Transcription
Hello everyone and welcome back to the world's best investing podcast. I just sent out my iron Q4 fiscal year 2025 earnings digest. I now believe it's very likely we're looking at a kind of Tesla for data centers. Why would I say that? Because Tesla's success over the past decade has come down to one thing primarily, which is figuring out how to build a machine that just turns out increasingly autonomous cars. With all the complexity that goes into that, all the software, hardware, automation both at the factory and at the level of the car, all of that, it's just a factory. There's just a machine that prints out factories and makes cars. Iron is now, as of this quarter, quite visibly, at least for me, a machine that just churns out data centers.
Um, why do I say that? One thing primarily which stands out in the whole report which is that iron actually printed positive cash from operations. This coincided to them temporarily forcing the expansion of the Bitcoin mining capacity. But it basically shows you three things which is one, the infrastructure they've deployed to date is capable of producing cash. Uh, which means iron's repeatable data center infrastructure actually works and it also means that most likely we're looking at an extraordinary management team. Just the scale of the complexity of their achievements I think cannot be overstated. Uh, in 2025 here are some of the numbers. 150 megawatts of additional contracted grid connections. This is fiscal year 2025 by the way. 550 megawatt uh of data centers in operation. 400% increase in Bitcoin mining capacity and 132 uh% increase in AI cloud capacity. All of this stuff, they stop the mining capac, they stop the mining expansion capacity just for a little bit and cash from operations just shoots up. Uh, capex keeps going up. In the written form of the update, you're actually going to see it as the line continuing to trend down, but that's because it's from the perspective of the cash flow statement. But then capex continues going up, but then free cash flow per share also inlects up. Meaning the the underlying infrastructure produces cash, but then they can continue investing in new stuff, which this quarter was essentially for the a, the uh buildout of AI data centers which inherit from the original uh repeatable blueprint that uh that they had at the original site and then they produce more spare cash per share.
This is worth a billion words. This is how I unlocked um conviction with him for example. I saw them doing something which was really quite difficult. Um, you know, the whole thing they have going on, I think they have a moat which many people, many investors actually think they don't. And then as soon as that produces cash, it's real. So that's what's happened with iron and I'm very close to unlocking conviction. The only thing that's really holding me back is that most of the revenue now comes from Bitcoin and so as many of you know, I'm not quite on board with the thesis. I don't think it's a productive asset. I know it's debatable philosophically. I'm very much on board with what it does. I'm, you know, the whole sovereign individual thing. I'm an absolute practitioner of that. I'm just not sure of how that thing goes up in value. Actually, it's it's about someone coming along and wanting to pay more than you did fundamentally. It's in essence a greater fool theory. I'm open to being wrong, by the way. Happy to be proven wrong. So, you guys let me know in the comments if you believe I'm wrong.
But most of the revenue comes from Bitcoin. However, the AI business is picking up quite a bit. Uh, management shared some remarks during the call in which they basically said that they believe the contribution of the AI business is going to be rising over the couple of next couple of years. My impression is that the AI business could get as big or even bigger than the Bitcoin one in say the next two years. Maybe one year would require extraordinary execution. I'd say maybe two years. These guys have a solid track record of just constantly doing what they said they were going to do and actually overd delivering while underpromising. So the thing, the um, their deployment, they they are advancing correctly in their new data center which runs on liquid cooled technology. It's such a scarce thing right now in the world, liquid cooled uh compute capacity and they seem to be, according to the remarks, they seem to be executing well. Now, I wouldn't take management's remarks at face value. Usually, I don't do that just for any company, but these guys just have a tremendous track record. Um, since my original deep dive, the stock is up threefold. So that's a little bit of a painful uh increase in share price without me being inside it. You could again say I was maybe wrong to to not um not identify I think the extraordinary nature of this management team early enough. Many of you guys have, especially many of many people on X. So congratulations to you guys for doing this. I now, as I said, I'm very close to unlocking the conviction, but this thing uh relies on Bitcoin a bit too much right now.
Total debt is outpacing total cash by $400 million right now roughly. Uh, but this gap now seems a lot more manageable. My point is, I guess having seen the work they did when I did the deep dive a few months ago, like the their ability to deploy compute successfully really quickly when no one else can do that was evident. But my observation was, it's very easy to turn this operation into a financial hell and they haven't done that. They're printing cash. All of a sudden my perception of the balance sheet changes because now they're actually printing cash. They may go back to not printing cash if they want to continue expanding. But now I know that fundamentally the infrastructure they are deploying is healthy. Right? So that's why I believe it's a little bit like Tesla for data centers. Simultaneously what we have is obviously not much compute capacity in the world relative to what seems to be a kind of glitch in civilization in which the AI scaling laws continue to work and accelerate. And so it looks like we're positioned to see exponential an exponential rise in demand for compute. It would be like, I think our brains are wired, as I've said many times, for linear thinking. So it's almost hard to believe, but um, I think it's happening. I think it really is going to continue accelerating. And so the combination of this repeatable blueprint with what seems to be like infinite uh demand right now just starting to take off, I think is going to make this a lot, I think most likely it's going to make it a lot more valuable than the current market cap of 12 billion. It's it's not about the capacity they've installed to date, which is already remarkable, 3 gigawatt of uh, you know, the whole operation, but uh, it's more about the blueprint. The blueprint is only getting started.
Um, it was very bullish. It was very interesting to see them having to scale down the density of the racks in Horizon 1, which is the one that runs on liquid cooling. 200 gigawatt is what you need to run the latest bleeding ass GPUs. And it seems that actually customers were demanding lower densities. It just shows you these guys are operating at essentially ahead of the curve and two, have this inherent flexibility to modify their infrastructure. So it's the blueprint. It's a little bit like Palantir a few years ago. If you believed AI was coming and you sort of understood that digital twins were going to be the central asset in capitalism essentially, you were right. And that's how you're now up 20, 30x. It's a very similar kind of situation here. And I mean this for me, it's it's been a it's been a remarkable milestone in the thesis, like producing cash like that is basically everything that I needed to see. Now again, the only thing that's holding me back is the Bitcoin thing. Uh, you know, I'm just not on board with that just yet. But their repeatable blueprint for data centers, I see it now. I see the demand. I didn't even look at the valuation multiple right now because I'm not thinking about buying it just yet. I will be looking at that. But just 12 billion, you know, in the face of just infinite demand for compute, no one else being able to do this and these guys just managing such tremendous complexity so successfully while producing cash. I think this is real. I think this is very real. The stock is up big time today and I think rightfully so. So, uh, I'm really going to think about this. You guys know I have just six stocks. This could be an addition. This is sort of I I told you guys a bunch of months ago, I think, that I had some I had a good degree of conviction with Wise. Uh, it's not quite at that level, but perhaps just below that. And this is something that I could definitely add to my portfolio. So, uh, so congrats on all of you guys that got this right. Uh, and I would say if you haven't bought this yet, the way I'm thinking for me is I think this is just the beginning. I mean, 12 billion, infinite compute capacity, repeatable infrastructure. So, quite remarkable really.
Um, coco Daniel Roberts was asked during the call if Iron was ever going to focus on software. And I'll just cap off the video with that. He basically said that their customers want their own software. So to to me, there it was very clear that iron is actually working with much bigger customers like hyperscalers and stuff like that. And Nebius is looking at something else. When I studied Nebius this summer, the impression I emerged with that I emerged with was that Nebius cloud could become the kind of software that actually iron customers do end up using. Right? So Daniel Roberts was a little bit dis, not I'm not going to say dismissive about software in the call, but a little bit. And and so it made me understand that Nebius could pull off a kind of innovator's dilemma sort of disruption to everyone in the industry, including iron eventually. He was he kind of was like, you know, our customers just want to use their own stuff. They're very sophisticated, they have their own software. And you, you could just sort of see Nebius cloud iterating over time and becoming something that enables um enables customers to make a much more efficient use of data centers than anyone else could. So that's one one thing worth watching. And then the other one, the rack density that I talked about, that was very interesting to see. So to cap off, as I said, close to conviction, I think this could be a massive exponential success story. It's kind of overwhelming how many exponential opportunities we have in the market today. A lot of these companies you look at that people on X surface because there's a lot of smart people there, in my opinion, really have, you know, extraordinary management, infinite runways ahead, highly repeatable sort of franchiseable value creation processes. So, so fascinating. It's kind of overwhelming. But uh, anyways, guys, thank you very much for joining me. That's it for today. I'm going to continue watching this company very closely. I think it's absolutely fascinating. If you enjoyed this video, could I please ask you one favor? Could you please share this with someone else whom you think will enjoy it because these deep dives and updates are for free and so the only way this grows is with your help. Thank you very much in advance.