The company at the heart of the AI bubble

0:01Hello and welcome to Decoder. I'm Nilay Patel, Editor in Chief of The Verge, and Decoder is my show about big ideas and other problems. So a lot of people think AI is a bubble. A lot of people, including people like OpenAI CEO Sam Altman, who keeps running around telling everyone that he thinks AI is a bubble, all the while raising and spending enormous amounts of money that seem like bubble indicators to everyone else. This is all pretty confusing. So we set Verge senior reporter Liz Lopato out to report on this supposed AI bubble, whether it's real, how it might pop, and what all this might mean.

0:34And she's joining the show today to talk about one particular company that sits right in the middle of all of it. That company is called CoreWeave, and Liz has spent considerable time digging into its history, its financials, and the truly fascinating story that all of it tells us about the modern AI boom. Liz actually has a big story about CoreWeave running next week on The Verge. So if you want even more detail after you listen to this episode, just stay tuned for that. But to me, the basics of the CoreWeave story are themselves fascinating. CoreWeave was founded by former commodities traders as an Ethereum mining firm out in New Jersey.

1:08And crypto. Like so much of the tech industry, when crypto sputtered, CoreWeave pivoted to AI. It started to use the same GPUs it was stockpiling to mine Ethereum to build its own data center operation. CoreWeave's big innovation is creative financing. It raises huge amounts of money to build data centers full of NVIDIA GPUs, and then leases those GPUs back to big AI companies to feed the industry's seemingly insatiable demand for compute. That's turned CoreWeave into a pillar of the obscenely expensive AI infrastructure build that's turned markets inside out over the past year. CoreWeave itself went public this past March, and it's worth almost$50 billion today.

1:45And just this year alone, CoreWeave has signed multiple contracts worth tens of billions of dollars, with companies like Meta and OpenAI for access to its data centers. But what Liz has found is that the story of CoreWeave is a lot more complicated than just a few Wall Street types making some smart bets and getting filthy rich. In fact, the narrative around CoreWeave, that it's just making the picks and shovels of the AI gold rush, is built around some very complex financial maneuvering, as well as a unique relationship to NVIDIA itself. In many ways, Liz says, CoreWeave could not exist without extraordinary levels of financial investment from NVIDIA, and from the value the AI industry is putting on NVIDIA's stock on its chip and the wholesale transformation of the entire economy that AI is supposed to deliver at some point in the future.

2:30So I wanted to ask Liz, what does it mean for CoreWeave to be so dependent on NVIDIA, for the AI industry to be so dependent on CoreWeave, and what might happen in a world where the AI bubble actually pops? Okay, we're at Senior Reporter Liz Lopato on CoreWeave and the AI bubble. Here we go.

3:00Liz Lopato, you're a senior reporter here at The Verge. You've also had many other jobs here at The Verge. You're reporting a lot on AI and the finances behind AI. Welcome to Decoder. Hi, it's good to be here. I'm very excited to talk to you about this story. When you start reporting on AI, you end up being asked if things are a bubble. And when you start reporting on a bubble, you land at CoreWeave pretty directly. That has been your experience. It's been my experience. Explain to people why even talking about the idea of an AI bubble leads you to core weave as quickly as it does. So one of the investment theses you hear that sounds really smart and sophisticated is, oh, we're not investing in direct AI companies.

3:40We're investing in the picks and shovels. Selling the picks and shovels during a gold rush, that does seem smarter than trying to go pan for gold. I thought, OK, well, let's go look at the data centers. Those are essentially the picks and shovels, right? The thing about CoreWeave is that it's one of a group of companies called a NeoCloud, and they essentially just make data centers that they then lease out to people. Because it has gone public recently, there is more information available about it than there is about a lot of other things because of that beautiful S1 they filed. I was thinking, oh, I'll just get some thoughts about data center economics.

4:14Can people get their investment back? Is this going to have a return? And then I fell into a pit. That was why we are now here, because I still haven't climbed out of this pit. So let me just put this in broader context. And you can explain to me how CoreWeave is the same or different than the cloud providers we're familiar with. Amazon built AWS. It is still the leading cloud provider in the entire game. Google built Google Cloud. Microsoft has Azure. Those, in my mind, were as much about the software and services that ran on the data centers as they were about the chips. If you use AWS, Amazon builds you a bunch of primitives.

4:52You can just stand up a service, an application, something in the background very, very quickly using Amazon's resources. They will manage demand. They will manage rolling more compute at you if you need to. And Amazon is going to take over a bunch of parts of running the data center for you. So you can just focus on your application. Same with Google. Same with Microsoft. And their lock-in was all of that service provisioning. They're going to do all this work for you. CoreWeave is just selling chips. It doesn't have that middle layer of infrastructure that the big clouds have. Is it really just we're going to put the chips in a room and lease them out to you and you can do whatever you want?

5:29Or is there something else about this company that makes it have more of a moat? It's a little more sophisticated than that, but not a lot more sophisticated than that. Because it turns out the way that you connect the chips really matters in terms of being able to get good performance and good uptime out of them. And it also turns out that it's really hard to get energy, and that is a major supply constraint on AI right now. CoreWeave has a pretty good track record of getting energy, and they also have relatively good engineering and can charge a premium for renting out their chips, essentially.

5:58But it's not a huge mode. It's not necessarily a durable mode either, because if they lose the engineering advantage they've got, then they are basically in a situation where they're competing on price. And that's not a situation where they can win. That's really the heart of it for me. You know, Amazon and Google and Microsoft, they compute to be these huge hyperscaler clouds. But they have many, many ways to compete beyond just price. CoreWeave is competing on what feels like price, maybe data center build out or energy contracts or preferential energy pricing. Who are its customers for that?

6:32So this is fun. Their customers are Microsoft, NVIDIA, Meta, and a number of other places that need essentially overflow compute. So if you do a contract with, let's say, Microsoft, it may turn out that Microsoft just doesn't have the capacity to execute your contract. And so then they will push something, maybe something less valuable. Who knows? I don't know how they make these decisions. They'll push it over to CoreWeave in order to execute the contract. That's the thing that really interests me, right? Is that if you have all of these differentiated cloud computing providers and they've decided they need a commodity and compute, CoreWeave just exists to provide that commodity to them.

7:15Can't something else provide that commodity to them? Shouldn't they build their own data centers? Well, they are, is the thing. And this is part of what makes CoreWeave interesting, right? because a lot of their customers may potentially become their competitors if the data center build out goes the way that I think it's going to go. It does suggest a long-term problem for CoreWeave, which is that right now their customers are the big guys. They have other customers besides the big guys, but Microsoft is their biggest customer followed by NVIDIA. And so people who are wanting an AI cloud host aren't going directly to CoreWeave.

7:52They're going to Microsoft. And Microsoft is keeping those customers by doing overflow compute. So there's this sort of like way in which Microsoft is basically getting to have this compute without having to take on some of the risks of it. Where did CoreWeave come from? Is this an invention of Microsoft? Is this a company with a history in the past? Is it wholly new because of AI? It is crypto. Yeah. Of course, it's crypto. So it's crypto. So these guys pivoted basically after the crypto boom in 2022 away from crypto. Because in 2021, Corwe have made all of its money by mining Ethereum. So they basically timed things correctly to really ride the AI boom.

8:41And, you know, one of the things that you see in the AI space that strikes me as a flag is that there are a lot of people who pivoted from crypto into AI because they thought it was going to be more profitable. And you may remember one of the reasons why NVIDIA was starting to have this incredible run was because all of these crypto companies were buying its chips. And NVIDIA was like not the most enthused about that because... Not at all, actually. No. I think they even nerfed one of their chips, but they didn't love crypto. And when AI started to really get frothy, especially after ChatGPT was released, I think that was something that felt better to NVIDIA and to a lot of other people than, yeah, we're going to mine Bitcoin.

9:28And so, like, Corwee was in the right place at the right time to really ride that. I understand that abstractly, right? If you're NVIDIA, you don't want to be in a race to do literally a single calculation to mine Bitcoin or to participate in the Ethereum blockchain because you'll get beaten by ASICs or other specialized compute over time. And maybe you just don't want to participate in that and you certainly don't want to reduce your ability to supply other markets that you think are more valuable long term. If you're CoreWeave, maybe you do believe that that is the future and you've gotten really good at buying and connecting GPUs to build data centers for specific tasks.

10:04AI comes along, NVIDIA says, this is the thing. This is the multi-purpose compute solution that our chips are made for. They can, if you believe it, can accelerate the economy in various ways. This is where the growth is going to be. And then you look at CoreWeave and you're like, well, you're already pretty good at building data centers full of our GPUs. Now do it for the thing that we want you to do it for. And then you just arrive at CoreWeave exists. It's just not possible for CoreWeave to exist without NVIDIA. And we can get more into that in a minute. But just like thinking about how this company works, if you look at the S1 filing, they say that they own more than 250 ,000 NVIDIA chips, which is big.

10:43And that's, you know, infrastructure that's necessary for running AI models. And in fact, it only has NVIDIA chips. It doesn't have any others. So on top of that, you have NVIDIA, which is an investor in CoreWeave. When, you know, the IPO faltered in March, NVIDIA swooped in and bought shares. NVIDIA has backstopped CoreWeave in a number of ways, including saying that they will buy any excess compute that isn't sold to customers. So they're essentially providing a guarantee of income, which is remarkable. It's like having essentially a knight, right? Like NVIDIA is the king and CoreWeave is their knight.

11:20And so NVIDIA is protecting CoreWeave in many ways, and CoreWeave in many ways seems to serve NVIDIA. So I appreciate the royal metaphor, but more directly on the ground, NVIDIA is CoreWeave's main supplier. And the entire business is we will buy these chips for X amount of dollars and sell the use of those chips for Y. And Y over time has to be more than X. And doesn't that get very complicated if your supplier is also buying the demand? Like there's something there that just breaks my brain. It's one of the reasons why you have been getting the occasional DM from me where I just say I feel crazy.

11:59Because I look at this stuff and I feel crazy. I think one way to understand this is that NVIDIA really wants CoreWeave to exist. And so they're going to do what they need to do in order to make that happen. There's a reason for that. Part of it is the data center build out and the fear that one might overbuild. And part of it is that one of CoreWeave's big customers who've just signed a huge contract with them and then expanded it is OpenAI. And OpenAI, as far as I know, is not profitable and hasn't been profitable and has been losing a lot of money. And so if you are a certain kind of person or bank, you might be nervous that a major customer like OpenAI doesn't have profit and may not be able to pay its bills.

12:45And so what this guarantee from NVIDIA does is it says, hey, these bills are going to get paid. And if you are a shareholder or a bank or whoever, it doesn't really matter to you who's paying the bills. That's not like your problem. It's just a matter of making sure those bills get paid. We need to take a quick break. We'll be right back.

13:17Support for this show comes from LinkedIn. Imagine if any of the movies that included the line, I need the right person for the job, settled for, I'll just take about anyone. How many heists would have failed? How many deals would have fallen through? How many secret spy missions would have ended in disaster? So why would you accept just anyone when hiring for your business? When you need the right person for the job, you can turn to LinkedIn Jobs. And now LinkedIn Jobs is stepping things up with their new AI assistant. So you can feel confident you're finding top talent that you can't find anywhere else.

13:51With LinkedIn Jobs AI assistant, you can skip the confusing steps and recruiting jargon. It filters through applicants based on criteria you've set for your role and surfaces only the best matches. So you're not stuck sorting through a mountain of resumes. Hire right the first time. Post your job for free at linkedin.com slash partner, then promote it to use LinkedIn Jobs' new AI system, making it easier and faster to find top candidates. That's linkedin.com slash partner to post your job for free. Terms and conditions apply.

14:33We're back with Verge senior reporter Liz Lopato. Before the break, we were discussing the origins of CoreWeave and its unique relationship to NVIDIA. Now I wanted to dive deeper into some of the clever accounting that's going on behind the scenes across the entire AI industry to pay for some of this costly infrastructure. You write in your piece that CoreWeave as an entity makes, quote, a horrible kind of sense that it exists to hedge other companies' risks and juice profits by offloading that risk into this company and offloading the risk of building data centers. somewhere else that can fail without hurting the big companies themselves.

15:08Explain that to people. It's taking on some of the risk of building data centers. And that is an undertaking. Construction projects, as I think everybody knows, often don't happen on schedule for reasons that have nothing to do with how good people are at building things. Sometimes if the weather sucks, you just can't build. That's just a thing that happens. And so there is inherently risk in building stuff. And then on top of that, if you think about chips, they frequently and very quickly become outdated. And so if you are doing a big spend on this generation of chips, and then NVIDIA comes out with a new, better generation of chips next year, you're doing another big spend.

15:47That is also something of a risk, right? Being in this business, there are these assets that depreciate. And so CoreWeave is taking on those risks. and Microsoft is just renting their compute. It doesn't have to worry about, you know, is this data center going to be something that I need to continue to operate forever? Are these chips going to be outdated next year? None of that matters. It's just, oh, this is a place where I can put excess business. We've seen this idea pop up in other places. Meta just did what's called a special purpose vehicle to build some data centers of its own. Companies that are situated much like CoreWeave that actually NVIDIA is investing in as well.

16:24Is this just the answer to the risk of we're going to build a data center, the cost might balloon, and we don't know what the return is? Is it just there's a problem and this is the obvious solution? It might be. One of the things that I've been thinking about here a lot are these special purpose vehicles. There are a lot of reasons why you might choose to use one as a company. One of them might be to limit risk, like the risk that something goes bankrupt or the risk that something might be affected by another part of the company. because these are like little contained fiefdoms that lets you feel a certain amount of comfort with what you're doing in a way that I think that a lot of these companies are embracing.

17:06And I, as a longtime reader and admirer of Matt Levine, he had written a thing about how these companies are maintaining their pristine credit ratings. Like Microsoft's is better than the US, I think. They're doing it via the SPVs. They're doing it via some of these deals where they're taking on debt without touching their credit ratings. And he has suggested that this is because they think it's just really nice to have a good credit rating. And he's not wrong. It is really nice to have a good credit rating because it means you can borrow at a lower rate. But I also wonder how much of this is because these special purpose vehicles aren't subject to the same kinds of regulations as parent companies and have certain kinds of structures that may be advantageous.

17:51I understand why it's profitable to be the sort of middleman operator between your pristine credit rating and like the muck of growing a business. There's a lot of money in that middle zone, traditionally, across every industry and every category you can think of. But Corey is a public company. Like, can you sustain a public company on, we will accept your risk and hopefully AI pays off? My impulse is no, but I mean, we're gonna find out. Like, you know, I could be wrong. It wouldn't be the first time I've been wrong. But I look at CoreWeave and I see how much debt they've taken on and it really makes me nervous.

18:28You know, there's been sort of this ongoing theme, I would say, in the last couple of months where we've seen a lot of debt being taken out in the markets for this AI build out. And like, there's good reason why you would use debt rather than profits, mainly that you can get a bunch of money really quick and that lets you do the build out really quick. And then potentially, you know, that matters for the lifetime of the chips. Because if you do the build out really quick, then you get the maximum lifetime out of these chips. And maybe you make your profits faster. Like that's the hope, right? So the chips piece of this is the other thing that just constantly breaks my brain whenever I talk to you about it or I read anything about it.

19:02CoreWeb is taking on tons of debt. It's taking tons of loans. It's collateralizing those loans with the chips that's bought from NVIDIA. The payoff of those loans is that the chips will do extraordinarily valuable workloads, but the chips wear out, and AI as an industry has not yet proven that it can do a bunch of really, really valuable workloads. There's a chronology problem here that suggests to me that AI won't pay off before all of the NVIDIA chips literally wear out. That OpenAI will just do unprofitable Sora video generation at such rate that the chips will wear out before someone invents an AI workload that doubles the profits of Citibank or something.

19:44I don't see how those timelines actually align such that this business can succeed. Is there a better thesis out there than that? Well, so this is the thing that I'm really curious about. And obviously, drop me a line if you know, because the chip depreciation cycle really matters here. And there seems to be a fair amount of disagreement about what it actually is. For those of you who follow Michael Burry, who you may recall from The Big Short as the guy who nailed the timing on the financial crisis, he's arguing that the depreciation schedules that we're looking at right now are, in fact, too rosy.

20:19That may be true. I've barely scratched the surface here, but the thing that I have found is that a lot of things vary, right? So where the location of the chips are, what kind of infrastructure is around them, how they're being serviced, all of these other things, how heavy the use is. there seems like there are a lot of variables involved in how quickly these chips wear out. And so that, to me, is one of the central questions about whether this works out or not is, is this good collateral? There's one way of looking at this where these chips are effectively the same as a truck. And we have vehicles that get loans all the time and trucks depreciate.

20:53We have a pretty good schedule for the depreciation of the truck for your business. And so you can get a loan with the truck as collateral, and that's not really an issue. But because this particular kind of financial product is so new, it's sort of difficult to figure out what's going on with the chips. And so to me, that is one of the central questions here about will this succeed? Is like how good are the chips? How long did they last? I want to be clear. It's not just CoreWeave that's doing this. They pioneered this kind of loan. They were the ones who did it first. And that's because their CEO and a couple of other folks in their senior management, they don't come from tech.

21:29They're finance guys. They came up with this interesting financial product, which I personally am fascinated by. Then a bunch of other similar NeoCloud companies immediately followed in their footsteps. So that's like Lambda, Crusoe. There are other companies that are now sort of making these deals with private credit in order to get debt and buy NVIDIA chips. Does CoreV have a path to being profitable inside of all this? Even if AI doesn't pay off, right, and OpenAI never makes a dollar, they're still going to pay CoreWeave for capacity. Is there a path for CoreWeave to say, look, we just have the infrastructure.

22:08On some timeline, we're going to figure this out. We can just ride the wave and make money along the way. This is why the debt is important, because the chip loans are only one kind of debt that CoreWeave has taken out. There's a revolving credit line with J.P. Morgan. There's some vendor financing involved. There are secured loans with investors. All of these types of lending have different structures. And depending on when the loans come due and how this is structured, that's going to affect Corweave's timeline for when they have to have the money. Maybe as long as they have access to credit markets, they can extend that some, right?

22:40Like they can take out a loan to pay off the interest on the other loan and give themselves a little more runway. But eventually, you start to look at the company and you think, given the size of these loans, we really need to have a very fast takeoff for AI. that is built into the assumptions here. And so if you imagine AI is a normal technology, like a big and important technology, but a normal one, that takeoff isn't going to happen that way. I would like you to think about like online shopping, right? It took, what, 25 years to be 25 % of retail? And when people were saying in 1999 that you're going to be shopping online, they weren't wrong.

23:21Pets.com was actually not a bad idea. It was just early. the timing was wrong, right? Because if you look at Chewy, what is it? It is basically a much later pets.com and it was enormously successful. So, you know, you think about like how people work around new technology and the ways that businesses work around new technology. And you think, well, maybe there's going to be a slower adoption. So like, let's take, for instance, a normal business like that sells physical objects. We'll say widgets because why not? And this widget business may not have 1 % of its income that it can invest in AI right away.

23:57So it's going to continue using legacy products until either it can't, or it has an especially Goodyear. And that is going to slow the adoption. If we look at sort of the ways that AI is being adopted right now, the people who are taking it on first are the people who have these relatively sophisticated businesses. So we're talking about banks, we're talking about software companies. If you think about the physical world, that's not happening at quite the same rate. We need to take another quick break. We'll be right back.

24:41We're back with Verge senior reporter Liz Lopato discussing CoreWeave and whether or not the big promises of the AI industry can deliver actual returns before the huge bills come due. The CEO of CoreWeave is Michael Entrader. Last week, he told the Wall Street Journal that all of this money that's being spent is, quote, not a lot of money if the economy doubles in size. That's the bet, right? This all looks normal if AI can double the economy in size. Is that a reasonable thing to say to the market? Is the market buying that? Looks like no. And like, in fairness, I also think it's delusional. But like, you know, a court we've just reported, it's third quarter earnings, and the stock immediately tanked.