like there's like a bunch of them, like Vantage data centers. Well, of course, there's like the guys like Crusoe Lancium and Libius, obviously, Core Scientific and Core Weave. I mean, obviously Core Scientific, I can't really say if all of their power is grid connected. So that might be the case, and that might be like one of the first iron comparable power providers, but obviously they are not a cloud provider. So, it comes down to your vertical integration and your power source. This is going to, not only de-risk your business, but it's going to make you able to standardize your process. And by standardizing your process, you can optimize and then you can increase your cadence of construction. And this is going to make you more attractive for players in all parts of the ecosystem. And from subcontractors to cooling vendors, Iron is now talking to a whole bunch of reference design vendors. So if you can say, we have all the people to install your stuff, you're going to get your stuff. And so that's basically the very short version of how this is going to work out. So, at Raise, you see an iron executive team that's like extremely confident about their layer one and layer two. And people are just nitpicking about how is your layer 3 going to be? are you, why do you say you don't need a sales team and you go around and you hire this guy or you acquire Mirantis? You know, that's just, that's just noise because, you know, if you, want to fine tune your customer portfolio, you obviously need A big variety in people that are able to hook up all these kind of customers that are in the different part of the ecosystem, like from the very small startups to the hyperscalers that requires a whole different range of employees to work with these people. So when Dan Roberts says, yeah, if you have Nvidia, you don't need a sales team. Yeah, that's obviously doesn't mean that just because you have Nvidia, you're no longer picky about which customer you want to have. So this comes down to the same thing where Aaron says, we could lease our entire portfolio tomorrow if we wanted. But, you know, and this is a direct quote from Kent, you'd be mad to sell your capacity today if you could sell it in two months for a much higher price to a bigger segment of customers. Because it is twofold, right? Selling your capacity today is not, you know, your capacity that you will bring online next year. If you would sell that today, not only are you limited in your pricing power, But you are also limited in the amount of customers that can sign because these smaller companies, they cannot sign for 2027. They just don't have the credit worthiness or the capital on hand to do prepayments of, you know, 25 to 50%, 6 to 12 months in the future. And because they are small, they're also not investment grade for, financing against. So Iron, instead of, turning around, laying on their back towards the hyperscale and say, okay, we need a 2027 customer now. So here you go. This is our price and let's go. And, you know, Iron Radar just keeps on building out the data center. And now they have NVIDIA behind them, watching over their shoulder, every move they make in Sweetwater. You know, they are looking at their construction process and then they have this whole team of guys that are like from all the reference architecture vendors. And they're also looking at the whole construction process. So it's not no longer iron is just single-handedly building something and they will come. It is a joint reference design. I mean, it's going to be a flagship site, right? So there will be a data center constructed that when it's ready in 2027, Nvidia will very gladly send those GPUs over because first of all, it's their flagship site, right? I mean, it's going to have like literally an Nvidia flag probably hanging there at the campus. But, you know, Iron can then contract these GPUs when they are being installed, and that opens up all these other companies that no longer have to, prepay for compute in the future, but for compute next week or compute in two weeks. And that also, you know, allows Iron to, obviously they will still need to finance a part of the build-out without the customer having signed. So this is the headache for a growing company like Iron. we should start looking at the way they have, they can finance their build outs that are signed at the last minute. So I think the reason why the 50,000 GPUs, 50,000 B300s have not seen any announcement is Iron is not going to announce anything if they are just going to you know, sell it to their existing customer base. And maybe they will even add newer customers that are even smaller than their existing customer base. Like right now, they're selling to, you know, together AI, fireworks, AI, base ten, those kind of you may. I think there are going to be even smaller players that are going to be introduced that, you know, take smaller clusters, but, you know, feel very happy to just pay down for three years or 50% or 40% at a high rate because at least they know they are good for their compute, So because of the high percentage of down payment, it also means that there is a lot less left for Iron's finance. But okay, so I guess what I'm trying to say, I'll just wrap it up because I'm going on an intense rant and I don't even know if everyone can hear me because my microphone is apparently a disaster today. So I think what's going to happen is over the coming 12 to 18 months, we're going to see iron be a differentiator in at least two layers out of the three layers. But just remember that if you excel in the first two layers, so your power, the cost of your power, the source of your power, your vertical integration and your data center capacity and the pace at which you can stand them up, that gives you a massive advantage in your second layer. So why is that? Because if you have one, NVIDIA will help you get to. And if you have the right partners, the partners will, um, you know, help you to get your networking. Because, you know, just think about it. If you're a, if you're a company that produces a product and your, your customers, uh, cannot, uh, you know, install your products, how can you, uh, make your procurement, how can you make your roadmap into the coming years if you don't even know if your customers can buy your stuff in the future? You want to sell to the customers that have the highest capabilities of being a long-term customer that are being, you know, so that's why Iron, by excelling in the first layer, they will be able to get the second layer as long as they, are good for the money. So I think the financing part is still, a topic that we will have to keep an eye on. But so by being vertically integrated and having the right partners, you will get the chips. And then with a team like Mirantis and, at some point, maybe more than that, everything in layer 3 will gravitate towards you. And then Iron will be able to get the right customer mix, which will ultimately be the driver of their financing mix. So it is very, maybe I'm not really clear on the details. It's a sort of a high-level approach, but it is very important to start seeing iron the investment from the eyes of the company because after hearing all these first-hand talks with people at Raise from Iron, it's really clear to me that, they are really living in a different perspective than we are. I mean, I think the stock is so poorly performing and the reputation of the stock on Twitter is so bad is because the company is so misunderstood. So I think what I'm going to do going forward is I'm going to try to approach it more from that perspective. And this rhymes a lot with what was said in the All In podcast. May not draw immediate parallels now, or maybe you can't get to that conclusion just from my big rant. But what Gavin Baker said, I just had a quote here. I'll just share this post in an S. It's not the specific quote, but this is literally iron, you know, vertically integrated AI infrastructure provider that's going to, you know, be able to get the lowest cost per token out of their megawatts. I mean, I will gladly hear anyone that has a different take on this, but I don't think there is many other cloud providers that own their own GPUs and their own data centers that are fully grid connected and vertically integrated from the land to the compute that will be able to compete on a token price war, for example. I mean, gladly hear someone who thinks differently. But I'm not going to I'm not saying that that's the strategy of Iron. Obviously, it's not because they're waiting to sign for to get higher margins. But, you know, if you turn it around, if there will be a massive dump in token price because open source is going to somehow flood the market, you're going to be happy that you're in the company that can play along with that and won't be losing money like some of the others. So I'll just pause there.

₿itcoin ₿utcher: Let's go to Robo, seed investor and a friend. Robo, how are you today?

Robo: I'm good, thanks, Butcher. I had a question for Franz. Just I know we a lot, certainly I did anyway in early days is of this AI model was that, you know, you were looking at colo deals being the sort of the, I guess, the foundations of Iron's balance sheet and the capacity to finance the balance sheet and that the infrastructure as a service would be the margin business. Now we're seeing that evolve into where I guess the bare metal, I use that phrase carefully because it can mean different things to different people, but that that potentially is if you guess, the foundation of Iron's balance sheet now and this business model that you allude to with smaller customers is where the juice for the company is. Do you see, do you have a ratio in mind of that, Fran? And also I saw that Kent talked about at Ray's that three years, but he saw the sweet spot for the term of deals. And I guess if we're going down to smaller customers, that their term will naturally be a little bit shorter. But I see Brian Fry, who's hopefully he's on this call, but I always like reading, talked about this morning or just posted this morning on the significant increased willingness lately to fund against sub-investment grade counterparties mentioned by Kent also at RAISE. So I just, I guess, That's the background of my question, Fran, but have you given any thought to what you see as an optimal business mix in terms of both term and customer?

Frans Bakker: Yeah, so the mix that I expect Aaron to go forward with is I don't see them go over 15% hyperscaler in terms of capacity. I see that as extremely unlikely, because if you look at, we can take it from a gross MW perspective, if you will, but right now, Microsoft is stated to be 300 MW gross for Horizon 1 to 4. I know it's not going to be actually that, but they are allocating that. So out of a, 5.8 GW portfolio, at the end of the year, you could say, I mean, I know not all of that is literally energized, but that's five, a little bit more than 5%. So I see them during this period, they're going to install these 50,000 B300s. They're going to start retrofitting data centers in Childress for compute next year. I think we're going to see increasingly more non-hyperscaler compute or non-bare metal only compute getting into the mix. I think all the other compute outside of the Microsoft contract is going to increasingly be more at least viable for Mirantis kind of bare metal plus compute. So I think they will sign a hyperscaler deal outside of Horizon 5 and 6. I think Horizon 5 and 6 is very likely going to Microsoft as well, but I think they will probably sign more NVIDIA deals. They will probably find one other anchor tenant or hyperscaler deal besides that. So, you know, I just think if you add all that up and just take Nvidia out of the mix because I think Nvidia is like a very special case. It's not really a hyperscaler customer. It's also not a bare metal customer. It's they are managed service contractor first first one. So they're probably going to be only managed service contracts going forward. So there's a lot of juice in there, is what I'm saying. I mean, $15 million of megawatts of IT loads in revenue, that's a lot of juice compared to, you know, what we got from Microsoft for Horizon 1 to 4. So even if there would be like 200, 300 megawatt additional hyperscaler or bare metal, you know, the poor bare metal rate on top of that, you would never get to much higher than 10% of the mix. So I think maybe over time, they will add some more if it's really needed for the capital. But I think it's going to increasingly shift to more cash flows and more asset-backed financing. So, I think the investment grade of the customer is going to be less relevant going forward. So I think 2026 and 2027 are the, coming 18 months is the massive sweet spot in monetizing your GPUs to the end users as much as possible. That's also what was communicated at Rays. Any compute you can deliver before 2028 is in absolute high demand. And you'd be seriously scoring your, shooting yourself in the foot if you would sign that away during that period to hyperscalers or to middlemen, if I can call them like that. So that's my, let's just say it's probably going to go towards 10%. It's maybe at some point in time, it's going to be between 10 and 15%. But if they announce a new site of another gigawatt, you know, you're going to see that the mix will go down. So it's probably going to be below 15% indefinitely, in my opinion. So that means 85% of Iron's portfolio is most likely going to be dedicated to anything below hyperscalers and middlemen. So that can be anything like Fortune 500 companies the Palantirs or big pharma or, you know, AI native companies, startups, it's going to be a mix of that, like the real juicy GPU, our business. So, Andy, sorry, Fran.

Robo: And do you see any need for to iron to maintain that sort of 15% ratio across each sort of field or each site they have? or that would just be a portfolio target?

Frans Bakker: I think as a portfolio target, I mean, I don't really see why it would be site-specific necessarily. I mean, obviously, the cash flow from site A will also be used to, you know, stand up data centers in site B, for example. So there, I don't see it as a side by side mix per se, unless there is something that is really that I don't know that does require that. But I would see that as not very important. I think it's more like a portfolio thing. But maybe, for example, for Australia, it's a whole different story. Maybe like in Australia for Bundy, they would need an anchor tenant for, you know, to set up relationships with, you know, maybe they need a customer that is like government approved. You know what I'm saying? So maybe maybe it's a bit different there. But I think in Texas, they will just take the bare minimum. They really the bare minimum of middleman customers that they need to succeed.

Robo: Thanks, Fran. Thanks for all your work. It was brilliant. Thank you.

₿itcoin ₿utcher: Before we go to Mega Man, I would just add Robo. I think if you were to ask Dan, and we're going to try and interview him before or after earnings, my readout from an outside perspective, we're not inside, but we'd like to think we have an educated view on this. You think about some of the hyperscalers, though, that you would potentially sign, whether it's Google, they have the TPUs, you have Amazon with Trainium, you have Meta with their own ASIC coming out. Those are all direct competitors to NVIDIA. And I for us to commit to five gigawatts to NVIDIA, I think we already are married to Jensen and his organization. And I don't I don't think it's I just think at that point, like you've cast, you made your bet on Jensen. It's not to say you can't do business with those organizations, but I find it less likely that we would for that reason. And maybe it's Microsoft exclusively because they're leaning into open source more so than those organizations. At least that's what they're communicating, which kind of plays into what Jensen wants to do. The other piece is I think the reason they did Microsoft originally was the whole premise of a hyperscaler was the credit accessibility and the credit worthiness of a triple A. Like the GPU financing market appears to be healthy. And if they are, to Franz's point, Fortune 100s with good credit ratings, I don't think it appears to me will yet to be seen that they'll be able to get similar financing terms through Dell. And most importantly, it gets back to Jensen. If Jensen's on your side, I just have a hard time believing that they're going to stray too far off the path and start doing more business with Google or Amazon for the reasons that I've mentioned. So those are just a few other considerations and But let's go to Jason now. Mega Man, how are you? Jason, I can't hear you right now. Are you on mute? Let's give him one more second. I'm trying to message him.

Jason: Hey there, butcher. Good evening. Sorry about that. I switched from a Bluetooth. Now I'm doing well. Thank you. Thank you, neighbor.

₿itcoin ₿utcher: Yeah, what do you what do you got tonight?

Jason: Sure. Great. Hey, listen, I wanted to hop in real quick and you know, agree with a few things that Franz was saying just in that man and this stuff is popping off and what I was going to get at is, you know, we do equipment finance for companies that compete with iron, let's say. You know, we've got one or two in the space, one particularly that is a top 20, you know, NeoCloud and the same things that we're seeing or hearing coming out of, you know, IRN that they don't need a sales team and things like that are pretty true because, you know, everything I'm seeing here out of the customers that I talk with are numbers that are growing exponentially, really. You know, and obviously forecasts are what they are a lot of times, right? They're kind of hockey stick looking in business. So but you know, the company that that I talk a lot with and I don't mind, you know, talking about them at all, really, or what I learned generally across, you know, a top 20 player niche here, you know, without saying anything about them, you know, specifically, I know I don't mind talking generally and I like talking generally about them, but Actually, I feel like the team who, you know, Mike Alfred, I can't say anything or whatever because we have NDAs and stuff. But long story short, now Mike's a legend. But, you know, things are going so well at these companies that, you know, I see it translating over to Iron real easily. So, you know, these guys have everything from bare metal line items going across the board to what they're calling now. inference as a service, which is really booming. And I'm actually going to learn a little bit more about the definition of exactly what that is. But to Franz's point and everybody's point, it's the smaller AI, you know, the native, the smaller labs and enterprise customers that are using the inference as a service. And I know OpenAI is actually one of those customers that's using inference as a service, they call it, and that line item is growing. You know, like I said, exponentially, whereas the bare metal stuff is actually running off kind of flat. I'm looking at the numbers right now going across the board. The projections really run flat for the bare metal stuff. So you can tell that they really do want to get into the sticky enterprise stuff as a as a real niche company that I talk with a lot does not own any data centers, did 164 million top line revenue. making about 24 million in net income, which you know is not bad seeing as how not a lot of people are making money right now at this right while everybody's scaling. But they've been in business for 10 years at this stuff and they have a small environment with AMD. So I was going to mention that too. AMD is doing a lot of the same things that Nvidia is doing where they do, you know, have a They've got someone else hosting GPUs and things for really, as I understand it, it's their back end software and things like that, but also they're growing as well. The environments that they're working in the background are also growing. So I actually want to learn more about the products and the services that they have going on with AMD, but. Long story short, things are trending well across the niche from what I can tell. And I wouldn't want to be in another place, really. I was a Cores investor a long, long time ago. Still have some stock in that company. And I like the energy infrastructure space. But I think that the GPU as a service space is actually, like we're all getting at here, the hotter way to go and better bang for your buck. at least what I'm seeing. So I'm going to keep looking at what I'm saying here and hang out. But I wanted to mention that, man, things are going well and it seems like customers are not the problem. It's standing up compute and getting financing arranged as well. So that's where I'm at right now. Butcher.

₿itcoin ₿utcher: Can you help everyone understand your company does the equipment leasing, is it, or can you help me understand your role with that? smaller cloud that you were speaking to?

Jason: Sure. Yeah, absolutely. That's exactly what we do is asset based finance. So we did a round of AMD. I got I'd have to get down into the get my propeller hat on and get the quotes out to tell you exactly what they were. But we yeah, we do GPU financing. Plain and simple is one of the, you know, things that we lease. We lease a lot of IT equipment. We do a lot of actually hydrogen fuel cells as well. Not not bloom, but other ones. But long story short, yeah, we do. We do private credit related things.

₿itcoin ₿utcher: So from your vantage point and I see God particles in here and this is something that he mentioned this past week and something I had been talking about. Everyone sees, I think one of the bear cases with iron specifically is you see the costs of the GPUs or the AMD chips, everything might be up, you know, 50% from last year. But with Iron specifically, the only proof, and if anyone from the company is listening, my biggest pet peeve that I will ***** about continuously, just like GP, is the fact that when you guide for $3 an hour on a B300, that the cost of it is almost 50, 60% higher, it looks like it's not economically viable. And that speaks to my impatience. But also, I know that those B300, specifically the 50,000 that they bought in March for approximately $70,000, and then the B300s for Nvidia that are closer to 80,000, let's say 78,000, that they were able to lease out at $377. This is a long way of me saying that I expect, just like GP, for those to lease out at least $4, if not $450 to $5. an hour. So my question to you, Mega Man, given that you're in the field, like our customers, how much does that impact financing with the increase in the size? Like it's probably better for you because the loans are larger and you guys collect your spread. That's one side of it. But from a your actual customer who's leasing their equipment and then the end users that are buying the compute, Would it be fair to say in your view that they're able to pass along their costs of increased CapEx to the end consumer and that there's not much pushback right now given the compute constraint?

Jason: You know, I think in a nutshell, I'd have to say you are correct, right? We look at things from a top line perspective and then a bottom line perspective. And most of the things we look at are backward facing, right? But that being said, right, that's where this stuff does get tricky because it's a forward story, right? It's what's on the come. So, you know, you know, that's the thing. These these these prices are bouncing all around. And actually, I am about to dive deeper into all this myself, right, to your point exactly, and get more granular with what what what an increase, you know, looks like. But You know, I'll tell you what, over the coming weeks here, we'll talk more about this specific, you know, use case and really exactly how we're underwriting it. But you know, the last round of financing we did was underwritten along the lines of the existing contract that they had in hand. So we actually we don't take a security interest in the contract, but we see the underlying cash flows. And that's, you know, those were set in stone, kind of take it or use it, right? Exactly that, take or pay. So, you know, as things increase and stuff, obviously, right, revenue increases. I'll have to look and ask more about what they see for future hourly GPU rentals and what they look like moving forward. So we'll talk again on that topic in the future. Look forward to it.

₿itcoin ₿utcher: Thanks for coming up. Franz, you had your hand up.

Frans Bakker: Yeah, I think my takeaway from what I heard from Ken Draper at Rays, and just to be clear, I wasn't there myself. I've spoken to people that spoke to him, so it's just so you know that. But what he was communicating is all the extra costs and it can be GPUs increasing in cost or just the memory factor, it will all be calculated towards the end user. So they have no concern about that. I think when we are breaking our, where we are completely surprised about an increase in GPU purchase price from one batch to another, well, they are not increasing their guidance relatively. That is just because, they are not telling us what they really think is going to happen. Iron is just extrapolating previous signed contracts into an implied GPU hour rate for the next cycle. And that is honestly, it's very distorting for valuing the business. Basically, If sell side analysts follow what Iren said, it looks horrible or it looks suboptimal. And if you follow, you know, what the market is telling you that the GPU hour prices are doing, and if you listen to what they are saying, yeah, we are waiting with signing the customers to get a better GPU hour rate. How can you honestly follow their own ARR guidance? It's just based on ******** like literally. So Iron is confident that they can, mark up the higher prices of the GPUs towards the customer. So they're not going to lower their, they're not going to see a lower margin because the cost of the GPU increases. So I just wanted to point that out. I see there's a lot of these accounts on X talking about DRAM and, how it's going to keep going up into, Iron sees that as well. They see a constraint into 2028 and it will probably be a factor well into 2031. So internally, that's how they are looking at it, but they are not concerned because if customer A doesn't want to pay for that higher price, then they have customer B to Z that will pay it. I mean, I think the whole notion of that there is some kind of limited demand where you need to fit a bunch of, check a bunch of boxes, otherwise they will not take it. There is no such thing. I mean, this is all or nothing. So these guys, they will, you know, Iron is not just, choosing between 10 customers for one gigawatt. They are choosing between hundreds of customers that want some part of that one gigawatt. So all they are really doing is they are like a DJ sitting in front of a big mixer, a big table where they have all these levers that they can pull. And we are standing in front of it screaming when deal, it's extremely silly that, we are not in line with the DJ right now. We just need to understand what they're what they're really doing. And so, yeah, it's frustrating that they're not really sharing this, you know, what you can derive from going to an event like Race it really gives you not just conviction, it just makes you understand what they're actually doing. And so it is our responsibility, in my opinion, within this community to keep hammering the table about this and pounding the table about this. And we will, you know, as far as I'm concerned, I'm going to do that weekly. So yeah, just wanted to put it out there. These higher costs are just going to be translated to the customer. So transferred to the customer. So, you know, I'll leave it at that.

₿itcoin ₿utcher: It's irrational at this point. If you're trying to lower your cost of equity, you have to show that your business can make money and eventually they're going to contract those. Hopefully we see something at August earnings and I know they haven't executed those contracts yet, but. If your end goal is to lower your cost of equity and acquire more long-only funds, which we had a member of the board come on this exact space and communicate that, the fact that they're not communicating the marketplace economics more accurately is just a lost opportunity. And I won't step away from that opinion. It's just kind of ridiculous at this point. It will flow through eventually. And I agree with everything you said earlier, Franz, that there's a difference between us screaming one deal and like there's actual evidence that. They're received GPUs at McKenzie, which most likely means based off conversations with Dan that they already have a customer in mind and have agreed upon something and maybe it's not executed yet. But if we get to August earnings and they have executed contracted ARR and they're not reporting an uptick in that, I just I don't understand it because you're not going to see it till the following quarter. And then that's another three months where people are doubting it because they can't see it in the income statement. And I'll just stop there. Someone else is requesting to come up and I'm going to vet them. What were you going to say, Franz?

Frans Bakker: No, I agree with you. I mean, I will give them like, I will give them the time, you could almost say the benefit of the doubt to get to August earnings and tell us, okay, look guys, our previous ARR was based on backwards looking. We are now three months further. We have had a range of customer negotiations and we have started to receive our first GPUs of those 50,000. As we have discussed, we like to sign these GPUs closer to when they arrive. Some of them have arrived. We have came to the conclusion that the implied GPU hour rate for these GPUs is a lot higher than our previously guided ARR. Thus, we are going to increase our year-end ARR guidance by this much. And I think that if they don't do this, then you could seriously consider them insincere, and that would not be fiduciary from them towards the stock. with the ATM in mind, they would need to be honest about their end of year ARR. And I think you can't keep sandbagging just for the sake of sandbagging because it is completely frustrating the sell side analyst price targets. I mean, it was shown before the ranges are, you know, from expected like $40 million of ARR AI revenue to 240. Like these things are so wildly off, you cannot make that up. I mean, all, for example, IR would have to do is pick up the phone, call 20 people and say it's not 40, it's not 240. We expect it to be in this range. straight after the earnings call in August, they do these debriefing with the institutions. They should just do all of those guys do so they have a proper way to value the business. You're not going to have these, you know, crazy disparities between these price targets where it's 45 or it's $120. I mean, obviously there are different methods to value the company, but If at least they have the right revenue estimates to a certain extent, it would make it so much easier. And, it would also clear up a lot of FUD that's being spread on this channel. It is just, because on the other hand, if they don't do it and they think that, oh, we're being sandbagging and legal told us to be very modest, and what you're going to see is during this period, during these quarters, Iron is going to be put away with not only a poor AI revenue profile in the past quarters, but also the future doesn't look too rosy either, right? It's going to be like, oh, so you acquired Mirantis and you can't get more than $2.80 for your B300. Look at how much it's going at, blah, blah, blah, blah, you know? And we keep getting back into this discussion. I think we're past the point of sandbagging is good. the whole under promise over deliver thing is it's just let's do a little bit more under promise and we'll take a little bit less over delivering for granted. OK.

₿itcoin ₿utcher: Yeah, just anyone who believes that they're going to execute those contracts at $2.92 an hour after they sign NVIDIA, who's arguably the largest Mag seven. They're in the top three for 377 an hour. I just simply disagree with you. And it's just kind of stupid that we're still talking about this, to be honest. So I'll stop there. And I think we've made our point. But Franz, what other from the construction front, do you have any? I also had I pulled up. I have two thoughts. You were talking about the three layers of the piece, you know, the economic model for iron, but specifically, do you know anything? I haven't had a chance to research the two gentlemen that they recently hired. I know one's the former Oracle exec and the chief, I believe that was the chief product officer. And then there was one other hire. Can you speak to those at all? Or otherwise, do you have any construction updates for everyone?

Frans Bakker: I'll just finish off with the last point to the previous topic is that, yes, the NVIDIA GPU hour rate is exceeds $3.50, but don't forget that there's still a backlog and a hyperscaler sort of customer. I know it's managed service, but I just I would like to remind people that the GPUs that are arriving sooner will be contracted for a higher rate. So I would expect those to exceed that rate. So I would not be surprised to see $4 plus GPU hours for shorter durations, three years versus five years, right? And GPUs that are received. So other than that, yeah, I mean my construction updates. I've just been away for a few days, so I've not really been into it much. I did share yesterday or today for some of you that Sweetwater 2 has broken ground, which is right around the time which you expect it to be. So 18 months before energization, they start to, you know, get their substation. I think what we're going to see is a utility substation next to the existing substation. And then Iron is going to build their own bulk substation to connect to that switch. So I would expect to see a lot more movement at Sweetwater too in the coming three to six months, like for the remainder of this year, because they will need a bulk substation in, you know, by the end of 2027. So Obviously, I can understand people not appreciating this kind of intel right now because it's very far out, but it still shows that, Iron is not standing still. And maybe, that looks like it sometimes in track order when there's not much news other than acquisitions and sponsorships, but they are really moving. And adding Sweetwater 2 to the mix is just going to add another playground for their data center development. And that just means that in parallel they can do more. And I think that's ultimately what's going to happen. So Childress, Childress is the last thing I saw is the Horizon 3 dry cooler installation is nearing completion. So that's very good. Horizon 2 liquid cooling plant is showing signs of near completion. And I think what was said at Raze is that Horizon 2 already has some GPUs inside, but not commissioned or installed or just arrived in the building. My expectations is that Aaron will announce something around the handover of Horizon 1 next week, so not in the week of 13 to 19 July, but the week of 20 July. That's my estimate. Yeah, so Childress is looking good. They are in the process of demolishing the Block 4 and 5 mining buildings. So those are 12 buildings that are going to be, well, I'd say demolished, but it looks, right now it looks more like disassembling. So it's not like they have a big slash hammer or something like that. It's not going to get blown up or ran over with a bulldozer. It's being taken apart and the parts are being like properly, you know, displaced. I don't know if they're going to like reuse some of these materials or not, but at least it's not being done in a manner that it's like actually being destructed. But I guess we're going to find out. But so they're literally working on multiple fronts. It's like the Horizon 1 to 4 is construction. Horizon 5 and 6 is in the like disassembly preparation phase. And then you have the retrofit for the GPUs that will be installed this year. That's just going according to plan. I need to read over some notes from Ray's, but I think Dennis Skrinikov has said a few things about what they expect the retrofit to look like. I think it was implied that the cost of a megawatt retrofit would end up somewhere near $6 million a MW. So it's a little bit higher than I initially expected, but it kind of makes sense because, you know, the bill of materials is going to go, you know, it's inflating these these things like generators and, UPS, et cetera, everything is getting more expensive in this part of the build out. So, I guess there is no, not really a big problem. I kind of like estimated that even with a $8 million a megawatt as opposed to $4 million a megawatt with a you know, with the Nvidia contract, for example, like $15 million of revenue per IT load, you're only looking at like a few more months of of payback. So it's I mean, obviously it's frustrating, but it is just what it is. But I think going forward, you're going to see this just increase the GPU hour rate on top of that. So I think it's not just the GPUs that are going to inflate the GPU our rates is also not just the markets, the spot pricing. I think it's a mix of everything. So GPUs, market pricing and, you know, under scarcity and then even the CapEx that goes into the buildings. I don't know exactly if it's like a one-on-one, but I'm not worried about it. What's more important is how quick they can do it. So I will, for now, that is my point, my standpoint. So it looks like they're, well ahead of schedule. They are already retrofitting data centers for 2027. So this is the whole thing that we learned from them during the Bitcoin mining periods that they don't just do 2 buildings and let the rest wait until the first two are done. They just do a whole bunch at the same time. So you get like a streamlined process where the first team goes all through the first building and then to the second building and the third building and then behind them, the next team will do their part. So it's like a staggered work, staggered build out, a staggered retrofit. So yeah. H1 2027 is really when we are going to, get our aha moment, our see, I told you so moment. And, we are just watching it from afar, but it's going, getting along very well. And they are also, building out the site in general. So the children's site is still seeing a new office being built. They're building new lay down yards. It's like, it's really growing. It's becoming very dense in terms of, footprint in relative to the amount of acres they have. So it's starting to look very attractive from a satellite point of view. And yes, Sweetwater is a bit of a mystery, but I'm going to, in my subscriber group, I've planned a new satellite image to be ordered today. So that's an $800 satellite image. So those are very expensive. But they give you the extreme high definition where you can literally see the reflection of the hood of a car, for example. It's like you can see things that are, very tiny and that makes you able to really understand what's going on. So with regards to Sweetwater, I can't really say where they are at. I just know that they are building, they've started foundational works for the data center for the DSX reference design for the 200 megawatt IT loads of next year. They've started that foundational works. It's, yeah, so more to be continued. And other than that, it's mostly secondhand information from Race. Prince George is fully contracted and I don't know if that means fully operational. I don't think that the GB 300s that they acquired, the 1200, I don't think they are actually operational now, but they will probably will be in the coming weeks or months. But it was said by Iron C-Suite that Prince George is fully contracted. So The McKenzie, as someone already said earlier, is also mentioned by Iron C-Suite that McKenzie has some GPUs, but they are not running yet. But, you know, we can estimate that it will probably take like at most two months until they are really billing. So I think it bodes well for some P300 revenue in the current quarter. It's not like guaranteed it will be very, very much or anything, but it could be like a little gravy on top of the horizon one revenue, right? So that's starting to look good. And on terms of construction, I think Dennis, our CTO, literally said that after the whole learning process in Prince George, McKenzie is really a rubber stamp. So, you know, they are not going to experience hiccups or anything that they may have had in Prince George. It's really now a matter of copy-pasting and best practices applying. So, and kind of flat since the 2027 thing, so not much to say about that. So yeah, that's my update from the sides.

₿itcoin ₿utcher: Let's go to Lando and then probably closing thoughts. Lando, what's up?

Lando: So I just wanted to take a second to thank both of you guys, friends and Butcher, for covering race. I know I first brought it up. three or four weeks ago with highlight again. I'm just glad that it kind of turned out to what it ended up being, you know, special thanks to all the guys that flew out or took a train there. They did some really good ED, but yeah, I just wanted to highlight like a few things. I don't know if you guys talked about it before I got here. I joined probably half hour ago, but surprising one. Either I agree with Franz. I think it'll be turnover next week, so not in the next 5 training days, but the following day or week. If I had to guess it'll be a Monday or Tuesday announcement. And that'll be huge. The Maranches acquisition is also supposed to finish up this month. I believe Kent said that to either Finn or Harris. I'd have to double check with them. And then this week, I believe it was Harris told us that the GB 300 exemplar cloud certification from Nvidia will most likely get announced. That's just like icing on the cake. You know, it probably won't move the stock price too much, similar to how like the B300 announcement didn't move it too much. But it's just kind of that validation now that they're still working very close with Nvidia and working on executing at a high level. But and then another unexpected announcement was the bare Rubin samples. We don't know when those are coming. I believe it's in the next month or two is when they're trying to get it for. But yeah, all good things from Ray's. I'm talking to a few other guests that weren't part of the main group. And they're going to get me all their notes and information because they also met with, I believe, Kent and Dennis and a few other executives at the conference. So I'm going to work on getting their notes and hopefully it's a similar information or something new. But yeah, that's all that I had. I just wanted to highlight those few things because I know share price and sentiment on X and just in general has been Not the greatest, but the good thing is that they are executing and they are working towards the iron cloud and whatever that vision has turned into. But yeah, that's really all I got. I appreciate all you guys for giving me the opportunity and I'm really glad the raise about ended up working out and not just being another conference. And then hopefully, I'm going to try to network to get Finn or Harris or other people to go to GTC when Nvidia goes to Europe. I don't know if I can, if I have the connections or strings to pull to get there, but we'll see. So just know that that's in the works and I'll get all the other information from the other Ray's guests as soon as I get it all. But thanks guys.

₿itcoin ₿utcher: Yeah, so I pulled up the matrix. I can post the link, but the only companies that have exemplar status for GB300 are AWS, Global AI, Azure, Nevius, and Nscale, and not even Coreweave. So I guess, you know, $15 billion, which of these is not a like iron, versus all those other companies. I don't know what Nscale's valued at right now, but our current valuation is a very good opportunity, not financial advice, but they're showing. I think that announcement, along with the B300, shows that the technical expertise is in the background. I see Bitcoin AI guy asking when delivery. I think that's the piece, just plugging these in and generating revenue and maybe management upping their guidance to what it actually is. I think those are like two or three things off the top of my head that would communicate a more commercially viable situation to the market, but we will patiently wait. Franz, any closing thoughts? I appreciate Lando, Robo. Mega Man, everyone else who came and attended, it's good to be back this week. I do believe, just to reiterate, that Iron will be the most productive, profitable token factory data center builder in the world. And that's why I'm still here. And I hope you guys see that our feedback from time to time is just that it's There's anything, there's nothing going to be 100% perfect. And if we just become some echo chamber that champions every single thing that management says, I think this loses value and I won't do it at that point. So I appreciate everyone's time. But Franz, anything to close up?

Frans Bakker: Yeah, I think instead of time to compute, maybe for an optics view or like a sentiment kind of thing, time to exemplar or when exemplar is maybe a better way to look at it. Because, if you look back at when Iron got their B200 or B300 exemplar status, there was probably a lot to say from the side of Nebius et al. We got that already like so many months before or last year or something. But now you're going to see Iron will announce the GB300 exemplar, and then there's like only a very small list of people that had that in this year, like weeks or months ago. So you're going to see the gap between Iron announcing something from Nvidia and Pierce doing that, because you're going to see it shrink. And I know that Coreweave was the first one to actually bring up a Vera Rubin rack, but there's no one who has Vera Rubin exemplar status right now, I believe. So I think it's going to be interesting to see, you know, if you look back at it somewhere next year, the time between the peers when they got their B200 or B300 exemplar versus the GB300 versus the VR200, I think you're going to see that the playing field is being leveled from a NVIDIA perspective. So I think the whole narrative of we are a preferred partner, so we get our GPUs first is going to completely disappear next year when it's all about we build our data center faster and better than you. So we got the GPUs before you because that's what NVIDIA wants to deliver to. So that's just a final observation around this whole thing. Other than that, I completely agree with you. It's just literally high level. It's just Bitcoin mining all over again. So yeah, we will see how it turns out, but I'm bullish. And, for now, we're just the stock is just going to, not do anything as long as there is no news coming out and macro is dominating. It doesn't look very good for the coming week with the Iran war, you know, restarting. So let's see. But yes, we are not an echo chamber. I just don't think it's worth my time to, get too much into these things that are, not meaningfully impacting the stock. I mean, be as it may, all the bad news or sponsorships or RSUs or whatever, it has not really, really impacted the stock as much as it did the sentiment. So I think that it has probably has more weight from us to explain all the things that are going to move the stock in a good way, rather than dwell on all the negatives that are more like a principle thing or, it's not to say I don't have any principles, it's just that the reality is if I want what's best for my own portfolio, it is better that I educate people on what is going to really move it. And I think on the downside, what moves iron is mostly not iron related. And on the potential upside, it is mostly iron. So there are a lot of accounts that cover macro and interest rates and all these kind of things that will pull the stock price down. But I think we do a lot better with covering the fundamentals of the company that will ultimately lead to pumping the stock price up. And it's not a pump story. It's just that Kent has said at Raze that there is no world where their GPUs will not be contracted next year. So I just want people to understand that if I say Iron's going to print EPS next year, then you can help me accountable for it. So I am very confident that next year the earnings are going to look really, really, really good. Even if maybe the EPS is not going to like be meaningful compared to the revenue. At least we are going to bring in a lot of revenue compared to the first half of this year and even compared to the whole 2026. 2027 versus 2026 is going to be night and day. So yeah, I just hope people can weather the storm until, you know, the iron is just a mid-cap right now. I mean, it's like we are valued at $14 billion or something at the current price. It's completely ridiculous. But this is what you invest in. This is a hyper-growth company with a very volatile stock with a small market cap. This is just how it is for now. But in 9 to 12 months, it's going to look completely different. So yeah, I'll just leave it at that. And thanks for hosting the space, Butcher. I gladly join next week again.

₿itcoin ₿utcher: We'll leave it at that. Good night, everyone, and we'll see you next week. Thank you for joining.