The New Tesla Is Here: Here's the Harsh Truth for Investors

Cold open

Phil Town: [Clip from later in the episode] You're in third place after qualifying. You're right behind that guy. You come around the track, they throw the green flag, and your car gets incinerated. [laughter]

Welcome and introductions

Hunter Hoffler: Well, welcome back, everybody, to the Rule One Podcast. Joined today by Austin, Travis and Phil Town, New York Times bestseller. We are all analysts for Phil and have about 50 years of investing experience between us.

Hunter: This week, pretty cool topic. We got to see a flying car, something that everybody thought they were going to see just 10 years from now. For 50 years, people have been saying, "In the next 10 years, we'll have a flying car." And now we do. That's Elon Musk's Tesla Roadster. They gave an amazing demonstration of what it's capable of, in a collaboration with a bunch of technology from SpaceX. So we're going to get into that and what that might mean.

Hunter: And then later on, we'll talk about some of these off-balance-sheet obligations from these hyperscalers that look nearly debt-free. We think they look like great Rule One businesses if we can get them on sale, but there are these little off-balance-sheet obligations that show up in the footnotes that are potentially very troubling and something that we need to keep in mind.

The Tesla Roadster demonstration

Hunter: So with that, let's start off with the new Tesla Roadster. This demonstration was amazing. Elon, with some of his engineers, showed this vehicle go up a ramp and invert itself and hang upside down, which everybody thought was the end of the demonstration, which was amazing in itself. And then they proceeded to hang this thing upside down, launch it upside down off of a ramp, have it spin around in the air and land itself. So we have a flying car. Later, the same Roadster jumped over all of the Tesla models in a flying-type capacity, and it was pretty amazing.

Hunter: So, guys, let's kick it off. What do you think this means? Big question. What does this mean for Tesla? Is there going to be any sort of actual applicable technology that comes out of this that people might be enjoying in their vehicles in the next 15, 20 years?

Austin Bowen: It's got these fans on the underside of it that create up to 10,000 pounds of downforce at a standstill. I mean, it's amazing that it can still roll, right? How does it even move forward if it's being sucked into the road like that?

Austin: Yeah. Well, think: 10,000 pounds of downforce is about 4,600 kilograms. That would be the equivalent of pulling two F-150s behind it, you know? So it's not that it would want to do that, but…

Phil: Whoa, look at that.

Travis Parker: I legitimately thought this was fake. I really did.

Hunter: I think everybody did.

Phil: I didn't even get that far. As soon as I saw it go upside down, I'm like, "Oh, somebody made an AI thing."

Hunter: Yeah, that's exactly what I was thinking, Phil. It was like, AI, somebody's tricking us. I'm not buying into this. No way. Yeah, it does drive. It reminds me of those Hot Wheels ramps they used to have that would spin the car around in a loop. I'd always aim it at my brother's head and try to hit him with it. Those things are dangerous little flying projectiles. That's exactly what it reminded me of.

Hunter: But yeah, a lot of you guys in here all race. And the first thing that I thought was, man, how cool would that be? I think what they said is that it increases the downforce on the vehicle to over 2 Gs, where with the best kind of setup and tires now, you're getting about 1.6 Gs worth of grip. So does that mean that you can essentially hit a corner 20% faster than you would have been able to before?

Austin: It's probably going to be a significant amount faster. He did say that it's going to break every track record out there for a production car. What most people don't realize is that even though the Tesla S Plaid is faster than any production car today, it hasn't even come close to beating any of the track records. That's because it weighs 5,000 pounds. In a straight line, it crushes cars like the Lambo or the Porsche GT2 RS, but when it comes to braking, weight matters a lot. If you're a heavy car, you can't brake as quickly. And then also cornering: you can't corner with that weight at the same speeds.

Austin: So this is going to enable Musk to build an EV that's lighter, with huge downforce, that can brake faster but also accelerate much, much faster than any gas production engine. So I think it's going to be really incredible to see what happens.

Downforce, wings and racing

Phil: The current technology on downforce, you can see it in F1. They've got the F1 wing, and in a regular car, you put a wing on the back, and the bigger the wing, the more downforce. One kind of car might have a small wing on it that you'd see on the road, and it would put 40 pounds of downforce, and more of a racing car might put 800 pounds of downforce. But the problem is, the bigger the wing, the more the drag, and you end up with a problem in your final maximum speed capabilities.

Phil: And what Musk has created here doesn't have, at least from what I can see, any visible drag. There might be some downside underneath the car, where maybe it's not as smooth as it would normally be on a race car, but it's kind of nice to not have a big fat wing up there and get two Gs of downforce.

Austin: It does have a wing that pops out on it. If you see in the video, it has one that pops out, and it changes angle and height based on need. So apparently, if you're taking turns, it'll increase wing size, and in straights, kind of like in F1 how the DRS system flattens out, it'll do that in straights to help flatten it out and reduce drag.

Phil: That's an additional help, but that's not holding it up upside down, right? That's some vacuum that is doing that. And then the idea is that vacuum will apply in a turn.

Austin: And so think of the logic: in a straight line, with a race car going 160 miles an hour, you might have that 800 or 900 pounds of downforce from the wing. But when you slow down in a turn to 60…

Phil: It all goes away. Yeah.

Austin: It all goes away. And so now this car, through a turn, can apply up to 10,000 pounds of downforce while not weighing that much. And just think of Barber. For everybody, Phil and I have raced cars for years; Phil had a racing team. But think of Barber when you're coming down into 13, when you bottom out around that bottoming turn, coming down to 14, 14 to 15 at Barber.

Phil: Exactly. It doesn't matter how much gas you apply there. You've got all that weight just settling and compressing the suspension. It doesn't matter how fast you're going; the car will not slip, because it's just stuck to the road right through that bottoming out.

Austin: It's so fun, because a lot of times in a race we'll be racing people that aren't as familiar with the track, and in a race, that's the spot, man. I can pass people almost with total certainty right there, because they don't think they can put the gas on, and you can just push it right there, and again, your car is just pinned to the road.

Austin: We'll see. I'm sure Randy Pobst is going to be the guy they hire to go set these records at Laguna Seca, Willow, Road Atlanta here.

Phil: To get this in perspective, when we race at COTA in Austin, Texas, a really good driver will do a lap in like 2:19 or 2:20, something like that. That's really driving the car. We're talking a Porsche GT3 RS or a GT4 RS, something like that. They'll get down to 2:19s. And the guys driving F1 have so much more downforce and acceleration capability in their car that they're doing it… I think Hamilton did a 1:22 or 1:24, about a minute faster on a two-and-a-half-minute lap. He did it in like 60% of the same time. It's just unbelievable. So what I'm wondering is, will this new car beat that?

Austin: I don't think it'll beat that. Those cars are… Musk did not put $410 million into the design and development of the Roadster. [laughter] So that one might elude him. But in terms of beating a GT2 RS or a Lambo, just off the top of my head, I think with almost certainty you would be able to do that at this point.

Austin: Just to put this in perspective for everybody, this car went 0 to 60 in 0.8 seconds, which is the same 0 to 60, on average, that you get for an F-18 taking off of a carrier, where that launch system has to launch it. So this isn't a jet going down the runway speed. This is getting-catapulted-off-a-carrier level of force. It's inhuman.

Travis: Well, for a $50,000 deposit and 200,000, it's all yours.

Phil: That's crazy. An F1 car does 0 to 60 in 2.7.

Austin: Oh, yeah. No, 2.7. They can't go any faster. They're so light, they don't have the traction. This thing has rockets out the back of it. An F1 car will just burn its tires if it tries to accelerate any faster, because they weigh so little.

Phil: Like, the fastest F1 is actually 2.1, and it's all about tire grip. So how does this guy get a 0.9? What's he using for tire grip?

Hunter: I mean, how much does this thing weigh? That's the thing. I bet this thing's really freaking heavy.

Austin: It literally has thrusters out the back of the car.

Phil: Oh, it's not dependent on the tires.

Austin: Not dependent on the tires. [laughter] It takes off like a rocket.

Phil: You're in third place after qualifying. You're right behind that guy. [laughter] You come around the track, they throw the green flag, and your car gets incinerated. [laughter]

Austin: Yeah. Imagine what the Department of Transportation is going to do when one of these things gets on the road. [laughter] They're going to freak.

Phil: There's going to be some gravitational limitations, right? And one of them is going to be the tire wear. I don't see how these guys could go a lot of laps with the tires taking this sort of abuse. I don't know. They're going to find the rubber's the limit. Austin, go find the clip of it jumping the other Tesla models.

Hunter: That's what I need in Atlanta traffic. I need to be able to just leapfrog…

Hunter: No, no, how about this? It's not even traffic. It's the guy in the left lane going 55 in a 70. Yeah, that's the new flipping the bird as you pass by: "I'm just going to fly over top of you." [laughter]

Austin: I'm just envisioning you being in the car and that happens, and it starts rotating, and you just smack your head on the glass and get a concussion from that. That's not going to be fun. If you're a driver in that car without a helmet, you aren't having fun right there at all.

Phil: Oh my God, you're getting smoked. I mean, you need a racing seat and you need a HANS device, and your head will stay stable. So you're just going to work in your crash helmet, your racing seat, your HANS and your five-way seat belts. Going to work in your suit.

Why build it? Tesla as a technology company

Hunter: All right. Well, big question: any commercial application here [laughter] that you guys can think of? This was just purely a fun announcement, I think.

Travis: Yeah, this is what blows my mind about this: he put this much time and effort into building this car, and I think he came out and effectively admitted that this isn't even going to move the needle, right? It's not going to move the needle. And if you ask me, Tesla is not even a car company anyway, right? They're basically a data company, a technology company, a robotics company. That's really where they're going.

Travis: So I'm still trying to wrap my head around why. To answer your question, I don't know what the application is for this. It's a really, really cool invention, but practically, I have no idea. And even more fundamentally, why did he spend so much time and money on this, when I think the future of the company, unless I'm wrong, is really robotics, data and building full self-driving units? So that's my question to you guys: where is the motivation? What is even the logical path for this car, in that cycle?

Hunter: I think it's really good marketing. It's great marketing, 100%. We're talking about it. Everybody's talking about it. It makes me want one of these. As soon as I can afford a $250,000 car, I'll get back to you on that one. But…

Austin: You actually can't get one with the jets. The one with the rockets in it is invite-only.

Phil: But how different is this than just the F1 marketing, you know? You've got Mercedes, you've got Ferrari spending hundreds of millions of dollars every year, and none of it's practical. Most of it is just, "We're the fastest thing in the world, and you need to buy our street car."

Austin: Yeah. Tesla famously says they don't spend any money on marketing, but this is their marketing spend. They'll probably see a return.

Austin: In terms of the technology, I was just talking with Christina about this. I've got a Rivian, she's got a Rivian. We're big Rivian fans, and I upgraded from a Jeep. So I went from one of the most horrible vehicles to drive on the highway; it literally should be almost a safety hazard. Not having to pay gas prices right now: awesome. We've got two chargers in our garage to charge them both up, so we don't have to go to Superchargers.

Austin: But there is a massive technology gap between the Rivians and the Teslas. Christina's dad has the Cybertruck and the S Plaid; he's a big Tesla fan. And she just went to New Orleans, and she's going up to Rivian chargers and they won't accept payment, because they're failing to connect to the app. And it's not like there's a lot of them. So she's really struggling to do this, and it's almost embarrassing for Rivian, because they're out there making dumb designs with the lights for Halloween, but Rivian owners can't use the charging stations that Rivian made, because their technology is so poor. The app, or the maps, in Rivian is horrible compared to Google Maps.

Austin: And then they started charging, I think it's like a $1,500 one-time fee or like $100 a month or something like that, to get their autopilot features, kind of mimicking what Tesla does with FSD. But Tesla's FSD, for those of you that don't follow the news: if I was sitting here in my driveway, I could plug in any address as far away as I want and just hit go, and that car with FSD would back me out of my garage, drive me out of my driveway, drive me to where I wanted to go and even park in the parking lot for me, without touching the wheel once.

Austin: It is as autonomous as you can possibly get without the legal barrier of not having somebody sitting in the passenger seat. Versus Rivian's autonomy, which is about the same quality as an Infiniti QX80 that I was driving in like 2017, where it basically just keeps you in the lanes and slows down for traffic in front of you. And that $1,500 upgrade allows you to flip your blinker on and change lanes.

Austin: So I agree with Travis completely. It's not about the car. Tesla is a technology company. And I think that as soon as they get clearance to have autonomous vehicles, personally, I would sell all my cars to not have to drive anywhere. Just sit in the back seat, read the news, listen to podcasts. It would make an hour-long drive nothing. I'd happily pay for that, if I could be chauffeured by my car everywhere.

What self-driving could change

Phil: We were just thinking about how the impact of this on children is so phenomenal. We had a couple people working on our farm, and they're at sort of a labor level of wages, so they're working hard to make a living, two jobs, the whole thing. One of their kids was about 12 years old and got interested in band and wanted to be in a band at school, but both parents were working hard and couldn't come after school and pick him up and take him to wherever that practice was. There was something involved with transportation, and they just couldn't do it.

Phil: And as a result, this kid ended up joining a gang when he was 16, 17, and just went down the tubes. He's in jail now. And you just think what a difference it would make for all lower-class families who are struggling to get their kids advantages that exist out there, but they need somebody to get them there. And now, all of a sudden, you've got this self-driving car that could take a kid anywhere you need them to go. And I'm telling you right now, if you have kids, you have a really good idea how much time and effort it takes to actually manage that process of moving your kids around after school.

Travis: From a purely selfish standpoint, I'm a motorcycle rider, and my biggest fear riding down the road is somebody swinging into my lane because they can't see me, right? So I'm always just paranoid. You get into somebody's blind spot, and they swing over and get you. Self-driving cars' ability to increase safety for everybody is huge.

Austin: They're going to make mistakes, too. That's the thing that everybody points to: "Oh, well, somebody unfortunately was killed." But you look at the data, and there's plenty of it now: billions of miles driven. They are infinitely safer than anybody driving, because accidents usually don't happen because somebody's a bad driver. They happen because they've got a kid screaming in the back seat and they turn around for a split second to manage a distraction going on in the vehicle, or something like that, and that's what causes it.

Austin: Or they're playing a video poker game at the same time. Or they're playing a video poker game on their phone while driving. You've got to make money somehow.

Hunter: I mean, that is the biggest hurdle for me: you've got to be able to point the finger at somebody when accidents happen. I think that's an innate thing for people, and geez, that is going to be really difficult when it's a widespread phenomenon and every day there are horrific wrecks happening with FSD as the culprit. But yeah, sure, it's 5% as much as it would be if there were all human drivers on the road. But that's going to be a hurdle for people to get over at some point.

Hunter: But we'll see when that happens. I think it's been "next year," according to Tesla. It's been, "Next year we're going to have FSD fully up and running." It's been that way for how long now?

Travis: Since 2016, I think, was the first time it was supposed to come out.

How do you value Tesla?

Phil: Yeah. Look at what we're talking about with Tesla in terms of it not being a car company. It's got robotics. It's got all this innovation going on. It's just an amazing company. And how do you figure out what it's worth? How in the world would you know what this company is worth? What kind of cash flow can you expect over the next 10 years coming out of this company, if you were to own it yourself and just have it run the way it's being run right now, with all of the R&D that's going on? How would you put a number on it?

Austin: I think it gets really hard right now. You get 50 million people subscribed to FSD monthly, and it gets a lot easier, looking at it like a subscription revenue service. So it's a great company right now, but trying to predict when it will actually get FSD is where people will ultimately make a lot of money, or not a lot of money. If it takes 10 more years, the people that are betting on it happening by next year are probably not going to make a lot of money. But if it happens next year… And I think that's what we're trying to avoid: getting into investment situations where we're trying to guess when a massive technological change will allow revenues and cash flows to show up. We just want them to be there already.

Travis: What's the current market cap? Like over a trillion, right? It's like 1.2, 1.3 trillion, something like that. Because if that's the case, then if you wanted to earn 15%, the thing would have to be like a five- or six-trillion-dollar company to justify that valuation, right?

Austin: Oh, it's actually about 1.4 trillion. So it's almost a 1.4 trillion market cap. That means you'd have to get to 6 trillion to make a 15% return.

Phil: Well, those numbers just don't mean anything anymore to me. They're so big. I've lost track of them in terms of what that means out there. I used to understand what a $500 million company meant. This is crazy.

Phil: What does that mean? So the other side of this coin, kind of the Cathie Wood side, if you will, is there are going to be gigantic winners out there, and we need to buy into them in order to participate in that amazing win. And I think to a certain degree, that's got to be a little bit of the logic that Buffett applied to Google. We're basically arguing that Google is a purchase by Berkshire because it's a better deal in the long run than U.S. Treasuries, right? It doesn't have a real high hurdle to become a better deal.

Phil: But it could be an ancillary positive thing to own, if you're thinking there's going to be some winners out of these seven companies and Google's one of them. "They're just going to be a winner, and I'm sure of it." You almost have to think Buffett thinks, "I'm sure of it. It's better than T-bills, and T-bills are a certainty." So he's not only sure that it's going to be better than T-bills, but it's going to be there. It's going to be one of the winners.

Austin: If search alone continues to grow at 12%, you could make an argument that buying Google at 350 today will return about 8%. If everything else in Google failed, just the search business would carry that investment for the next 10 years, above a T-bill rate. So everything else that Google does is just upside beyond that, as long as search continues to be a dominant form of digital advertising.

Phil: Speaking of which, I'm finding… We're going through and kind of eliminating some of these AI search models, like Perplexity and ChatGPT and Grok. You've got them all competing out there, and Gemini is just so easy. It's just right there on my Google search. It's just so much easier than the rest of them. I'm starting to find myself using it more and more, and for very specific stuff, I go over to Claude, and that's it. I've got Gemini and Claude.

Austin: And the reason I do that is because Gemini has all the recent information from YouTube and the news that search collects for it, which was always our thesis. It was always a data issue, and Google has the data. I think Claude, for advanced tasks like building a really good presentation, or especially coding, is light-years ahead of Gemini. But Claude has to go search Google if you want current news. [laughter]

Austin: Yeah. Yeah, 100% agree with you.

Robots: Optimus and the physical world

Travis: I guess my point, relative to Tesla, is: is Tesla one of those? It's got the car company; it's got profitability in that regard. If you look at the Optimus robots, I actually think there's something there. I'm not saying it justifies the current valuation, but I look at the Optimus robots… I think I was telling Austin this story. It was in the late '90s, robotics, where we were experimenting with those in the factory, and they've come a long way.

Travis: I think there's a real case in the future, down the road, where they're going to become mainstream in manufacturing and households and everyday life. The question is how long that takes. I'm not suggesting it justifies the current price, but I think there's a lot of runway for the company there, and all the data associated with that.

Phil: Did you see these MMA fighters fighting a robot, or was this some AI BS that I just saw? [laughter]

Travis: I did not see that.

Hunter: There's no way I would be getting in the ring with some robot that would break my neck. [laughter]

Phil: It looked real. I don't know if it was real. You guys tell us out there if you think this is real, but go look at it. It's like one guy after another taking on this robot. These guys would hit this thing and try to trip it, do a thing to try to take the robot down. And the balance level of this machine was insane, man. The robot would just rip your head off.

Hunter: Yeah, maybe if it wasn't punching back.

Travis: Yeah, if it wasn't punching back. Or because they actually have these things behind cages, so they don't accidentally just tear somebody's face off.

Austin: Look at the robot. OK, so he's wearing, like, motorcycle gear, almost. Check the robot. It's the Terminator. [laughter]

Phil: So is this AI?

Austin: No. No. This is real. It was on Fox News, Phil, a month ago.

Phil: That's my nightmare. Give me a break.

Travis: That is crazy.

Phil: When he gets this robot off… That hurt. He gets this robot off balance a little bit. A human couldn't have recovered. I think there's one… God, man, that's nasty.

Hunter: Yeah, I would have bet a substantial amount that it was going to end up being AI that you saw, but I guess I'm wrong.

Travis: That's incredibly impressive, actually, because I saw a video not too long ago. There's a company in Corvallis called Agility Robotics, and I saw this video where a guy was trying to knock it over with a stick. They were behind a protective screen trying to push it over, and this robot was on one leg, balancing and using its arms to recover. It was crazy. That one looks like it's about 10x more advanced. That thing's moving like a human.

Phil: All right, we're getting there. We're getting there.

Hunter: Well, that's the last frontier, right? When you think about the use cases for AI, it can handle more and more of our daily workload when we're doing knowledge-based tasks. At some point, that technology presenting itself as something that can come and build a fence in my backyard, or come and fix plumbing in my house, would be insane. I'm using AI to try to figure that stuff out now, but operating in the physical space is going to be a huge market, especially something that doesn't live in an industrial warehouse somewhere. So in Tesla's case, if they are the leaders in that industry, that could be explosive growth potential.

Phil: I mean, this is a great use case for robotics, honest to God: putting in a fence. We needed a fence that was half a mile long, so we were looking around for somebody who could put this fence in at a reasonable price. And we found a guy, and he was a really good fence builder. The only downside was that he also was using meth. [laughter] So the fence would progress and then stop for a while [laughter], and then it would start again.

Austin: Did you do the cost-benefit analysis? You're like, if we'd paid a professional, it would have been twice as much but finished in half the time. So in the end, it all worked out the same.

Phil: We had that exact same thing happen on a house we were renovating. We had a guy that was such a good price, but it turned out he worked about an hour in an eight-hour day. [laughter] Every time we went over there and looked, he'd be outside smoking and hanging out. It turned out to be not all that efficient.

Phil: So there's a real case here. And obviously, how many workers are coming in from overseas to take the jobs that Americans won't do, these menial jobs? So this is where the robots are going to come roaring in.

Austin: Absolutely. There's an amazing, not politically correct South Park episode on this that is completely true. Outside of the Home Depot in this episode, there's all the white-collar workers whose jobs got taken by AI, out there trying to find some work, and then all the contractors are rolling by in like a $200,000 lifted truck. [laughter] That's pretty true, honestly. [laughter]

Phil: Oh my God, man. I've got a friend of mine up in Wyoming who runs a landscaping company [laughter], and all the guys who work for him are Latin American guys who probably have various legal or illegal status, and he pays them really well. His top guys are making $35 to $40 an hour. His bottom guys are making 30. So it's not a wage issue, per se. So I was asking him, why is everybody from Latin America here? None of them are American guys. And he goes, "Yeah, WWW." I said, "What's that?" He goes, "Why do you won't work." [laughter]

Hunter: Well, we're getting canceled on our second episode, everybody. Third episode. And we are done for.

Phil: That's the one group you can mess with where you can't be canceled. We're in good shape.

A small, fun position?

Travis: Well, I still don't know how to put a value on this business. So unless it gets to six trillion, guys, I'm out.

Austin: I'd say that if there was ever an entrepreneur that could do it, they've got the jockey to do it. I think 20 companies in the entire history of the world have reached a trillion-dollar market cap, and Elon has founded two of those companies. So…

Travis: All I'll say is, if he tells you it's going to be there in 10 years, just double it. Call it 20.

Phil: Double it to 20, and you're still at about 8%. So you're good. But what I think is fun is to think about a portfolio that is dominated by really wonderful businesses that you do have a clear view of the value of, and a very clear view of what the cash flow is going to be, so you buy them at the right price. But I do think there's a lot of fun and a lot of energy in owning something like a Tesla, taking a small part of the portfolio and getting it in there. It just makes investing fun in a way that buying your next chewing gum company doesn't quite cover. [laughter]

Travis: So you would advocate… so this could be a risky business play, right, Phil? Let's say you take like 1% of your portfolio and throw a little bit in there.

Phil: Yeah. And that way, you can say, "Yeah, I bought Tesla." You can tell your grandkids, "Yeah, I bought Tesla when it was a trillion."

Austin: All right, fellas. Monday. Let's go.

Travis: Yeah. Here we go.

Austin: Let's roll the dice. We got the green light from the boss.

Hyperscalers' off-balance-sheet obligations

Hunter: All right, guys. Well, hey, let's move on, speaking of the AI talk here, to a subject we've been meaning to get to the last couple of weeks that's really interesting. There's been some consternation out there about what's going to happen with these hyperscalers and all these huge off-balance-sheet obligations that they have, which could come back to bite them in the future. I know a lot of us want to talk about what the potential debt implications could be, but there's this other aspect of it, which is the lease obligations, that I'd like to start with.

Hunter: A lot of these off-balance-sheet obligations, as of right now, are commitments to enter into leases with data centers to provide some compute for these big hyperscalers. And the danger there is not really evident, because it doesn't quite show up on the balance sheet. You see a lot of these obligations in the footnotes, but they can have huge, massive balance sheet implications in the event that there's some sort of default on these commercial leases.

Hunter: In this case, this would be an over-obligation. I guess these hyperscalers have obligated themselves to too much compute, too much data center space out there, and they don't really need it anymore. It's at least been a question of ours, maybe not a prediction, that the amount of compute that's going to be needed, at least in the short term, might not actually be what's being forecasted out there. There's a lot of the idea of the use case for AI right now being largely saturated. There's this idea of the fact that the chips themselves are going to become much more efficient, so the square footage that's needed to operate these huge stacks of chips is going to be smaller. So it's a possibility.

Hunter: But what happens first is that you're going to get a notice from the owner of these locations. It's going to say, "Hey, you're late on your rent. You haven't paid it. What are we going to do about that? I know that you're not using the space, and it's not producing any revenue for you, so why would you?" But what happens further down the line, after an eviction lawsuit has been filed, is that the owners of these locations can ask for what's called rent acceleration.

Hunter: So if they have signed on to, say, a 15-year lease, or even longer, to build out one of these data centers in one of these locations, then the owner of that building can come back and say, "OK, you haven't paid your rent. I've tried to collect it from you. Now you need to pay me back in a lump sum: the $30 million, $40 million, whatever it might be, probably even more than that, that you owe me for building this location, all in one lump sum."

Hunter: And what makes that even more dangerous than just owning a building, or having a loan to build a location, is that you can always refinance that. But in this case, you're dealing with one owner of a location, and they might not be particularly interested in trying to find another party to occupy that space. And in the hypothetical that we've built out right now, there isn't anybody out there that's really wanting to occupy that space.

Hunter: So you could have billions and billions, hundreds of billions of dollars' worth of obligations coming due on space that you're not using, that's not profitable for you, if you are one of these hyperscalers, like a Google that we just got done talking about. So that's kind of the first domino that would fall here, and it will actually start to show up as an obligation on your balance sheet when a lawsuit is granted and the rent-acceleration clause has been exercised, and now you owe hundreds of billions of dollars.

Austin: One curious thing I have with these leases… Phil, I want to pitch this over to you in a sec. Let me bring up a chart that I was working on in Bloomberg, their AI tool, just so you guys can see what I'm looking at here. As of last quarter, I brought up Amazon, Microsoft, Google, Facebook, Oracle and Broadcom, which I consider the biggest six spenders that are taking on debts and liabilities.

Austin: Then I've got what's currently on their balance sheets in terms of debt and leases, and then off-balance-sheet leases, future debt if they've mentioned any, but also these purchase commitments to buy chips and stuff from Nvidia. And you can see the numbers are staggering. On balance sheet right now, it's almost a trillion dollars of total debt and lease liabilities, forecasted, with the off-balance-sheet purchase commitments, to grow to 3.6 trillion.

Austin: Before I pass it to Phil, because Phil, I'd love for you to talk through it, since I know you have some very familiar experience of what happens when you miss a debt payment with a company: Hunter, are there typically clauses…? And just so everybody understands why I'm asking Hunter this, his father successfully built a major real estate company up in the Northeast, and Hunter has been in the commercial real estate game his entire life through his family. So if one property goes bad, is there ever anything that could systematically trigger other properties? Or, as long as you're paying the rent on the others, is there no ability for an owner to trigger a closure?

Hunter: Yeah. I would say in this case, it's probably not an issue. These are really, really creditworthy businesses. If you're a smaller tenant, then you might have to personally guarantee some of these leases, or put up other properties or other leases ahead of the one that you are paying on. In this case, that's probably not going to be the case, since these institutions are so creditworthy.

Hunter: So the first thing that's probably going to happen is a lockout. You're not going to have access to the facility anymore. And then typically, as part of the contracts for these leases, the owners are able to go in and take possession of anything that is inside those buildings and garnish that from you, and then turn around and sell it. In this case, that's going to be a bunch of fairly valuable equipment that the owners of these locations can turn around and sell off. So that's going to have another balance sheet effect as well.

Phil: Yeah. And if you take Meta, for example, good numbers here: you're looking at a grand total of over a trillion dollars, when what they're showing on their balance sheet is, what is it, 100-something?

Austin: 112 billion.

Phil: Yeah. OK. So they're 10x: what they actually owe, compared to what they're showing. And the problem is that it changes your multiple dramatically on what the enterprise value of the business is, where enterprise value is where you take the debt and add it to the market cap. So you just have a complete dislocation of reality if these things are legitimately debt to Meta.

Debt vs. leases vs. purchase commitments

Austin: And that's the thing that I wanted to touch on. Before we dive too deep, I want everybody to understand the difference between a debt and the risks there, versus a lease like Hunter just explained, versus a purchase commitment. And Travis is actually the perfect person to explain these purchase contracts, because he did it for 26 years at HP. But Phil, could you explain to people what happens if you miss an interest payment on one of your loans?

Phil: Yeah, and some of you have been through this with me. It's devastating, because you basically turn over to the management team the option of how they're going to deal with this missed interest payment. First off, you have the lender, who is going to go to management and say, "OK, you're going to pay us, and we have these conditions." They're probably going to negotiate. And then one of two things is going to happen. Either the management team is going to find the money to pay this from someplace else, they're going to sell something, which is typically what happens, they're going to unload in anticipation of avoiding a bankruptcy. And a good management team is going to do that.

Phil: So if we were to take a look at Meta, they essentially owe $900 billion, and they've got to pay it, I think, around 2027 or 2028.

Austin: 450 billion of that is purchase commitments, which is different than debt. Of their total debt, they currently have 80 billion on their balance sheet, and it's forecasted to go up. But a big portion of that is leases and these purchase commitments.

Phil: But let me just go into the second thing that can happen here, and that is that the management team can decide, "Let's just take this thing into bankruptcy and eliminate these issues," which they can do in a Chapter 11 bankruptcy without pretty much upsetting the operations of the business. The whole idea of a Chapter 11 bankruptcy is, in fact, to not upset the business operations: to keep everybody employed, to keep the products in the market, and to go through a really, really quick debt restructuring that, at a minimum, is going to chop the debt down dramatically, typically by over half. And often it eliminates the debt completely, and it's just a start-over time.

Phil: But of course, when it eliminates the debt… The debt is a higher level of obligation for the company than the shareholders have. So in order to get the holders of the debt to agree to chop their debt, the first thing that has to happen is you have to eliminate all the shareholders' rights in the company. You completely restructure the company. You wipe out all of the existing shareholders. You put the company into a bankruptcy form, and down the road, 45 days to a year later, you're coming out of bankruptcy with essentially the same name of the company but a new form, and that new form now has this completely restructured debt.

Phil: So that's basically what can happen, and I've been through it, and it is amazing what management teams are willing to do, particularly if they are not the ones that put the whole problem together in the first place. So the biggest thing that can happen here is a bankruptcy. And shareholders who are looking at a company like Meta, nobody's thinking bankruptcy with Meta, right? Nobody. And yet look at the load that they're obligated to. And I don't know, Austin, that they're not obligated to take care of that $900 billion.

Austin: That's what I wanted to bring Travis on for. But before Travis, before you come on and talk about purchase commitments: wasn't there the case with Horsehead, where they missed an interest payment? They had, like, what, $400 million in debt? Didn't they miss like one interest payment on like a $10 million loan?

Phil: It was on 40. It was on a $40 million loan to an Australian lender, and those bastards took it into bankruptcy, even though they had elements of the company, a couple of subsidiaries, that were worth over $100 million, and they refused to entertain offers.

Austin: Didn't that trigger a systematic triggering of their other debt, too? When they missed that interest payment and took that loan into bankruptcy, weren't all the other lenders like, "This trips a clause where now our debts are also due"?

Phil: That is probably part of it. In other words, every lender is going to go into bankruptcy together, and one is going to trigger the rest, right? But man, what a bunch of bastards running that company, honest to God. [laughter] Really. I would love to see them in jail. We tried hard, but we didn't get them there, unfortunately. We won our court case, but it was just a nightmare trying to figure out how to really make them accountable.

Phil: So one of the things we learned from that is that you can't assume that rational things are being done with a lot of debt. You can't assume it. So I guess I'm really interested, Trav, in what the level of obligation is here. Can this off-the-books debt force this company into bankruptcy?

Austin: So the last piece of this, Travis, just to tee you up, is that between these six companies, there's $1.7 trillion of off-balance-sheet purchase commitments to buy chips and equipment. So it's not debt, it's not leases; it's a commitment, for 10 years into the future, to buy X amount of chips per year.

Travis: Yeah, these long-term purchasing agreements, they're calling them, right? So I think it depends. This is not going to be a simple answer. I think it really depends on the structure of the contract. I've been reaching out to colleagues in the industry, and I've been getting different kinds of responses in terms of how committed, or how obligated, these companies are to purchasing these chips. It really depends on the contract.

Travis: I actually had a really interesting conversation with a former colleague who told me, "Yeah, we've got a long-term purchasing agreement to buy memory, and it goes out to 2028, but it can be canceled within lead time." And what that means is manufacturing lead time. Let's say, for example, you're buying memory from Micron, and it takes Micron six months from when it starts the wafer to the time the chip comes out and is shippable. Even though you're committed to buy a certain quantity out to 2028, you could cancel those orders up to six months before they start.

Travis: Anything that's started, you're paying for, right? They already started it, so it's going to get shipped to you whether you like it or not. But that's six months. That's nothing, right? That's not a real commitment. They're booking the orders, and they're going to have that allocation committed to them, but they don't have to take it. But there are other contracts that are, I think they're called take-or-pay, right?

Austin: Yeah. Meta has 454 billion in take-or-pay contracts.

Travis: Right. That's a different story. Now you are obligated to take those. That stuff's coming to you whether you like it or not. But here's the reality, just from living in this world. I'll admit this is a little bit different situation, and it's also a little bit different time frame, but you could have these long-term take-or-pay commitments, and if and when the music stops, you're going to go back to the table and renegotiate that contract, right?

Travis: You're going to get duration changes. You're going to ask them, "Hey, I want a year before you ship this stuff to me. I'll pay you after that year's delay. We're going to renegotiate this." Because there's only a few people that have the kind of buying power to purchase that much material, and you only have a few suppliers. Those relationships are usually long-term. They want them to be productive. You're not just going to shove all of this… I mean, it's possible, but I doubt that they would just force them to take it and pay.

Phil: I think you're making a really good point, but that's the same thing with regular debt. Companies go in and negotiate their debt all the time. But the real question here, I think, is that the legality of those contracts is, I think, identical to debt. Everything you're saying is absolutely true. They're going to try to renegotiate. They're going to try to push it off. You're going to have all these other business operations going on. But at the end of the day, they can't get out of it without doing all of that, just like regular debt. So, not treating it like regular debt, when it is just like regular debt.

Travis: Yeah. No, I agree. It's an obligation. It's debt. It's a real thing. But to your point, you can renegotiate a note, and this is going to be a little bit more flexible, I guess, is my point. You have a little bit more flexibility here, right? I think that's really the difference…

Phil: Because of the interest in the company in trying to keep that relationship going.

Travis: Exactly. It's mutually beneficial.

Austin: Notes are a lot harder, because there's usually multiple owners of them. Once the debt is issued, it goes off to tons of third-party buyers that hold it, pension funds, things like that. So renegotiating something like that requires getting everybody on board, versus this, which is a one-party-to-one-party renegotiation. So that's on the take-or-pay. Then you've got all of these leases. Meta has another 355 billion of uncommenced leases that are binding future debt. They may have a little more flexibility under those, but I don't know how much.

Travis: And if they haven't been commenced, they're not even on the balance sheet, right? They don't show up on the balance sheet until it's committed. They could have signed it, and it never shows up, right? Until they commence it, that's the only time you see it. As an investor, you have to go out and understand what the signed-but-not-yet-commenced leases are to understand this.

Inside a data center joint venture

Austin: So take Hyperion, the Hyperion facility down in Louisiana. Basically, what they did is they bought a piece of land for a couple billion dollars, or maybe even less than that, and then they formed a JV with a company called Blue Owl. That's like a separate business, and that JV is what owns the land and the data center, and Meta is going to lease it back from the JV that it's a 20% owner of. Blue Owl owns the other 80%.

Austin: Now, Blue Owl, the 80% owner, goes to a lender and secures a loan to build the data center. The reason it doesn't show up on Meta's balance sheet is because the lender isn't making the loan to Meta. They're making the loan to Hyperion, which is owned by Blue Owl. So it's off-balance-sheet entirely.

Austin: But the only way the lender will lend money for this data center is if Meta agrees that they will basically guarantee, I think it's 16 years in that case: "Whether we use it or not, we'll guarantee 16 years of lease payments." That's what the lender is using as security, because the JV has nothing. It's basically a shell company that right now has no revenues, no cash flows, no nothing. So yeah, whether they use it or not, they are absolutely on the hook for paying the leases on these facilities.

Phil: It's so sneaky, really, when you think about it. They take a minority position with Blue Owl, and that keeps the loan off the balance sheet, and they're fully guaranteeing that the lender will not lose money.

Why take the risk? Forecasting demand

Hunter: Yeah. And from my point of view, I'm sitting here wondering why these companies are willing to take such a massive risk, why you're willing to go out on a limb and take these risks. It's one of two things to me. Either there's such confidence that the demand for these products is going to continue growing into the future that they know they're going to need all of this compute, or the other option is that they don't know. There's no certainty, but they know they're going to die if they don't take the risk here. They're going to die on the vine, and somebody else is going to potentially unseat them and take market share away.

Hunter: So I wonder which one it is, and I don't know when we're going to get an answer to that. Because at some point, all of this might make a whole lot of sense to us. This is a huge, massive market. These guys are continuing to consolidate this market. But we want to look for some sort of indicator to say, "OK, this is starting to make sense. These data centers are coming online. They're being used. We see the demand for this product continuing to climb, continuing to rise." What do we feel is going to be that indicator? When's that flag going to come up?

Travis: Well, when I look at this, to that point, I think the real question is: how are they even forecasting this demand? This is the part of the equation that gets really difficult for me. I understand the supply side pretty well. The demand side is really hard to wrap my head around. So I'd be really curious to see how they're doing that and what they're basing it off of. I'm assuming that most forecasts are off by orders of magnitude, right? And good business operations figure out a way to mitigate that risk and make sure that they don't buy too much material and equipment and spend too much capital on something they're never going to need. So they're playing that game right now in a big way.

CoreWeave and the credit markets

Travis: And this is where I think companies like CoreWeave come into the equation. They're in the mix. Are you guys familiar with CoreWeave? Just maybe for an overview for everybody: CoreWeave, I guess the best way to think of it is that they're like a commercial landlord for AI, right? They basically sell compute to companies like Microsoft and Meta. Microsoft and Meta need all these Nvidia chips for their models, and they need the buildings and the data centers, and they need all this electrical power, and that can be really expensive and slow. It can take two to three years to build a data center. But CoreWeave has gone out and built up some of this capacity and can do it for them, on a lease basis. So basically, CoreWeave's collecting a rent check for providing this compute power to these companies.

Travis: And so that's one way I think these companies are trying to figure it out: "Hey, I can lease some of this instead of building it." It's the make-versus-buy decision. If you look at what CoreWeave is doing, I think their CDS opened up like a thousand points or something over the last week.

Austin: Their bonds went to junk.

Travis: Yeah. So what were they priced at before, Austin? It was like 200 basis points or something, and it went to a thousand.

Austin: Yeah. We've been looking at them for a little while. Just to put this in context, we're talking about credit default swap spreads, which is what Michael Burry famously bought to basically bet against the housing market, and he made billions. Steve Eisman made billions. Same thing we're talking about. You're basically buying an insurance policy that says, "I get paid out if this company goes bankrupt."

Austin: Two years ago, even a year ago, with the AI buildout, a lot of these companies like CoreWeave and Oracle were being priced like there was, let's call it, under a 1% chance of a default event occurring in the next year, and maybe a 5% chance over the next five years. And CoreWeave, just in the past two months, has surged to where now I think it's a 15% chance of a default in the next year. That's what the insurance price is saying; that's what the insurance costs. And a coin toss, 50/50, that CoreWeave defaults on a loan in the next five years. So that's what they call junk debt, essentially. You're getting 15 to 20% yields per year to own the debt.

Austin: Now, Oracle has surged to, I think, a 30% chance of a default in the next five years, which is just crazy to think: that a company as great as Oracle, with a guy as rich as Ellison, would ever put his company in a position where it could potentially face a default. But man, it's happened to plenty of rich people before. They don't need to do it, but they just can't help themselves. That would scare the crap out of me, if I had more money than I could ever need because of a company I founded. The thought of putting it in a position where there was a chance of default… I just can't put myself in those shoes. I get it from a startup perspective; you're trying to become the billionaire. But once you've made it, what possesses you to potentially give it up?

Travis: And that's why I brought this example up, to that exact point: why? They're not only building them themselves, but they're working with companies like CoreWeave that have these massive obligations. CoreWeave's total debt is going to exceed their equity if they keep up at this pace, right? So they're putting themselves at risk, which tells me, assuming these guys are rational, that they believe the demand is real and that they have to do it.

Phil: My view of this is that this is going to commodity pricing, and if that revenue stream stalls, then the free cash flow that you're bringing in from your absolute fortress business at Meta or Google turns deeply negative and sucks up your core cash reserves. And then you're in a situation where you've got to issue stock, share dilution. You've got to get debt.

Phil: I don't think any of these companies are in any kind of scenario where you could say they could go bankrupt in any kind of reasonable way, but it would just crush their stock. Man, that stock value would disappear. You'd be looking at these companies at 10% of what they're currently seeing on the stock market. And it doesn't even have to be rational to take it down that much, because once it really starts, then every mom and pop, 401(k), index, everything, they're all selling, and it just goes viral. So I don't see a probability of bankruptcy. What I do see is a significant probability that the revenue does not come in. The demand could be there, but the dollars might not be there, in which case this whole thing falls off the table.

Austin: I think the reason that CoreWeave's bonds are priced the way they are, which is different from an Oracle or a Google in a big way, is because of the way I've listened to the founder talk about how he structures these deals. He goes, "I put things in a box. I've got a Microsoft box, and Microsoft gives me a guarantee of compute over the next eight years. Then I put some cash in the box, and I go to a lender and say, 'Here's this guarantee from Microsoft.'" He basically said something along the lines of, "If that box goes bad, it doesn't impact any of the other boxes."

Austin: Which is completely wrong, because if that box goes bad, nobody's ever going to put more debt in any of the other boxes for the next two or three years. They're already at junk bond status without one going bad. And CoreWeave's got like $30 to $40 billion of commitments that they need a debt market to fund. They specifically do not have the cash flow to fund these commitments themselves. They have to have an open debt market.

Phil: And the rates just went way up on that debt. Way up, right? I don't know, they were probably getting 4 to 5% before. I don't know what they're paying now. It's got to be north of 10%.

Austin: If I had to make a prediction or a bet, I'd say if AI cracks, you'll see it in CoreWeave cracking before anything else. That would be the first domino to fall: them not being able to meet some future commitment that they've guaranteed they'd provide compute for, because they can't get the financing to actually build that commitment out.

Travis: And if that happens, I've got to believe the contagion spreads to the hyperscalers as well. That's the whole thing here.

Phil: The difference being that they do have a cash flow to fund losses over here, though. That's the difference.

Travis: Exactly. That's exactly right. They've got other businesses that can fund it, and all they have to do is turn off the capex spigot, and they've got tons of free cash flow, and they're going to survive. But I still think, to Phil's point, the contagion spreads to the stock price. That's when I want to buy these companies. Obviously, we would evaluate what the actual situation is and make sure that we're not wrong about our assumptions here, but assuming they're going to survive it, that's the event where you want to own these guys.

Austin: And specifically, I would prefer Amazon, Microsoft and Google over all others, because if they overbuild these data centers, eventually their cloud businesses will absorb that overspend on compute. They actually have a business use case. If Meta overbuilds, they have no cloud business where customers are coming in and sucking up that unused space. They just overspent. They're basically spending to improve their ad targeting and make an AI model, Muse and their Llama models. So they don't have this fallback that the other big ones, Amazon, Microsoft and Google, have, where the demand for cloud will, I think, suck up a lot of that capacity in the long run.

Wrap-up

Hunter: So yeah, with that, everybody, I guess we will go ahead and conclude this episode of the Rule One Podcast. Thanks for listening, and look for another podcast episode from us next week, hopefully with some more topical news. So we will close it down, and for Austin, Travis and Phil, I'm Hunter, and we will see you guys next week.