$iren Sunday Night 5.31.26

Hosted by @₿itcoin ₿utcher 🥩 🐑 🐷 · 2026-05-27 · Tags: IREN

TLDR

The speakers evaluated IREN’s transition from Bitcoin mining infrastructure into a vertically integrated AI cloud and data-center operator, focusing on Nvidia’s architecture, Mirantis software, Sweetwater, and near-term execution risks. They were strongly bullish on the long-term opportunity but acknowledged delays at Childress, uncertainty around contracts and hardware delivery, and the need to convert projected recurring revenue into reported results.

Speakers

Notable quotes

Transcript

₿itcoin ₿utcher: We're going to get this started. It's 3 minutes past nine on the East Coast. I'm Bitcoin Butcher. My partner in crime, Franz Baker, took the evening off. Might have a co-host join me. Might not. No big deal. I still have some things to talk about. And let's kick this off. I saw a request. I guess the request got canceled. But a few things that we can talk about tonight. A lot of stuff going on in the industry. It's actually 9 o'clock in Taiwan. And if you're in the know, in two hours, Jensen will be having his keynote address at GTC in Taipei. Hope I pronounced that correctly. And I expect that speech in particular to give us a little wind in our sails this week after how we ended on Friday. Friday, I kind of talked about already. It was the *********** roller coaster where things were up, things were down, and then we kind of closed going into the weekend. So until we have some more fundamental progress, Although there were some things that we can talk about this week that moved us forward, we're still a stock that is highly influenced by market makers and high frequency traders and hedge funds, which makes, if you're watching it day-to-day, it can make you dizzy. But at the same time, we still had our highest monthly close on record at 63 and change. Things are headed in the right direction, but day-to-day for guys that I, and gals for that matter, who try and day trade this, it can be a little taxing on your mindset. But all that noted, we're going to get things going. Some potential topics that I'm going to touch on here that caught my eyes this week and things that I've posted in the nest. I first noted the Wall Street Journal article that the company, it appears, sponsored an article in the Wall Street Journal featuring John Gross, who's the new chief innovation officer. And did, you know, if the Wall Street Journal sought it out or an independent reporter sought it out versus us placing it with an ad, would that be a little more Credible and newsworthy probably, but I do like that IR is starting to try and sell the message a little more in the marketplace and something as simple as John Gross being on the front lines. I think Brian's in the crowd and Brian and I talks privately that maybe marketing in itself isn't going to bring the institutions, but. brand awareness and being on top of people's mind once they do announce Sweetwater or McKenzie or the next Childers Steel, anything that can get our name out there, I think is better than nothing. And in the grand scheme of things, whatever they paid for that article is relatively immaterial to what we're going to have to spend in CapEx to construct data centers and buy GPUs to run compute. So Notice that not much more to discuss than that other than that gross is in the field. I think my gut feeling is he was initially brought in to help solve some of the ills of whether it was Childress or Canada, but we've had, everyone's aware of the delays in execution where we didn't meet the 500. annual recurring revenue by the end of the first quarter. We had the 33 million and, you know, that's the negative. But it looks like along the way we've picked up a talented guy like Gross who my gut feeling was brought in to consult the problem initially and then eventually they just decided to bring him in-house so that we could avoid those pitfalls going forward and he could help standardize a building process that is replicable as the company looks to build out. It's five gigawatts on record, which is more likely 10 to 12 conservatively. So I think where we're going, everything is dictated by process. And I'm reminded of the book Atomic Habits, where it's all these small little data that day, habits that the company's building right now and we don't see it, but I believe it's just a matter of faith and belief in the guys in the background that it's an iterative process that they're figuring out how to do this best just like they did for Bitcoin mining, and eventually they'll be able to build Sweetwater at scale fairly quickly in 2027. I think that's the end goal. And then ultimately, Sweetwater being the flagship that carries over to Oklahoma or any other new Texas sites. I believe someone had mentioned actually that batch zero for ERCOT was being finalized this week, potentially, or in the next week or two. So we might see some new sites. I wish Franz was on because he could speak to it more intimately. But my gut feeling is based off of or there's Mark who's requesting to come up. I'll add him come up. Based on our research, you know, if it's 10 to 12 gigawatts, I have no reason to believe with all the sites that are popping up and how early Dan and Will were in Texas that there's at least another gigawatt or gigawatt and a half. That is just my gut feeling, but Franz could substantiate that better and maybe he'll write something on that this week. The other piece I included in the nest, those of you who don't understand the supply chain and how it works, we all know Iron has these sites in Canada and Texas and Australia and whatnot. Mark, I'm going to put you on mute till I finish up here. So related to that, you have to. Mark, I want to finish speaking if that's cool, but it's hard to with you on without mute on. All right, so we were talking about Dell. They're an OEM, and what OEMs do is they buy GPUs from NVIDIA, and then they construct the servers, which are these big computers that they iron purchases from Dell, and then they put in their data centers before they sell compute. Previously it was just bare metal like they did to Microsoft where they just saw the raw compute power. But now with the Mirantis acquisition, they do these managed services. Mirantis is most known for Cordant, which helps orchestrate and organize workflows. And that's kind of where I'm sure this is what Mark wants to talk about because he's pretty passionate about NVIDIA and the NVIDIA GTC, like where this is all coming together is Now that Iron has this partnership with Nvidia, or backtrack a second, the Dell earnings, short answer is they crushed. They had record earnings. I have in the Nest some of the highlights of those earnings, record revenue, but most notably their AI backlog, their AI orders. And their AI revenue was over 700% year over year. So that's very telling as to where the industry is headed. And that's kind of a precursor to what's going to happen to Iron here. And that's what I look forward to seeing flow through our financials eventually. And I expect more GPU orders. But how that relates to Nvidia, now we have this partnership with Nvidia. Nvidia has this DSX architecture, which is essentially how you design a data center or an AI factory. I hope to hear more from Jensen about that today. And with that, there's what's called their ZGX hardware, which is essentially NVIDIA taking more ownership and being more involved. And ultimately, Lenovo or Dell will build this DGX equipment that the DGX equipment with NVIDIA will be installed into Sweetwater most likely, and ultimately that optimizes the efficiency and then maximizes token production, which in turn, Iron gets to sell to customers. So that's kind of a baseline of where things are headed, but I'm sure Mark wants to speak to that. So we'll just let Mark up on stage here. Mark, how are you tonight?

Mark: I'm great. Thanks for having me on. Been listening to you guys for a little over a year now. So it's great to be on stage with you. I mean, you really summarized it pretty nicely. I was looking for more of a, you know, we could go back and forth on a few things, but you were talking about the fact that the DSX architecture is going to be a big deal. It is. I think we in the group here know that there was a little bit of a hiccup maybe with the initial construction on Horizon One. We shouldn't expect any of those kinds of things happening here once we are full-blown DSX architecture with NVIDIA. NVIDIA has that all mapped out. Some of the people in the group may understand that, NVIDIA has had this program for a while now, but ultimately they map all this out basically as a digital twin. So it's a simulation and they have a blueprint to the exact specs of exactly how they want a data center built. And that goes into the whole total cost of operation and throughput advantages. So there's a lot to be really excited about with being partnered with Nvidia. And last week I was going to try to come up and talk about it, but I got kicked out because of the space went. Anyways, that's not important. But ultimately, I don't think we were really leaning into how important this partnership with NVIDIA is going to be. Dan said it. I think everyone thought he was serious. He wasn't just tongue in cheek saying, you don't need a sales team. We're not going to need a sales team. NVIDIA is going to route enterprise and sovereign partners right through the Iron data centers. And I think what we're going to see with Sweetwater One, the first 200 megawatt tranche, is we're going to see revenue probably close to 20 billion over a five year period based on some, you know, some technical checks that I've looked at and things. And there are a lot of things that can be learned just from the Nvidia site alone. But I've been a big fan of Nvidia for a long time. I've been a shareholder since 2015. And this is the real deal. This is who we want our wagon. You know, we want to hitch to this wagon without question. But there's a lot of other, you know, things we can talk about in detail. But the total cost of operation, you know, this is going to be profitability like other companies aren't going to see. the profitability of these DSX architectures, along with the very latest hardware, you're talking about Vera Rubin 200. And then, of course, they're going to-- which is 56 volt architecture, which is going to change with the next architecture that comes along. And then it's going to be 800 volt architecture. And the name is escaping me right now, but...

₿itcoin ₿utcher: Let me ask you, Mark, I had a question for you, if you don't mind. Yeah. You mentioned $20 billion in revenue over five years. I actually republished something I had sent out last week, and I come to a similar conclusion as you that at about $5.50 to $6 an hour, that comes out to about $4 billion per year, which over five years meets your $20 million estimate. And it's fairly in reach and it's not too crazy when you look at the pricing increases that Navius has published. I think my question for you Are you, and first, maybe if you can give us a little background. Do you work in the industry or is your experience purely related to your investment history with NVIDIA? Why don't you tell us about yourself before I ask a follow-up question?

Mark: Sure. I am not a technical expert in the field whatsoever. I am a former financial advisor who's been retired for about 10 years now. And so really I'm more focused on the business side. of the game. But I found myself really loving the industry quite a bit with the explosion of Jensen's vision. And then, you know, I finally became an iron shareholder in early 2025. And I've been very excited about it ever since. And so it's really just more of like a passion of mine here at this point. And I basically try to read everything I can read as to what's going on, not only in the industry overall, but particularly what this partnership is going to look like, you know, what are some of the nuances and so on and so forth. So, but, but most of the stuff I know is because of, of, of having read a lot of the technical specs and the NVIDIA blogs and all the information coming out of NVIDIA has given me a decent handle as to, you know, what we should expect.

₿itcoin ₿utcher: Well, that's good context. I think we're, and certainly I didn't have a background in AI until 18 months ago and certainly I've never practice it, but I think I know enough to be dangerous by now and enough to ask the right questions from smart people who happen to frequent our program or someone like Franz who's doing due diligence full-time. What I would say related to what you're speaking to though with NVIDIA is just if anyone had the chance to read Agrippa's breakdown of where things he believes are headed, and I tend to agree with him, is we signed this 60 megawatt deal with Nvidia and Mirantis, but Nvidia's designed the DSX architecture, designed all these GPUs, like they don't need Iron's technical know-how. Like it's almost in Agrippa's mind, and I tend to agree, and I think God Particle had contributed to this thesis that the Childress that, which was actually released this week. So let me maybe backtrack a second. Guidance this week was raised from 3.7 to 4.4 billion annual recurring revenue. And that $700 million difference, or if you round up, is the $3.4 billion deal over five years, which is $680 million a year. That was officially added to guidance, even though it was recognized as a deal at earnings. So that was potentially what led to that mini gamma squeeze midweek where we, I think we tapped as high as 68 or $69 a share just from the algos and people seeing a headline that we raised guidance when really it was something that was already published. Why that's important is Nvidia's as technically sound like they invented the space. So yes, they're trusting iron with managed services with Mirantis and Mirantis has this history as a trusted vendor with Nvidia, but like. I view it more as a staging ground in a practice round before Sweetwater. Can they handle the 60 megawatts, work out the kinks with the B300s before they receive the Rubens and kind of scale up to that 200 megawatts? I think the question I'm trying to understand now is I've read a lot of posts related to Sweetwater and Ideal DeFi. I don't know if he attended tonight or frankly has had some great stuff on Mirantis. And one of the things that Mirantis enables with their orchestration is the idea that you can house multiple enterprises or sovereign entities under one roof, call it Sweetwater, all at the same time simultaneously while protecting each individual user's data. And I don't have the technical know-how and there was actually a post ideal DeFi wrote in response to my question. But just assume that Cordant and Mirantis has this ability to using open source divide these work streams without interfering and protecting everyone's data. So I come from more of a retail background myself with our my small business and have a little commercial real estate myself. So it almost I would be interested in your thoughts, Mark or anyone else who wants to come up. But I'm starting to think of Sweetwater, the whole two gigawatt site as almost as a millennial. I grew up going to the mall with my, you know, my mom or my grandma and you'd go to the food court, whatever, like, There's this like feeling of a digital shopping mall that's forming and that Nvidia is going to start flooding these tenants and whether anthropic is the anchor or if Nvidia itself takes on space or some other hyperscaler, those are almost your anchor tenant, like your legacy. Sears for all the now I'm aging myself, but what used to be a Sears or a Macy's or a Nordstrom like these larger tenants that take up more space that might not pay the same rate as your smaller enterprises, but those smaller enterprises will eventually come to play at Sweetwater. And now that there is this. technology known as cordant that can help divide those workloads and protect everyone's work streams. And again, if someone in the crowd is tech oriented, please come up instead of having a meat cutter speak to this. But the fact that it's possible, I've seen enough to just and the fact that. Jensen's willing to bet on Dan and Dennis and the team. I don't know how you couldn't be bullish. Like I think both things can be true where I'm short term a little disappointed in our. execution and our ability to scale, but we're also finding ways to fix that. And I think we're going to see that in August. And it just sucks to have to wait for August to prove what you think's already happening or has occurred. But long term, like anyone with a doubt in the long term vision of this, like I, I can't really say much else. Like you have the number one guy in the industry who basically created this pivoting from what was originally designed for, you know, graphic cards for video games and now is on the cutting edge and everyone looks at his every word and is willing to buy the company that we hold at $70 a share in the future because he knows things are headed in a certain direction. So I think the question baked in there, Mark, was how do you envision Sweetwater playing out? Do you think it's a mix of tenants or there's a hyper who tries to take the whole thing or as a business strategy, where do you see this going?

Mark: As a business strategy, you touched on the aspect I think it's going to happen. It's going to be multiple tenants and they can all be of different sizes. You were talking about the corded program that comes with the acquisition of Mirantis. And that runs on top of Kubernetes, which is the which is part of the full stack. And what that does is it, like you already mentioned, it allows different companies data to be separated into different containers and managed. You know, that's the orchestration level and managed so that there can be dozens of tenants Sometimes people tend to forget that 200 megawatts of IT load is a lot. And they can service a lot, a lot of different entities, whether it be, you know, I don't know so much about sovereign there at the same time, but enterprise for sure and a hyper. possibly of some kind. I think the, thing that iron ore is prided itself on was optionality. And I think they'll have the optionality here. But Jensen knows what he wants to do with Sweetwater. And I'm sure he's going to get what he wants. out of this location. And that's not a bad thing. I think some of the iron diehards might feel like they're losing some of their, you know, might be losing some of their individuality here. But this is honestly the greatest thing that has happened to iron in its history right now. And Jensen is going to make iron into probably the greatest neo cloud of them all and maybe even, a true huge hyperscaler, years out. But that's what's going to happen. I'm almost certain of it. This has been a blueprint of Jensen's for a long time. I know we see, you know, Google is going to be selling their TPUs now out to, you know, not just using them in-house. And quite honestly, that kind of sucks for a huge Nvidia fan because, Jensen really put the, put the kibosh on, his venture of almost becoming a hyperscaler himself. And I think he's going to really show what he's made. And he, like you said already, Butcher, he This is all because of Jensen anyways. I know this is an iron sight and I'm a huge iron fan and actually a big stockholder. It's my number one position now over Nvidia. But there's going to be a lot of things that happen in the next 12 months and they're going to be where iron explodes finally onto the scene. And some of the people who have been with Iron a long time, tired of hearing all the naysayers and all the negativity around Iron not being able to get contracts and get deals like some of these other NeoClouds are doing. But again, it comes down to where's the most profitability going to come from? And there's no doubt that this model that Iron is following is the way to go. Sure, Nebius could back into it just coming from a different direction. But I like starting where Iron started and having the land and the power, having a asset-heavy business, and then bringing on the right pieces to drive revenue and actually, most importantly, drive greater profitability then you're going to see from the Core Weaves and you're going to see from Nebius. And many of the others also, Applied Digital, Cipher, you name it, Huddy. These guys are not in the position to launch like Iron is. And it's going to be really exciting to watch. And Jensen's looking forward to taking away a lot of share of that business away from the hyperscalers.

₿itcoin ₿utcher: Let's dig into what you mentioned. The so formally this project that he put the kibosh on, I think you're referencing was DGX Cloud, which was an attempt to create a marketplace where NVIDIA would channel. users into DGX providers, whether it was Iron or someone else. But part of the concern with hyperscalers who still represent the majority of NVIDIA sales was that they were competing head to head. So that's why that program discontinued. But now to your point, Mark, and I've speaking about spoken about this previously, but with the sale of TPUs and Trainium, those ASIC chips that are developed by Google and Amazon, respectively, now that they're selling those outside of their walls, they're encroaching on territory that was previously Nvidia's, and that threatens their business model long term, even though Nvidia still has the best premium chips. they'd assume just sell everything as opposed to having competition from these guys. So I think the trade off now is to navigate this environment. Those equity investments in Coreweave in Nevius were an attempt to just find new customers that they could grow with in time to replace the hyperscalers over time. And I think What really clicked for me was there was a post after we acquired Mirantis that it simply said that the way their orchestration worked that an enterprise could replace a hyperscaler and that was kind of the moment where it clicked for me where. that was kind of the path to iron becoming the next hyperscaler or certainly as powerful and neocloud as any of them. You know, previously it was just an infrastructure play that they tacked on bare metal, but they were still hadn't shown their software chops where now that they have Mirantis, it's just a whole different value proposition. And you see that through the managed service revenues that they were able to generate on the Nvidia deal, which at 377 an hour, like they've been guiding 296 for McKenzie and Childress. And I've just been screaming about this for months that it's a ******** number and I'm impatient by nature and we're just not going to see it until it's contracted. But my hope is maybe it takes longer with Sweetwater, given the scope of the project and how big of an undertaking it is. And I'm confident at that long term, but it feels like the easy wins for iron and the interim is you just had this **** *** deal with Nvidia where you're just under $4 an hour. And I got to think Nvidia gets the hometown discount for essentially giving you their playbook to DSX architecture, like I wouldn't mind them starting to be a little more explicit. And I can't tell if this has to do with the customer relationships or if the deals aren't executed yet. But we do know as of earnings that they showed the contracted revenue of 3.1, which was the 1.9 for Microsoft and then the roughly 500 for Prince George and then the other 600 roughly for Nvidia. So I have no reason to believe that they've signed Mackenzie or Childers yet that 50,000 GPU purchase that they had in March as of, you know, May 8th, that was so like 3 weeks ago. So if I'm them in the interim. Are they finalizing Sweetwater given the job postings? That's one possibility, but it feels like given the emphasis, look how much the market changed their valuation, like five, $6 a share just by upping the guidance to 4.4, which was already done. Now imagine they start executing some of these contracts and the whole value proposition that Dan keeps pounding the table on is time to compute. And the easiest win they have right now is to finish Canada and the Childress air cooled pieces. So that's kind of if we have a chance to interview Dan again, I personally like long term, I'm all about the vision with Jensen and Sweetwater and we're going to ask about that. But short term, the bridge to that future goes through Childress and Mackenzie. And I just I haven't seen enough to we're just we're kind of waiting still. And that's kind of feels like the low hanging fruit where they can generate some momentum that snowballs and eventually self-funds this instead of keep going back to the markets for more externally solicited equity and instead retaining those earnings and reinvesting them in the business. So any thoughts you wanted to add to that, Mark?

Mark: No, you know, to be honest with you, I I think we're obviously going to we're going to increase our revenues there for sure. I've been so I've been so excited and so focused on what's going to be happening for Sweetwater One, the first 200 megawatts that I've kind of I've kind of just taken my eyes off of, you know, the remaining capacity opportunity for, you know, for Canada. But I think Sweetwater's going to be amazing. I think in the, you've got a post in the nest there where you, I think you mentioned Sweetwater won for second-half of 27, which I think is the safe call. I actually had it targeted for burn-in for May of 27, which is pretty close to second-half of 27. But But, I'm so excited about the revenue opportunity for Sweetwater One. And I really feel like this is going to be the true beginning of Iron becoming a dominant player. Like this is the real, I think the Microsoft deal was great, but I think the great stuff is still coming. And it has a lot to do with Nvidia, has a lot of has a lot to do with Jensen being a part, having his, fingerprint on what's going to what's going to happen there. But that's something to celebrate. There's going to be I mean, there's two gigawatts there in Sweetwater. So we got 200 megawatts coming. And I couldn't be more excited about it. So the deal for that will come somewhere in late 26, early 27. And that combination of increased revenues that they're going to report in the next two quarters, you know, actual revenues, not ARR. And then the step up in ARR from the revenues that are going to come from Sweetwater 1, 200 megawatts, this is really going to be the launching pad for Iron's big move.

₿itcoin ₿utcher: Yeah, I tend to agree. I think what I also put in the nest this week that is noteworthy is the completion of Fourweave was able to complete testing on Vera Rubin. I believe it's all burnt in and officially, whether it's producing revenue or not, that's a little more where I would probably consult someone like Jim to better understand what they accomplish. But the fact that the GPU was delivered and that it's plugged in and that it's closer to producing revenue, if not already, leads me to believe that I think that's the biggest thing why I think it's going to be second-half of 27 is first, they have to learn how to operate the way NVIDIA wants them to, which I think that learning curve gets sped up given that Mirantis is a legacy preferred vendor of NVIDIA, but they still have to probably learn a few things internally at Childress that they can apply to Sweetwater. Then the second piece is the supply chain certainty, which I don't believe, even though they have now moved up the priority stack given their partnership, I still think there's hyperscalers and orders to or we've like we just saw that probably take priority and how fast they can produce is another question, but Looks like someone's requesting to come up. I'm going to add them on. Mark, you're up. Do you have something to add? Maybe Mark requested by mistake. Mark.

Mark: Well, I'm still here, but that's the same Mark though.

₿itcoin ₿utcher: Yeah, it shows his mutes off though, so maybe. Mark Hill I was referencing, looks like he got off stage. Sometimes he has a habit of inviting himself unknowingly, so maybe that's it. But circling back to what we're talking about, I think my biggest thing with, like, we still haven't delivered Childress yet, like, and that can't be understated. The significance of that, like, forget the revenue, the economics of the deal, they're sub $3, like that's pretty ****** in this environment, but we're past that. So assuming that there's still an element of execution risk that even those GB300s probably would go for $4 to $5 now in this market, like that's not the point. The point is showing that they got through the problems at HorizonOne and that you were speaking to it earlier that horizon two and four are a lot smoother. And I think that answers part of my question earlier, why we haven't seen more in children specifically. If you, in the post that I had provided, there's 750 megawatts total at Childress and 300 of it gross that I'm speaking, not the critical IT, 300 of it's dedicated to that first deal. And they've committed to delivering 60 into early 2027 with the NVIDIA deal and another 50 by the end of the year. So that 50 and 60 is 110 plus the 300, that's 410, that leaves you 340 that still have to be allocated. And I guess part of the problem is until they finish that 300 and they have 2000 people on site focusing on that, and that's the number one mandate is delivering that on time, it's hard for them to move forward. on these other projects, albeit retrofits that will take a lot less time. But my gut feeling is the reason they haven't signed these sooner, I'm hoping they're working on it right now, but they were trying to work through the production problems at one through four before they bid off more and put their operation team in a worse position by having more contracts that they had to answer to, but there's no doubt the environment's favorable and we're expecting. I would also see if we talk to Dan soon, if he would commit to any language on when they're delivering GPUs. So there's a lot of things I have questions for him for, but I think he's open to seeing us again and we're working on that in the background, but I'm trying to think what other topics, Mark, if there's anything we haven't touched on. I had a post actually up there that McFly had put side by side in Australia the different roles related to anthropic and iron specifically and how they overlapped each other and appeared to mimic each other. And that was kind of more circumstantial evidence of a potential partnership there. But what I'm confused by, and Agrippa had spoken to this in his analysis, was we're spending millions of dollars to advertise in Australia, yet they didn't commit to any anything being built there in 2027. And my working theory on that right now is Dan went on this rant and I can post it, but essentially he was just telling the Australian government, whether it was local or federal, that they screwed up big time. And he's listed on the NASDAQ when he could have been on the Australian stock exchange and that they're behind the ball and it's because of bureaucracy essentially, and that they can't move fast enough to be a primary space for AI. And then we see Needham, I believe it's pronounced, but that research firm speaking to imminent approval of grid connections in Australia. So I can't help but think that he was just trying to light a fire under someone's *** in the Australian government and kind of press them to get the grid connection that he was seeking so that he could move forward potentially on something. So to be conservative, if they didn't have the grid officially secured, then even though maybe they will build something there next year, potentially, and this is all speculation on my part, but they can't do so without announcing that great connection. So that's another piece I've mentioned of like short term catalysts where, you know, Dan was in New York speed dating, but New York happens to also be home to two of his main banking partners in Goldman and JP Morgan. So there could be additional financing that we see coming soon. certainly the 50,000 GPUs that have been bought sent, you know, for the better part of three months now that may or may not have a partner designated, but haven't officially been contracted yet. The elephant in the room is Sweetwater. But yeah, I think there's a lot working in the background and it's just a matter of how they want to release it or what they're prioritizing at this moment in time. And, you know, we're still waiting. So waiting and watching attentively. I do think, though, that this address, the GTC, we're not going to be able to cover it tonight. Maybe we'll probably cover it next week, maybe midweek if small cap does a space on Wednesday night, but I would be really encouraged if Jensen speaks. I mean, I know there's a panel on AI factories. I just, I don't know how you can talk about AI factories without mentioning Iron and Sweetwater. And I think if he mentions it in that keynote, and certainly that panel, that that's also a potential short-term catalyst. But people have to start, it was as simple as taking a photo with Michael Dell and Jensen and Dan being in the center of it, like we have to start to track those institutions that have been accumulating in the 50s here. There's an execution piece of it, but there is something to be said about just being in the right room with the right people. And I think if we can associate ourselves with Nvidia and Dell, that more people are going to be willing to look past the numbers short term and take a bet on this business in the future, given the compute and power constraints that are plaguing the AI market. But Mark, were there any other topics that related to Iron that we haven't touched on that come to your mind that you want to speak to?

Mark: I don't think so. I know there's speculation that they're going to have Rubens. The Vera Rubens now change, they're going to change out the Blackwell, the Grace Blackwells, and they're going to substitute them for the Vera Rubens. I think that was for Horizon, was it not two through 4?

₿itcoin ₿utcher: Yeah, I, we had gone through, so I have in the, I think given the delivery time, I think the consensus amongst the guys who I do due diligence with in the background that three and four was a possibility, but more likely if we had to guess, given the tear down of some of the mining halls, it appears that, and Dan had referenced five and six, so it looks like you're going to have another 150 megawatts, which will be 100 critical IT dedicated to Rubin and it makes sense. Well, they're liquid cool, I should say. I'm making the assumption that it's Ruben. Maybe it's GB3 hundreds, but at the end of the day, it's the specs of the customer, which given horizon one through four being Microsoft, I think it's fair to assume that five and six most likely would go to Microsoft, but we don't know that for sure. And I think I think the reason they would do it, even if the economics aren't. It's yet to be seen with the pricing. I mean, they can renegotiate everything now, and it is a separate deal. I want to really see if they can get Microsoft to pay up this time. Otherwise, if they take another deal like they did for one through four, I'd start to question out loud the merit of that relationship. If they're unwilling to pay and we're in a scarcity driven market, like we've already shown the ability with NVIDIA to generate more money with managed services. And if we have control over the infrastructure that we're building, I, they already have the brand recognition of, I'd rather give a deal to Google like that than take another smaller deal with Microsoft unless There's plans for Australia given the investment that Microsoft's about to make in Australia. But then it's the same question we get back to like how many times with a hyperscaler are we supposed to give them a sweetheart deal? Like at some point like they should value our ability to deliver on time in a high quality product. So I that's another. I'd be curious how Dan and the team think about that. How much he would reveal to us is another conversation, but that's a long way of me saying I'm not sure three and four get upgraded, but there's five and six, which most likely is Rubin, like unless Rubins aren't available and there's this emphasis on time to compute and the client's paying for the contract anyways, and maybe they fill it with GB300s, but Those are some of my thoughts on that, Mark.

Mark: Yeah, I just they had the replacement. They had the replacement clause in the contract where they could change them out once the Vera Rubens became available. But I think Jensen's going to get what he wants. And I think that I think the days of Iron having to, you know, just take what they can get. I think those ended with this partnership with Nvidia. And that was the most exciting thing to me because of the credibility that Iron now has being partnered with Nvidia. It just it just can't be understated. You know, it's been understated too much. I think, you know, I think that Iron communities, you know, sometimes I feel like were too were too pessimistic or too down about things on some of these on some of these spaces sometimes. And it all changed when Nvidia came to play. And I don't know why it took so long. I honestly thought this partnership was going to happen before it did. And I don't know what slowed it down. And that's not really that important at this point. The important thing is that they're now partnered, I think that puts them close to the front of the line now with the distribution of the hardware. So I feel like those are going to be far more readily available now than they were two months ago. And, you know, it's as simple as I mean, it just it sounds overly simplistic, but a lot of the times you can see something coming. And you just have to be patient enough and positive enough to understand that it's going to happen. And I thought Iron and Nvidia were a perfect match in early 25. And that's really why I actually, you know, became a shareholder, because I just saw this coming. And in my mind, there was nothing that was going to stop this partnership. and it finally came about. And now I think they're at the head of the class because they have exactly what Nvidia wants. They have everything that can be the specs exactly the way they want it. They don't have to deal with Coreweave. I mean, they love Coreweave, don't get me wrong. Jensen loves Coreweave, he loves Nibius. Don't get me wrong, he loves those guys. But those guys have to deal with landlords and other other technicalities. It's just Nvidia and iron here. You know, it's just Jensen and the Roberts brothers, right? I mean, that's where we are. And there shouldn't be anything to stand in that in that way. It should be we're going to get you the blueprint. That's DSX. That's what they've talked about. We're going to get you the top of the line hardware. That's under the DGX portfolio. So that's Grace Blackwell's. It's liquid cooled 300s. It's VR 200s. There's nothing that's standing in the way now. And that's really what I think everybody at Iron's been waiting for. You know, all the Iron shareholders and everybody that shows up on these spaces every Sunday night. I think they know it too. They see it. It's going to happen. there's a lot of things you can always think of. You can always think of all the ways that something's not going to come together. But now that the partnership has happened, this is Jensen's game as much as Iron's game now. And Iron is going to succeed because Jensen's going to succeed. And it's a very, very simplistic, you know, way to put it. But they're the best. And this, all of this, all of this exists because of Jensen Wong. It all does, I mean, you can, sell, Trainium. They can, Amazon can have their chips. Google can have their chips. They're just chips. They're not AI systems. I know AMD is trying to catch up with that. Never been a big concern of mine. It never will be because as they, as all these guys try to do their best, the pie keeps getting bigger anyways. And I never listened to the Talking Heads on any of these financial channels. You just have to do your own research, and you have to dig in, and you have to convince yourself what makes the most sense. And this is what makes the most sense.

₿itcoin ₿utcher: Mark, let me ask you this. Do you still-- how is your-- without getting too personal, but are you primarily-- I don't have a problem sharing with everyone. The majority of my I hold a little NUAI, Cipher, and a few other small holdings, but I'm well over 80 to 85% in iron. Do you have a split on your portfolio and is it primarily iron at this point or do you still hold a large amount of Nvidia if you're willing to share just percentage wise? How can you have it constructed?

Mark: No, it's 80/20 now. I have 80% iron and 20% Nvidia. Yep.

₿itcoin ₿utcher: So let me ask you this. Where I see an opportunity is I've had this working theory that even if you believe Nvidia is going to succeed long term, like Iron and the other NeoClouds, be it Navius or Coreweave, are becoming the way it's getting positioned, they're essentially going to become a levered play on Nvidia. And the only thing I can think of is just people's risk appetite and Nvidia is the safer one. But at 5.7 trillion or whatever their market cap is now, I just wonder at what point when people are looking at this industry like It's not to say that Nvidia is not going to keep doing well, but if your goal is risk-adjusted returns, maybe our stock trades like bipolar, whatever right now. But if you really understand the space and what you've presented tonight, Mark, and what I've spoken to with the architecture and the DGX hardware, like if all that's flowing into iron right now and iron's a sub $25 billion market cap, I don't understand like that pair trade. I'm not suggesting shorting Nvidia, but my point is like if you're an Nvidia shareholder, like why you're not allocating more heavily to iron after this relationship or let's assume that they start getting compute online. Maybe that's your concern. Like it just feels so obvious. Like what? What's more likely to 10x is iron or Nvidia and and it's not even close. The answer is iron. And I'm just I'm wondering at what point when you have the talking heads like Chamath or any other financial influencer who think they're the first person to realize that there's a power shortage and that data centers aren't being constructed fast enough where they look at that valuation and then you're like, oh, these guys are projected to have roughly $4 billion of revenue. And then this guy from Detroit says it can go to 14, like I'm not expecting them to take it at face value. But at least call ******** on it and then look at it and then realize that there's truth to what we're saying here. I like that's what I just have a hard time understanding. Or you're going to have this IPO in 12 days with SpaceX and what SpaceX is doing. Like one of the main things in their IPO documents was that they were leasing space out to Anthropic because XAI hasn't had as much usage with Grok as they thought. And yet that's somehow that company is going to be valued at a trillion dollars plus, yet we're still $25 billion. Like I just, we're definitely in an echo chamber in a bubble. And I'm just waiting for that moment where people, it is pretty crazy that we're $25 billion and have only shown quarterly revenue income. Our record is $33 million. So like both things can be true, but like the moment that Horizon shows up on the income statement and they're showing 250 million a quarter. And like this is a billion dollar plus business with a huge portfolio where it's super hard to secure power right now. And their competitors are having to go behind the meter and use Bloom Energy and spend an extra few million dollars per megawatt. I just I know that you were saying earlier, it's just about having this general sense of optimism and a vision for the future, but there will be a day and we're going to be talking about how much of a joke this was that we're trading at $60 right now and I just buy as much as I can. in the interim while people still misunderstand this. And that is the alpha. And when these spaces, instead of I'm looking at how many people are, we have 335 people in here right now. There's going to be a time in the future without Mike Alfred that we're going to have two, 3000 people here on a Sunday night because we have the most popular stock in the stock market. And I believe that to be true. And I just kind of put my head down and keep working towards that day and prepare. And there's too many talented people in this community. If people start looking, you'll be amazed at what they find and just like why I want to almost print copies of Agrippa's novel and just go to every financial advisor I know and be like, please read this and tell me after reading, you know, investing two hours or whatever it takes in this, that this isn't a hundred or $200 billion company. So kind of stuff that we already know, but just, I appreciate you coming up on off the bench tonight where I was supposed to, I'll give Rosenberg **** later for that co-hosting with me, but you've been fun to talk to here. And I think we both share that same sense of optimism long term. And Nvidia, like we already won. Like it's not, it's what's the final score going to be? And we have to wait in the interim. And then there's some of us who try and time it with options or get cute with covered calls, but The reality is like if you're just buying and holding this five, 10 years from now, like if you have a position of 10,000 plus shares like that you're at, you know, 600 grand, like that's going to five X in the next five years. Like you're a pre-millionaire, you're pre-rich, whatever you want to call it. Oh, 600,000 is a lot of money, but it's just, it's not if, it's when. And so I appreciate you coming up tonight, Mark. And I, Everyone's kind of being shy tonight. We're an hour in, and the last thing I want to do is drag this out for the sake of it. I don't have much more to talk about. I do know Jensen's going on in an hour, and I do think that's going to move markets tomorrow, but I don't think I'm talented, but I don't know if as talented as you are, Mark, if we can talk an hour about nothing here. But I'm just hoping he says the words iron in his speech. and reference to Sweetwater. And that in itself is worth four to five points this week, in my opinion.

Mark: That would be nice, wouldn't it? I think, yeah, I just, I believe in being positive when the fundamentals are there. When you can, when you can verify, when you can kind of see the writing on the wall when you have the information, okay, just like Mike always talks about, this is the way to invest. You want to be some, you want to be early or somewhat early and you want to have unwavering conviction and you have to put a timeline on it where it's reasonable for your particular situation also. I mean, everyone in this room wants iron to be, $300 a share, by the end of the year. Well, that's not going to happen, right? I mean, I think we know that's probably most likely not going to happen. you never say never, but I mean, you got to be logical about it. So you have to have the appropriate, you know, timeline. You have to have the understanding of where you need, where you're going to be at a certain point in time and how to get there. And how you get there is you don't jump in and out of, jump in and out of stocks. I was very early on a company called Cloudflare, which is, you know, pretty well known today. But I was, very early on that. And, you know, I just wound down out of, wound down out of that in the last six to nine months. And But there have been a lot all the way. But you have to stay in and you have to continue to be positive about where you expect yourself to be. But ultimately, the specifications of the data center and what Nvidia is bringing is incredible. It is really the most brilliant and most capable technology that we're going to see. And it's going to be at Sweetwater One. I mean, I think I saw on the nest you put a you put a blanket. I think I think you were just being conservative also, I saw on that chart you had a PUE of 1.5. Sweetwater One should really operate around 1.1 based on Nvidia's specifications. So that changes profitability there by quite a bit. You know, you're talking about 20% on the expenditure piece. So you're, you know, 20% savings on energy. based on how technically proficient this DSX architecture is. And that's something that doesn't get a lot of headlines, you know, at all. And another reason they are, another reason why this AI system, this AI factory is amazing is because they have another design another architecture called CMX, and that has to do with the memory. And this doesn't get a lot of headlines, but the way that the Nvidia AI Factory is going to utilize memory is better than anybody else is doing it. It's a system. It's not pieces here and there. It's a full-fledged system, and it's called CMX. And what they do is they allow the really important high bandwidth memory that's coupled with these VR200 racks. It allows them to do the important work while they're able to siphon off some of the other memory requirements to other pieces. And that's a big deal. You can actually you can actually find that on Nvidia.com. It's talked about very articulately, more so than I can do at the moment. But Jensen's thought about everything. He's thought about where to, you know, where to conserve and where to get the most out and how to alternate systems and have and have the greatest use of power. I don't see anybody out there matching what NVIDIA brings to the table. Yeah, these ASICs, there's a place for them. But again, that's the overall pie getting bigger also. You know, everyone keeps talking about how how, yeah, Google, they're going to, they're going to corner the market on inference and no AMD is better on inference and blah, blah, blah, blah, blah. Okay, well, there's no technical specifications right now that show anyone capable of doing all the things that Nvidia's systems are capable of doing. And in many regards, the Nvidia AI system is better at inference than these ASICs and other GPUs. And that data is available as well. So it's just a perfect partnership. And this group, this wonderful group of people here tonight should be smiling because it is going to be an amazing, an amazing run that Iron's going to go on with this partnership with Nvidia.

₿itcoin ₿utcher: Let's touch on the You said CMX, and then I think previously a few weeks prior, we had a gentleman, Marcos, on, and he was speaking. Can you speak to direct liquid cooling, chip cooling? Have you seen anything on that? I mean, there was a little back and forth this week with Jim was kind of getting into it with Nabius Bull over what was being done at Horizon related to the and it was well within the GB 300 guidelines, but they were doing an RDHX heat exchange like on the back of the server that was, I think, intended for non core functions that kind of it was a hybrid cooling approach. And then there was liquid to chip. Have you seen anything like Marcos had been talking about a few weeks ago? And again, this is out of my realm, but it sounds like you've been reading the architecture pieces and the product specs a little more closely than I have. But you had spoken to I have the PUE at 1.5 out of conservatism because that's what they did for Horizon. And I think it actually makes my estimates, and I agree with you, like if we have the smartest guy in the world in AI, Jensen, designing our AI factory, like it's gonna have a better PUE than 1.5, but I also. think that when you're having a conversation with people who think we're just talking **** and that we're moon boys, that you have to have some conservative assumptions and to build an audience or at least have people from the outside willing to discuss it further. But what have you seen? You you spoke to the memory piece that they're trying to improve and that makes sense given the bottleneck, but. I'm just curious if you've seen anything else with direct cooling or anything else that you wanted to talk about that gives you hope on that not only are they going to be able to cool the chips and extend the useful life of them, but what other things are they doing to conserve power? Because ultimately, I think where this is going is Dan's favorite thing to quote, or at least in the past, was Javan's paradox. And again, those who aren't familiar, there's this misconception that if token prices go down over time, that iron screwed. And if chips are more efficient, then there's less need for power. But it's kind of stupid when you look at the growth in AI. already. It's exponential and the infrastructure is growing at a linear rate. So there's no chance of that happening. And to the extent that token costs do go down, there's other use cases that Dan's spoken to. Like right now, LLMs are primarily just tax. Like what happens when we go to audio and visual, and that's going to require more tokens to process. So all of this is Then, and that's what I think this is what you were speaking to, Mark, like Jensen's 10 moves ahead of everyone and is already considering all of these demands in the future. So all the more reason to have Rubin that's three to five times more efficient, and then you're going to have Rubin Ultra and then Feynman like. He's already like, I think Dan has spoken in 2024 about working in 2029 and being five years ahead. I would argue that Jensen's ten years ahead of everyone right now and maybe 20 years. And that's what's even more encouraging. Like certainly AMD is a formidable company. Certainly all these other guys at Google and AWS are going to have success here. If I'm hitching my wagon to one guy in the industry, you're damn right it's Jensen. Like the guy, to your point, like created the category. So outside of that praise for Jensen, have you seen other, whether it's direct liquid cooling or any other technology that beyond memory that gives you evidence to that the DSX architecture, like how that that upside and efficiency that we're not pricing in right now, because certainly we can't imagine it because nothing's been really released yet, but we can at least speculate on the direction that it's going.

Mark: There's a lot of efficiencies. I don't have them verbatim in my head, unfortunately, but it just it's the whole system. I mean, I think everyone, I hope everyone realizes at this point that, this AI factory, it's self-contained with everything that it needs for efficiency. So that includes the Spectrum X Ethernet. So it includes the Nvidia Dynamo, if I'm remembering correct, the Nvidia Dynamo, that's their brand. It orchestrates inference requests and it keeps the it keeps the caching in a different container, a different pod, if you will, away from the high bandwidth memory. So the high bandwidth memory is not eaten up by what would be a mundane task and the high bandwidth memory is handling all the high inference use case, you know, the use cases there for high inference. That's kind of what it does. So it's CMCMX is it's a they call it a pod level like a rack rack scale context memory tier. And it stores and shares reusable keys and values to cache efficiently. That's kind of what that does there. So it allows for more effective use of the GPU context memory without any kind of hiccups so they can perform seamlessly. So CMX is powered by, here, I got it, actually, I got it here now. CMX is powered by NVIDIA Bluefield. They're four DPUs, so they handle storage processing, data movement, and offload tasks closer to the network. the NVIDIA STX reference architecture for native storage, the Spectrum X Ethernet, and that's for high bandwidth RDMA connectivity for predictable caches and stuff. This is probably a little boring for the space, but these are some of the technical specs. But ultimately, this allows for five times higher throughput and up to five times better power efficiency. and lower latency and better GPU utilization. So the bottom line is with the CMX memory piece, the performance is 20 times better for time to first token. That's going to be that's going to be vernacular. That's going to be language that you're going to start hearing a lot more of in these quarterly reports, these quarterly calls by companies in the industry, you're going to start hearing about time to 1st token exponentially. It's going to be crazy how often you hear this. But again, you said it already, Butcher, it's, you know, Jensen's thought this stuff out years in advance, years, years in advance. And really, you know, the AI world that we're living in and going to, you know, continue to live in is really was created by him. So it's great to be great to be a part of that. But there's a lot of stuff that we can talk about. I'll just I'll make some better notes and prepare a little more for some of the technicals.

₿itcoin ₿utcher: Oh yeah, I I'm getting you're a hit tonight. I so we have the subscriber group only friends and they're like, where did this guy Mark come from? We need more of that. And I think that that chat specifically is about 300 some subscribers and we're spoiled by the amount of diligence that Franz puts in. And we have some guys that are very good traders and it's a good eclectic group of individuals, but sometimes everyone gets a little ahead of themselves and impatient. And I think it's good to have someone like you up and re-emphasize being positive about this. And I do believe, even though at times I ***** and complain in the short term, I think it's just because my expectations are so high because I believe and have so much conviction in this. I've never once wavered and thought I wouldn't own this by any stretch. It was just more wanting it to happen tomorrow instead of over time. And I'm beginning to get a little better about that. And but yeah, there's no doubt in my mind, like when we're talking about iron in five years, like anything short of $100 billion or $200 billion with what they're doing right now would be a disappointment in my mind. And I don't, we've said it earlier, I don't think it's if, it's just when, and unfortunately, we're all a little impatient on here at times, myself included, but someone like yourself, it's refreshing to hear that you've kind of walked the walk before with your other investments and seen them through. have been on this road before, so the ups and downs of it. And yeah, it's a good reminder that it's okay to have a little self-critical attitude where we don't say everything they do is right, but even when they screw up short term, if they're working towards improving and correcting it, which I know they are and If they're doing that, it's going to take care of itself over time. So it's. I got Jeff requesting to come up though. Let's see. Jeff, can you hear us? Jeff's still on mute, but Jeff, when you're ready, we were giving you. Yeah, we were giving you **** earlier for. ditching me, but thanks for, but Mark saved the day. Mark is my co-host despite the lack of title. So Mark has served beautifully, but welcome Jeff. How are you this evening?

Jeff: I'm doing fantastic. I feel terrible that I flaked out on you that's not like me at all. I, one, was having dinner with my mom, didn't see the text message you'd sent me till about, I don't know, 30 minutes later. And then I was thinking we started at 7 P.m. Pacific for some reason. So I was like, all right, well, I'm just going to have to take dinner to go, Ma. I left dinner, well, at her house. She made some great salmon, by the way, which I'm now eating. But left and thought I had was going to arrive, 5, 10 minutes late. And then you're like, no, ********. We started at 9 P.m. Eastern time. I was like, oh, well, I guess I'm going to be an hour late. So apologies. Sounds like you guys are having a fantastic conversation. Mark, I probably learned more in that 3 minutes from you than I have learned in the last 30 days. So kudos. That was that was.

Mark: You're too kind. Thank you.

Jeff: So I don't know where, how I follow that, Butcher. I'm certainly not the technical person that Mark is, and I'm not sure what we have already discussed. So why don't you take the next thing and I will chime in.

₿itcoin ₿utcher: Yeah, we've kind of, just to recap, but I feel like we're running a radio show here where they, or internet radio, whatever you want to call it, but some of the topics we touched on earlier tonight were the Wall Street Journal article with Gross, the guidance update to 4.4, Dell earnings crushing. I didn't actually touch on Anthropic, and maybe this is a good spot for you to come in. I believe Anthropic, whether they closed a new funding round, but in excess of a trillion dollar valuation, and that's certainly Everyone's eye is on Anthropic right now, so we could probably talk about Anthropic here in a second. Coreweave closing on Rubin, where they had a tweet that Michael Deladry posted about. I don't know whether it was burned in and now that it's producing revenue, but essentially Vera Rubin coming online, which is encouraging for the industry, especially the ramifications that it'll play into Sweetwater. And then most notably like speculating on NVIDIA GTC, that's the conference being held in northern Taiwan and Taipei and Jensen should be speaking in 1/2 hour and has his keynote and I'm personally. I just, I don't know how he can speak to AI factories and not mention the new Iron partnership. And I have a feeling that we get name dropped in that. And do they announce a deal? My money's not on that, but I think at a minimum we get mentioned and the more times if the. If the words iron come out of Jensen's mouth and we can create a clip of that and just recirculate that and pound it into people's heads, even though we have a photo with him and Dan and Michael Dell like audio that saying that iron's Sweetwater campus of two gigawatts will be the flagship. partnership property for DSX architecture that will use DGX hardware to maximize token optimization and become the world's preeminent AI token factory. Like, if I could get something like that from Jensen, like, I would sign up for that right now. And that's what I'm looking for. Whether he does it or not, it's a whole other conversation. And then lastly, the Australia piece, there was a piece by McFly and some of the job postings and just curious, like when we interviewed Dan next, like what? It's hard to tell if I think they're given the grid approvals being signaled as imminent by Needham. Like, I think there's something missing that they're telling sell side and some of the institutions that retail. I mean, to be fair, they mentioned a global strategy and they're saying 2028, but like if we have a grid approval for Australia tomorrow and maybe they just haven't formed the construction team and in their defense, we still have a grid approval in Oklahoma, but we haven't committed to building there until 2028 or 2029, but just kind of what's going on in Australia right now and then After reading Agrippa's piece, I didn't mention this earlier, but like what is their, what are their plans in Europe with respect to Spain? You know, they have just under 500 megawatts there and what, what's their plan to go to market there given the growth of AI and Europe is not, I mean, it's certainly not. It's an active market. It's not the US by any stretch, but there's still, you just paid money for this company. Like when can we expect to see a return on it? Whereas like Mirantis, day one, you're already seeing the benefits of managed services come through in the margins of the NVIDIA deal. But I don't know if Jeff was about to speak.

Jeff: Yeah, I was, I was gonna say, well, Spain is already a data center that's up and running, right? So it's just a matter of how long it's going to take for it to be recognized on our balance sheet, I would imagine. Is that accurate?

₿itcoin ₿utcher: I got to look closer at it. I know the megawatts. I don't know if it was fully developed yet, but I know there's like a construction team associated with it. I think that was kind of one of my other. This is what this is going to sound. I everyone can judge however they want. I think if I had one critique of last quarter, there was so much great news that I don't understand why it wasn't spread out. Like I think consecutively they could have built on this narrative and maybe release the Spanish acquisition now or like kind of as opposed to like, Between reading Agrippa's novel, which was like, I mean, I spent most of my Sunday sitting on my couch, my wife watching a movie, and me trying to read line by line every word that Agrippa wrote. And part of the reason it took so long was there was so much to cover. And again, it was great. And if anyone isn't subscribed to Agrippa, or at least I think he provided a free version of it, like it's to die for. At the same time, there's so much to digest and I consider myself in the top one or 5% of people on the know-how of iron just because I spend the better part of a few hours a day dedicated to it. And even I was like, I miss this. So then at the same time, when you ask yourself out loud, how is someone on the outside who's not paying attention to this, how are they supposed to digest the story? There's so much to consume. And that's where I think they're doing a better job though, since earnings and the little piece with gross and these little... I think of it like a boxer, like a little jab instead of this huge haymaker being thrown. If you can just slowly throw that jab over time and all these little singles, they add up and people generally, I think it raises the floor instead of looking for one huge announcement at earnings. And then there's like six or seven things going on that we can't fully appreciate because the most important one certainly is the Nvidia partnership, which Mark spoke and do beautifully tonight. But there's still a lot of other like, like we don't fully understand what Maranis can do for us or what the Spanish partnership or I should say the acquisition, like what that means going into Europe.

Jeff: It's about the Spanish acquisition. It was just kind of mentioned in the earnings call, if I'm not mistaken, correct?

₿itcoin ₿utcher: Just a footnote, like in Agrippa did some research on the head of the new European operation. I forgot the gentleman's name, Spaniard, but you know, his working theory, whether we confirm this or not, was that given Dan's experience in infrastructure investment banking at Macquarie back in the day, that these guys, they were doing partnerships with the renewable energy in Spain and had this relationship or at least knew who the guy was enough to walk in the room. But that certainly was never on our bingo card in Europe. Like we just assumed APAC in North America and that was it. But they're now they're going into Europe. So like there's so much to unwind here. And like I said, there's a lot of us in this, you know, 300 some of us in here that kind of live and dive at every word that's coming out of PR management or what France or whatever. And like even us, we're kind of like forgot to talk about that piece because there were so much thrown. So I just I have to have context sometimes and take a step back and be like, if we're all this financially, emotionally, whatever you want to say, invested in this and we're struggling to comprehend all this and the greater big picture, like your average investor has no, no ******* chance. So like that's an opportunity long term and I think it's going to be self-evident and reveal itself. But yeah, there's just been so much going on and now we're kind of waiting here and the, you know, the tweets of one deal and all that are kind of coming up now where some of those things from earnings might have been. Better spread out, but I mean, I can't be upset at the moves themselves. Like it points to us becoming, if we're not a hyperscaler, we're going to be the biggest and baddest neo cloud or duke it out with Nabius and Coreweave. So yeah, those were just kind of like after reading that today. for the better part of a few hours and kind of skimming, to be honest, towards the end, just from like reading fatigue. And I apologize in advance and grip up, but like I'm gonna have to read that another two or three times to fully appreciate the depth of that and how much hard work he put into it. But that's what inspired me after reading that to look back and update a few assumptions that I had in that little napkin math. ARR model, like I still need to do share count and some of those things. But I think the point that everyone should take away is Nvidia is going to take us to the next level. And I think the Nvidia deal at Childress sets a floor for what we can expect for B300. It's like the only way we get, we sell a GPU under 377 is if we sign Anthropic. And even then, I think Anthropic is so flush with cash right now and so desperate for compute. I have a hard time believing we can't get 377 or higher. But did you, Jeff or Mark, did you see, I can look it up while you guys comment, but I don't have any details on the latest funding round that just closed with Anthropic. But if any of you guys had thoughts on what I said or Anthropic in general, feel free.

Jeff: I don't know the numbers off the top of my head though.

Mark: I know it was massive. It was 65 billion. So they're now the now another, it's now 99, 995, 900 ninety-five billion. So 65 billion now at a 900 ninety-five billion valuation.

Jeff: Let's just call it. Let's just call it.

Mark: Yeah.

Jeff: They made it 9 ninety-five to make investors.

Mark: Exactly. Discount.

₿itcoin ₿utcher: Yeah. So I'm looking at their blog posts. They're saying 65 in their series H at a 965 post money valuation. I don't think they've issued anything newer lately, but the latest number I remember was as much as I think $40 billion annual recurring revenue, which they had just raised from 30. So those guys keep growing and it's, I don't know how it's possible when they're growing that exponentially and we're the only ones who seemingly have, it's not to say we're the only ones they can do business with, but given their needs, how we don't do business with them eventually just feels almost mathematically impossible. given the lack of compute that is coming online right now compared to their growth or at a minimum, if they're eating up the rest of everyone else's compute, that means everyone else flows to us. So overall bullish for the industry and peers. Open AI's newest Forgot the name of the newest model, but that's kind of closing the gap on Anthropic again. And that kind of rivalry is hitting head to head, but it does appear that Anthropic just has a better grasp of the enterprise market at this point in time.

Jeff: Well, yeah, I mean, so I said this a while back, right? So ChatGPT really started as a B2C model, and now they're trying to break into the enterprise where Anthropic targeted the enterprise first. and it became more of a consumer product afterwards. I've said it in the past that I like the Anthropics business model better because it allowed them to tie up big money for long periods or multi-year contracts, which is huge, right? It's but what I wanted to actually touch on and I don't think it's getting enough attention is perplexity's new AI system. Computer, which is, have you heard of it, Butcher?

₿itcoin ₿utcher: I'm not familiar. Enlighten us, please.

Jeff: So Perplexity Computer essentially is 1 subscription and it kind of re, it's really hard to, kind of hard to explain. So it, basically redoes a computer, right? So it gives you subscriptions essentially to all the other major models. And you can think of this as like a overarching AI agent where you tell it what to do and then it will leverage ChatGPT for what it feels like ChatGPT will do best. It will leverage Anthropic for what Anthropic might do best. It will leverage Google Gemini for what Gemini might do best and it will tie them all together and have them work in conjunction with one another. The reason why I brought that up is I'd be interested if anthropics compute needs are going to go up exponentially because of that as well. Really, really cool. Highly suggest you check it out. I might be explaining it a little bit wrong or I might be a little bit ignorant to it, but it's It's pretty fascinating. It's essentially 1 subscription that gives you subscriptions to all of them. And it's Perplexity Compute is kind of an agent that sits on top that will bring in all the other different LLMs as needed. And it will pick and choose which LLM will perform which task best and then kind of brings them all together so that they work in unison. Have you heard of this, Butcher?

₿itcoin ₿utcher: No, I mean, I like the concept. I think The flaw in my approach, which I'm trying to work on right now, is like I had someone this weekend pitch me on a small boutique consulting firm or startup, whatever you want to call it. And the truth is like I'm not I use Gemini through Google search and I use graph because I have X. But I still haven't even used, you know, Claude or like, and then I'll see on Instagram reels, there's a few AI things I follow just to like learn a few new terms or kind of almost have awareness and hear of certain different LLMs that are out there. And a lot of times. They'll ask like a rapid fire, like what's best for financial analysis? What's best for whatever? And if that's automated through perplexity, that would make my life a lot easier.

Jeff: You know, like you absolutely should because you get subscriptions to all of the big LLMs and it will choose which one is going to be optimized for the specific tasks. that you want and it may leverage multiple at one time and create an agent on top of it that is essentially like an orchestration agent pulling the strings and making each different LLM, you know, perform their specific function in conjunction. It's called Perplexity Computer.

₿itcoin ₿utcher: So whatever happened, what's your understanding? I kept hearing in the background that There were, everyone kept talking about a potential acquisition by Apple of perplexity and incorporating it into Siri. Do you have any, you're pretty versed in the space, Jeff, but what Apple's plans with Siri are at this point? 'Cause it just, they have the most popular phone, I mean, certainly Droid or, you know, it's not the only operating system, but They're still cache to the Apple brand and Apple phone, but at the same time, like it just feels like they haven't. I know that they had a deal with, I believe, Google, and I don't know how that works with cloud services, but have you heard anything with Siri or Perplexity?

Jeff: When you open up at Apple's native Internet browser, it will open up to Google. That's what that deal was. In regards to the LLM, there was a big rumor that they were going to purchase Perplexity Incorporated and Redo Siri. That has not happened. There has not been any any new news on that front. I don't know if it's dead. It's in the works. However, I do think if they are going to be purchasing Perplexity, the price for Perplexity just went up exponentially because this new platform computer is it's pretty remarkable. But no, honestly, that's the trillion dollar question, right? Are they going to do it in house or are they going to buy? I would think that they are so far behind at this point that purchasing something and incorporating it is probably their best move and only move. Siri is so far behind I don't know if they can catch up in this day and age because everything is moving so fast and getting so much better. I think they're kind of I think they've kind of pigeonholed themselves where they're going to have to look for an acquisition and who that is going to be, whether it be perplexity or whether they just partner with somebody else for a ton of money. You know, I think that is a trillion dollar question, legitimately a trillion dollar question, and I don't have the answer to it.

₿itcoin ₿utcher: I don't know how they'd stay relevant without making that move. I mean, it just, yeah, it's a nice cash flowing business, but.

Jeff: We purchase the apps in the App Store, right? And just, and put it onto the phone. The hardware is what people want, right? People are buy Apple for the familiarity. And that's enough to keep them relevant. But And plus, like I said, you can add any LLM system into it, right? But they really do need something like inherently built into the phone.

₿itcoin ₿utcher: It just feels so natural. And I don't, I, you know, jobs has been gone for some time and Tim Cook had a lot of success, but I mean, what's their biggest innovation? I just feel like I bought a new iPhone 17 mainly because the, they wouldn't, I had a 14 that I, for the longest time wouldn't upgrade 'cause I didn't wanna go to the Verizon store, but okay, the camera's a little better. Like I, you know, I just, they have the Apple Watch, like I've never felt the need for that or the, and now you have companies like the, I think my best experience with Apple more recently is the Jon Hamm show, my friends and whatever it's called, but that's the point. They're focused on television when it just feels like they've been MIA the past six, seven years where AI is just these formative years where I just feel like if jobs were around, they wouldn't be in this position right now and they can brag about all the shareholder buybacks and the return on equity and all that. But at its core, it doesn't feel like they innovate too much anymore. And and that's just the opportunity for some of these other guys to disrupt and take their space as the most valuable companies in the world. Like, that's not to say they won't be around, but they're just not going to be as relevant going forward. And everyone's going to be talking about companies that are on the rise, whether it's, you know, you got the IPO of SpaceX, you got Anthropic and, and both those guys have a long ways to go, but at least they're trying to push the envelope. And I just, I don't see that with Apple right now.

Jeff: Well, I don't think they're going to be going anywhere and I don't think they're hurting, right? I mean, you still have laptops. They're the dominant tablet player. They, they own a huge chunk of the cell phone market. They have Apple TV, all the accessories that go with it make them money. The big thing with Apple, right, is if you kind of analyze them, they are not a company that traditionally makes big acquisitions. Like they are pretty stubborn and like to develop everything in-house, keep it closed vest. And so maybe their stubbornness kind of got in the way and of them getting out in front of this and purchasing one of these companies when they really should have at an earlier age. And now it's going to cost them a lot, a lot of money. And if they don't purchase them, they're going to, a partnership's going to cost a lot, a lot of money. Although someone like Google may pay them, right, to ensure that, hey, if we're going to be the landing page on the native web browser, right, let's link that into Gemini. So that could be an easy fix too. I mean, I think it's going to be one of those. I think it's either going to be some sort of partnership with Google that's expanded upon from what it is today, or they're going to have to make an acquisition. I don't foresee them being able to compete by building an LLM in-house at this point.

₿itcoin ₿utcher: Maybe bringing things back to Iron for a second. I think where I saw it, where's stock analyst pros in here. Maybe someone else with a tech background like him can come up. But I do want to touch on. I think there's enough that Iron's dealing with where they don't want to bite off more than it can chew. And there's enough to incorporating Mirantis into the company. But long term, if Iron's going to become a hyperscaler or if that's their vision, I've seen the, I just, I don't have the technical wherewithal, but I can at least ask the question out loud, like where are the gaps in. Mirantis and Irene as one, like, what do, what are the new pieces that long term that will enhance the customer and enterprise experience that they need to incorporate into their business? Like, are there additional acquisitions? Are there features that they have to maybe invest in the Mirantis team to build out so that they can compete head to head with these guys? Because I think earlier I touched on it, Jeff, you weren't on, but it appears that horizon five and six are a go for liquid cooling based off of the earnings call and Franz's research, which I think it's a fair assumption is most likely Microsoft. And there's been talks of hyperscalers potentially as tenants at Sweetwater, but I'm kind of at the point where I don't want to do business with hyperscalers just to say that we do business with hyperscalers. Like, I see what the opportunities are in the margins. Now, you could argue, do we use them short term to basically de-risk the infrastructure, maybe four or five-year deals, and then at the end of that four or five-year deals, you can widen your margin. And there's certainly advantages to having brand recognition, the credit, but I just, my fear is that we, I hope we're past the stage where we have to give this discount just to say that we do business with a hyperscaler. And it looks like the company's already, by emphasizing Frontier Labs and Enterprises is already kind of, Hinted at that direction where it's like if the hypers don't wanna pay up like we're willing to do this with others. And so that's kind of an open-ended question for the crowd or anyone else. But I I know like like Nabius, you were former shareholder like they've had acquisitions related to I believe it's search or. I don't quite understand it as well, but like them or Corwi's software layer, like what is Mirantis missing that Iron has an opportunity to add on in the future to enhance the margins even further?

Jeff: I think there's two parts to that, right? And I don't have the answer to that question. And I think a big reason why I don't have an answer to that question is because we're not, it's really, it's been kept in the dark, right? What Iron's software capabilities were prior to Miranda's, right? We've been told that they have software capabilities, but that hasn't really been expanded upon. And until we know like what they had already in-house, where they can plug in holes, right? It's hard to know where the gaps are. Like we have a good understanding of what Mirantis does, how it helps them from a Kubernetes perspective. So that's one thing, right? Like we need to know like what Iron was already capable of. And that's really hasn't been disclosed in detail as far as I'm aware. The second thing is though, and this might take it off on a little bit of a tangent, But I think it's a little bit dangerous to try to go after the enterprise space with the software play on top. And the reason being, and Stock Analyst Pro I see is on here, who's from Microsoft and could probably comment on this. Look, if you're going to try to go and compete with the Azure's of the world, the AWS's of the world, these guys have been in the game for so long. There are so many integrations they have. They have so many different software pieces that they can layer on top of their solution. I don't know how Iron competes with that because these companies are also so big, right? That if a deal got to, let's say, a billion dollar deal with the enterprise space, someone like Microsoft or AWS could go take that as a revenue play, undercut Iron, and have a more robust service from a software perspective. that I think that would be extremely challenging for Iron. Now, I could be wrong, but I do know how robust Azure is. I know how robust GCP is. I AWS, right? Like we're not going to be able to compete with them from a full package perspective. There's there are new solutions being added to AWS like weekly. They already have thousands of different things that you can do, right? Like things that we won't be able to compete with today and probably not for the foreseeable future. So that is a worry for me, right? I think where iron is really going to excel is in the bare metal space. If we start trying to to compete with these bigger boys from a software perspective and like all the bells and whistles, I struggle to see us being successful going head-to-head with Azure or Microsoft, given the amount of services and software support that they can layer on top of it.

₿itcoin ₿utcher: Yeah, it's just something that requires more research on my end, but unfortunately...

Jeff: Well, I don't think there's any research we can do, right? Iron just hasn't disclosed what the...

₿itcoin ₿utcher: I think what there's an opportunity. I think one of the interviews that changed how I thought about Iron was maybe we can have Brian Fry back on. But Brian Fry, like last year, I forgot when we had the interview, but that was kind of like a turning point for me where I initially got into it because I was a legacy. I mean, I still own a small piece of Bitcoin, but most of it I sold to get into iron. But yeah, from a Bitcoin play into an AI play, it was like, as I started looking at the power shortage, but then just like talking with Brian, the one space I had with Franz last year, that was kind of, so I think someone from the industry, or certainly if we could get, the people from Mirantis are pretty, Responsive. So if Dan, you know, now that it's under Dan's domicile, if Dan would agree to it. Certainly when I talk to Dan above anyone else, but even if we spoke to someone from Mirantis to better understand. How they add value to iron and actually walk through when it's one thing for me to read bullets that say they. to infrastructure as a service, you know, bare metal implementation or platform as a like, what does that mean? And can they go head to head with those guys or where? And of course, they're going to give the bullish end of it, but that would at least be a starting point. And then people with that background might be able to ask them questions and we would get a better feel for it. But that's the only idea I have top of mind right now.

Jeff: Just to give you an example, right? So AWS has a software service called SageMaker, right? SageMaker is used to do video analytics. The NFL uses it and they use it to assess. They pretty much watch SageMaker watches all the games and can like assess risk to players on specific plays. and things of that nature, right? So these are like how complicated and how complex some of these services are that can be layered on top of these already hyperscalers, AWS, Microsoft, right? Plus you have the Azure marketplace and the AWS marketplace where you could purchase SaaS solutions to layer on top and run. on that infrastructure, right? So you pretty much have everything and every software service at your disposal through the marketplace already, plus all of the other in-house services that they've designed specifically for these specific platforms. So like, that's my concern, right? Like Kubernetes is one thing. I just I think where Iron's going to excel is delivering power in that raw compute. If we try to get into a battle of offering an array of services and all these bells and whistles, I just I don't know if anybody compete. It's not something it's not against Iron. I'm not, you know, and I don't know if anybody could compete with that at this point.

₿itcoin ₿utcher: Now, I would have to reread it. And again, it's I think Agrippa kind of addressed this, that point that you're speaking to where NVIDIA via CUDA, that platform, and they have tools developed that can be layered on top and that can be sold as the package. But again, I- It's really.

Jeff: Software that helps the chips run better. It's not like, like software to run video analytics on things or, they're not business intelligence and analytics solutions that you can run on top of the compute that off islands providing that that's kind of what I'm getting at.

₿itcoin ₿utcher: Gotcha. But I think I don't have an answer for it. It's something self admittedly I have to research a little more and get better versed on instead of wasting everyone's time here on it. But it's that's kind of my open question. Like I think it's a fair question. Like if you're talking to Alfred or anyone else who mentions the word hyperscaler, like that's if you are going to be a hyperscaler, you have to be able to compete against them. So what are the holes in their game and what are they working on? And there's so much like before we get to that point, that's like, probably a few years from now, to be honest. But I know that they're thinking about that in the background when they're talking about multi hundred billion dollar valuations and even trillion dollar valuations. At least Mike is. But I mean, I don't think Mike is the only one who's thinking that behind their doors. And I don't think the moves that they're making right now are indicative of a company who's just happy selling No, but I think they're.

Jeff: Going to have to target really mature companies that already have a big IT presence in-house, right? Have all of their niche solutions in place where they can pretty much just bring it and layer it on top of what Iron is providing. That's where I see them being successful. If that makes sense.

₿itcoin ₿utcher: Yep. So we're we're two hours in that some of the guys were trying to watch Jensen instead of listen to me speculate. So I'm probably going to wrap this up here.

Jeff: Wait, hold on. I think Mark was going to chime in and I didn't want to cut him off and be rude.

₿itcoin ₿utcher: Well, yeah, Mark can make closing remarks here. And Jeff, do you have any closing remarks? But certainly my closing remarks are appreciate everyone's time this week. I'll leave it to the two guests on stage to wrap things up here, but unless there's reason to continue the conversation. But Jeff, thanks for joining. Mark, you this week's MVP. Hope you come back up in subsequent weeks, and we enjoyed your perspective.

Mark: Well, thank you very much for having me. I enjoyed getting to talk about NVIDIA and Iron together. I think something we can look forward to in the next couple of weeks. We can talk about, you know, Nvidia's reference architecture and just where, you know, Nvidia, Mirantis, Iron, what those three players, how they come together and how they begin the journey towards becoming, you know, a real player in, you know, for enterprise. I mean, again, remember, Mirantis does have 1500 clients, enterprise clients already, not all of them are going to migrate to, the iron power structure. That's not that's not likely in the short term, but in the long term, I think that'll be the case. And then, yeah, we can we can talk more about how Nvidia's reference architecture and everything they offer. will be why their partnership will put them in, good shape in the space.

₿itcoin ₿utcher: All right. Well, thanks again, everyone. We'll see you next week. We should have Franz back and hopefully we got a higher share price to. talk about after Jensen in the next minute or two speaks about a potential dealer at a minimum, mentions the Sweetwater AI Factory. We'll talk to everyone soon. Thank you.

Jeff: Have a great night, Aaron. Sorry again, Bud.