AI Bubble: ‘Something profound has changed’ | Ed Zitron

I have the books of a major data center company and they are losing hundreds of millions of dollars, hundreds of millions. This is most of the AI industry. The usual ability that the AI industry has to put the toothpaste back in the tube. It's not working anymore and people are starting to realize that the maths does not work. The more truth in numbers that comes out, the quicker this end. Hi, I'm Isaac and I'm glad to be joined again today by the writer of where's your Edat? And the host of the Better Offline podcast, Ed Zittron. Thanks for being here. Thanks for having me. So it seems we're in the midst of another tech sell off as it seems like the reality of a slow demand and expanding timelines for the predicted returns of investments are increasing. We've seen this a few times now. I think this actually kind of makes the third time this year if I'm memory serves. And I mean we've spoken before this. You didn't hear this, but I had, I can't remember exactly where, but an analyst saying that kind of the markets are expecting hyperscalers to be cutting back their capex and the data center builder at this point. But what are you hearing? Do you think we'll be back to blind optimism in a week or are you seeing anything different about it this time? So eager beavers who've been watching and listening to me for a while will notice that I spent last few months getting into the debt markets and that's because the stock markets are manipulated. They have been for a long time. They are emotional beasts. They go up, they go down, they go side to side, but the debt markets less so they tend to sure they can be emotional, sure there are dumb investors, but they tend to boil and when they go over the edge, it's kind of hard to get the the thing back in the cup. That's a terrible metaphor, but you get my point. So this week core weaves debt. So when bonds sell off the expected yield goes up for the debt. Long story short, their debt touched 16%. One of their fixed term loans touched 16%. So we are, and I just checked just before coming on it was all of them are above 14, one of them is above 15.3 point is these are personal savings, they're personal loans rates at these points which makes just borrowing for core weave and by extension most neo clouds and most data center SPV is just a ton more expensive. So the stock market is puking for reasons we'll get into, but it feels like Suddenly everyone started doing the maths. I did a year and a half ago. Year and a half ago. Gartner, bloody Gartner. Who they've been fan of the metaverse, whatever. So October 2025, Gartner said Neo clouds are reshaping AI infrastructure and that tech service leaders who fail to integrate them into their portfolio risk higher costs, slower innovation and diminished competitive advantages in the AI services race. So Gartner said yesterday that by 2030, 80% of neoclass will be dead. So what's happened is suddenly everyone went, can CoreWeave pay its debts? Could you? They've raised tens of billions. Could they pay that if something goes wrong and suddenly everyone's saying no? Nope, they're not sure they can. They're not fully accepting that core weave. They've not pledged themselves to the graveyard smash quite yet, but they have definitely seen something ugly. And I think it's just this. There is something profound that's changed in the last few weeks where you've got all sorts of economists going. Goldman Sachs came out with a report, I think it was last week, week before, where it was like, how much money would hyperscalers need to need to make just to cover the CapEx that they're spending in 2026, 2027? The answer is to just break even. I think they need like $360 billion of brand spanking new AI revenue at a time when they've got maybe a buck 40, buck 50, buck 70 if they're lucky. Now it's 425 billion if they want to have a 10% return on invested capital. So suddenly everyone's doing these things and you've got a bunch of people saying, oh well, we need trillions of dollars of AI revenue to make CapEx worthwhile. If only someone had said something, Isaac, if only someone had said now it could be anyone. But Gartner also added one other variable, and this is Gartner. These are the hypiest hypesters. They are an analyst group, but they're always analyzing to the benefit of the tech industry. But they said through 2030, AI data center owners would need to generate a cumulative $8.5 trillion in revenue to, to cover the announced capital spending, the financing cost and a 12% return on invested capital. Now these are Companies with like 20%, 30% ROIC on their investments in general. So this is what we call bad in the financial industry. So the stock market right now, we'll get into the anthropic stuff in a bit. Stock market's going Down. Everyone's worried. The Korean stock exchange, the cost BK PI has dumped, it's puked two days straight, could go back up because Jensen Huang goes out and said actually the numbers will be even bigger next year. And people go yay. But something has shifted where the usual ability that the AI industry has to put the toothpaste back in the tube, it's not working anymore. It's not working. Confidence is not being restored. And then there's this anthropic stuff that I know you want to get into, but there is something that has happened and I don't know where it is, but there was an information story saying that. I don't know if you remember, I've brought up MUFG and smbc, the two Japanese banks. Remember I brought them up in the past. Yeah. There's an information story about a week and a half ago that said they become more selective, more selective, more selective now on how many data centers they're going to invest in. Now these are the loosest legs other than Blue Owl. They're actually funding Blue Owl as well in the data center financing. So it's the little things I've been warning about the last few years. They're all starting to come together. And the problem with the AI boom bubble, whatever you call it, is that this is not something that's being substantiated by revenues or returns. It's hype, it's interpreting numbers and saying, I've read the tea leaves and the, the run rates and the big numbers in the analysis, the analyses and I decided that AI will be the biggest, most hugest and wettest sun ever. The problem is is that the sun is dry, this boy's never touched the water. And people are starting to realize that the maths does not work. And the problem is is that anthropic was the last key to be jingled. It was the last thing they could show that would improve things. And well, the numbers came out and they weren't what people needed. The sell off hit core weave the hardest as far as I can tell so far anyway. And so obviously I had to take a little look at the share price going back the years and year to date, which is one that really surprised me is that they are down 18% from the date of recording a year ago and they are only about 20% above their debut IPO price as well, which is also not good for a year and a few to several months later. I mean, I saw a great headline in the FT which kind of sums it up, which Is it basically said core Weaves bonds are screaming in financial distress? Yeah, which sums it up really nicely. And I was just wondering like, how close you think we are to a bit of a tipping point with Core Weave. I think we're. I think we're there, but a tipping point is more complex than just think bad. So let me explain. I've kind of mentioned this with Will, but I want to go over it again. So the way you should look at debt in general is whatever. The last time, the debt of the company in question priced at, which means when Core Weave has a loan or a bond. Pardon me, that sells off on the market. When it sells off, it's cheaper. Thus the effective yield, because you're paying less on the dollar for it, is higher. Whatever Core Weave's debt currently trades at is roughly where the debt will land, if not higher. If they borrow from someone. This means that the average cost of borrowing for Core weave now, it's not going to be below, it's not going to be below 10, might be 11, 12, 13. Ridiculous rates. This is also Core Weave is arguably one of the best capitalized, best connected, best funded Neo clouds out there, which means anyone raising data center debt right now is both in the toilet and being flushed at the same time. It's just, it's going to be nasty. Volta, which is this brand new neoclass, barely a few months old, apparently raising a leverage loan right now, real toilet loan there over 11%. And the banks in question, I think it's JP Morgan and Goldman. I think it is. Pardon me if I'm wrong on that one. They are not sticking around. They're trying to sell off more of the debt. They're not keeping it. Now, generally, when the bank thinks it's going to get paid, the bank keeps the debt. They keep it because they want all that money. So why wouldn't they want to keep it? It's almost as. And yes, there are complexities to the debt markets. I know people sell things, they buy things, whatever, but. But they're not holding quite as much. Why? It's probably because they don't think they're going to get paid long term. And to quote said the CEO of the company from margin call, they're selling to willing buyers. But the thing is, it's not so Core Weave and the AI bubble writ large. One of the mistakes I've made in my analyses is the. The early ones, I would say, is that I've been looking for an explosion. When you're looking for, to quote my mate Casey, A dead body, it's a slow boil. You find corpses here and there. Things die, things don't happen, contracts don't get signed, debt doesn't get raised. So it's less that Core Weave suddenly explodes and more Core Weave misses payments. And Core Weave cannot raise more debt. There was an analyst I found that said that Core Weave needs to raise another hundred billion or so in debt by 2030. So across the next three years at what, like you're talking 10, 11, 12, 13% rates. At some point, even if you think AI is the most important thing ever, there is a point at which this debt is unsustainable against the potential return. Because even if this company is cash flow positive, when you remove all the costs, when it's just like when it's just on a pure operating expenses basis, the problem is depreciation, which is an upfront cost amortized over time, is still going to eat into their profits forever. At this point, like just most of the chips they bought haven't been put in service. Most of the data centers that Core Scientific is building for them haven't been put in service. So the depreciation hasn't even started for most of this stuff. So Core Weave's just, it's more like a disease ridden company than it is going to explode now. Crazy things could happen. Nvidia CEO Jensen Huang just said, we don't need to backstop Core Wave. We don't need to backstop them. Which is funny because they've already backstopped corewef. There was a story in Business Insider back in February saying Core Weave says that Nvidia has its back on leases. Backstop like, come on, man. But the thing is, why would I possibly rely on the financial press to push back on this? Why would I possibly imagine the height of the financial press would say, hi, Jensen. Hi. Did you know that you already did that? Why did you say that if you already did it? But now why would we need object permanence? This is only the literal underpinning of the entire bloody stock market. I don't know. I'm very black pilled when it comes to financial journalism right now. It drives me up the goddamn wall because it's like even I should I talk about anthropic already? I'm go for it. Go for it. Eager to go for it. Okay, so we should launch into the anthropic. So the reason that the markets are puking today, which is Thursday, so this will be running on Friday, is because anthropics Q3 numbers came out in a leak from an investor meeting from leading up to their IPO. Now, their Q3 numbers were they had $16.8 billion in revenue. Sounds great, right? Well, they had $11.5 billion in Q2. This means their growth rate is 46%, quote for quote, which is huge. Wow. Except it's not enough. It's not enough. They had 4.8 billion in Q1. It went from doubling to not even going up 50%. Dario, that doesn't sound so good. But then when you dug into it more. Oh, we have an operating margin of 11%. Adjusted operating margin. You know, it'd be profitable if it wasn't for all these costs. And then they had a gross margin of 85. No, sorry. Adjusted gross margin of 85%. I wish I could do this with my, like, I want to buy a 50 billion dollar house. Yeah. If you just remove all my costs and if you extrapolate into the future, I'm going to be a trillionaire. Give me a mansion. Give me a give. Build me a palace in Las Vegas. I want one. But the other problem is, is that you remember run rates, right? You remember my complaints about run rates? So annualized run rates, take a period of time of revenue, multiply by a number. No one knows what the revenue number or the period of time means, but you get a magic number out of it. Latest one from anthropic was 65 billion in July. Wow, what a huge number. And then Anthropic neglected to make up a new one. They didn't include one. So AI boosters are all bringing the manila, like Bud Dwyer style, bringing the manila envelope up to the podium today. Because they've all kind of realized that despite this being genuinely impressive revenue, that's a huge amount of money. It isn't enough. Anthropic has $413 billion worth of non cancelable commitments, per Reuters. They also revealed that 5 to 7% of their revenue came from distillation, as in Chinese AI model companies using Anthropic's outputs to train their own models. And they. Anthropic claims they've killed it. Yeah, it's really bad, man. It's so bad because it's like making so much money off people stealing from us. Also, the stealing machine gets stolen from. But the funny thing is, people like, oh, that's fine, it's fine. It's not a big deal. Would you say that if Microsoft said that 5 to 7% of their revenue went away? Hmm. Would anyone know? No one would. They'd be like whoa, what are you talking about? 5 to 7% of your revenue. That's bloody huge. But then there's the other problem. That revenue is decelerating at a time when it has to accelerate. Anthropic has to be they were meant to per investors with rebar in their skull meant to have 100 billion annualized run rate. Who gives a they were meant to have this by the end of the year. It's obviously not going to happen. But you can see the AI boosters, they're all starting their cars in the garage a bit because they're all realizing that they heard they, the boosters, they, the AI companies had made these huge promises. They'd said we are going to grow eternally. Reuters reported a few months ago that Anthropics IPO their valuation hinges upon getting to 180 to $200 billion in revenue by 2028. It's just the numbers are not there. And the problem is that it's not just oh, Anthropic can't afford its bills, it's that Google has taken on hundreds of billion, sorry 800 something billion dollars of off balance sheet agreements to provide compute to Anthropic. It's Broadcom has taken out a $35 billion private. It's a private credit backed thing with Apollo and Blackstone where 30 billion of the debt is backstopped by Broadcom as in if Anthropic doesn't pay, Broadcom will have to pay. Either way you've got hut 8 fluid stack all of these connected counterparties that must have Anthropic grow because if Anthropic does not grow they do not get paid the money. But then there's the other problem which is even if the distillation wasn't taken out, which it's gone they've said Anthropic the story ran three days ago that said oh we've reduced the attacks by 90 to 99%. Just meaningless numbers again. Even then Anthropic will have to just to match its current growth rate they will have to make $24.5 billion in the next quarter and the quarter after that. I can't even do the math off the top of my head. I just did it beforehand I'm going to admit is this to keep up with the growth trajectory it has set itself. Anthropic needs to be cooking at like 35 to 40 billion dollars worth of quarterly revenue in the next year without fail because otherwise you just, you're not going to make it. And this is the funny thing, I think much like Core Weave, the numbers have always been there. The numbers, they, they've always been floating in the ether. Everyone's kind. I've been saying this for like a year and a half. People have said, ed, Ed, you're wrong because it hasn't happened yet. And by the way, I was early, of course I was. I know. Here's the thing. The things that people are criticizing Anthropic and Corey for are the same things I've been saying, which is to do all the things you are promising to make all the things you have said will happen happen, will require you to grow at such an aggressive rate that everything has it's price for perfection. And this is not a perfect company because even with this rapid revenue growth per ramp, 80% of Anthropic and OpenAI's enterprise revenues are from the 1% of their customers who are mostly AI startups who are selling services where you can burn for $20 a month, $100 a month worth of tokens from Anthropic and OpenAI. And even with significant discounts, that money don't make sense. For 200 bucks a month on Cognition you can burn $1200 worth of tokens from OpenAI and Anthropic that don't make no sense. But that is the revenue underpinning this growth story. Which means that Anthropic can't really sell its way out of this story. This isn't one where they can go to it's not like Anthropic's customer base is mostly big businesses as in a JP Morgan or a Goldman Sachs and say okay, we need you to pay more. Oh, we're up in the cost. No, it's mostly venture backed startups who will gladly let their users burn as much as possible, except even they are starting to feel the credit crunch. And so the way this all falls apart is a burn, a boil rather than an explosion. It's going to be a decay of these underlying numbers because at some point Anthropic is going to have to do a prepayment to for all of this compute capacity that it's promised at some point. And we haven't even mentioned the largest wettest sun in the valley OpenAI. But OpenAI faces every single problem I'm talking about. Plus plus plus because clammy Sam Altman has made even more promises. The Anisa Gardizzi at the Wall Street Journal reported a few months ago $750 billion worth of planned commitments compute commitments through 2030. Where's the money coming from Sammy? The answer is nowhere. He's not both of these companies. So a few weeks ago I did an analysis a week ago even where I said I think that both of these will get rated triple C. So pretty much as bad as it gets credit rating before they are literally just saying this company is in default, they'll never pay their debts. This isn't one where they can like SpaceX trick their way out of it and say oh actually we're going to be, we're actually we're going to, we're a bit profitable. We just, if we pull back on the capex. These companies don't have capex. OpenAI and Anthropic do not buy chips. Their sugar daddies do. The Microsoft's, the Googles, the Amazons, the Core waves, the Nvidias in some cases. But fundamentally the mathematical mismatch is becoming too hard to really bear. And while these companies might not run out of money immediately, the problem is is that the counterparties, these various data center companies run on debt, tons of debt. And the debt, even though interest rate, the 10 year treasury bond has dropped a bit even then they're now being priced both on that, which is still elevated, still above 5.08 I believe percent. They're also paying the data center premium and the data center premium just to be completely upfront with what that means, the data center premium is hey, are you going to exist in three years? And that is no longer a foregone conclusion. Much like Anthropic growing to this meteoric size of a business that's bigger than Microsoft and Google is no longer a foregone conclusion. And it had to be. The whole AI trade hinges upon the idea that nothing bad will ever happen. Hard to follow that succinct summation of everything that's going on. But I just want to really drill down on something there and maybe try and use one of your analogies, maybe maybe ruin it in the process. You talk about it being a burn or a boil and the leaked figures from Anthropic and OpenAI a month or so before that is the transparency essentially of the finances behind the AI boom. AI bubble. Is the transparency essentially oxygen to the burn? Well, the opaqueness. Yes, yes. So you'll notice that it's actually, that's actually a very good point. So my newsletter this week was about how the AI bubble is basically inflated by fake numbers. You'll notice that everything starts falling apart when people do real numbers. The market didn't freak out because Anthropic didn't give an annualized run rate. They freaked out because Anthropic showed its actual numbers. The reason people are freaking out about coreweave is not because Core Weave even did something new and bad. It's because they looked at the real numbers. You'll notice that when people freaked out about OpenAI, they freaked out about its real numbers when I published them. People are freaking out because the AI bubble, like every scam and con, is built on making you look always in the future, never in the now. Never think about what's in front of you right now. Never look at the reality. Look into the imagination of the Gavin Bakers of the world. The genius, the. The dunce. The dunce apocalypse is here because the dunces can no longer fool you with these numbers. And people are freaking out because we really haven't had good data. We really have. It's crazy when you think about it that this is only like the third or fourth time we've really had tangible numbers on Anthropic. Yeah, first report. First report was. No, actually May only be the second because Berbidgen, the Journal reported in. I want to say May. Good pull. If I did May, he reported Q1 and Q2 for anthropic 4.8 and 11.5. And then this Q3 is only the next one. Sorry. Four times Reuters had the 2025 numbers. 4.6 billion. We also don't know Anthropic's costs from these quarters. These latest ones, they. The fact that we don't know those costs is usually because they're not great. People have worked them back. Perhaps Anthra. I've heard, I heard from a source that perhaps Anthropic pulled back on training deliberately and apparently a lot of the cost problems with these companies are they're constantly oscillating between how much they do on inference, how much they do on training, and when the service gets dumber. Any. Any of the audience who's had a model get dumber or experienced or heard about that, that's because they pull back on how much inference is available because they have to train more models anyway. That push and pull is probably how they're pushing even their fake adjusted margins up. But when you look at the real numbers, people freak out. People don't like it. It's why we use run rates. Because think about it. Doesn't 65 billion sound better than 16.8 billion? Isn't that it's a higher number. I like big number. Why not use big number versus small number? And you've got. I'm going to be honest, the tears I'm drinking, give me more, I want more. Email me all of the funny AI booster things with them freaking out. But these people are saying things. I saw one of the most noted boosters yesterday or today say we had too high expectations for these companies. It's not a sports team, it's a company you do like. This is not. This is not a sports game. This is not the Detroit Lions. We are not the. Sam Altman is not Dan Campbell. Damn Camp Dan. Dan Campbell is a better coach anyway. But putting all that aside, the boosters are freaking out because suddenly. Because actually, let me take a step back. Suddenly we've gone from this point where boosters kind of controlled the narrative because boosters would flood the zone with all of these fake numbers, with all the hundred billion dollar number ARR number run rate number that I've seen that came from boosters. There was a number in the Reuters story yesterday where they said, oh, $110 billion ARR was at run rate. I got to keep saying it because ARR is different. 110 billion run rate was what Anthropic was saying. Reuters then had to roll that back and say, no, it was a sell side analyst that said that. And I must be clear. I would trust a homeless man on the street more than I would trust the sell side analyst on anything to do with AI right now. The sell side analysts are bought and sold. I don't know where the hell they're like, I don't know. The Dan Ives of the world and the various analysts who have boofed on this one deserve punishment for their failure. But suddenly the boosters and the sell side cannot flood the zone with fake numbers anymore because the real ones are here. And I think Anthropic knows. And the reason they're sharing the cost is more data allows people to come to intelligent conclusions based on reality that might not look so good for the AI trade or anthrop. And we just don't like that in this business. We need to keep the con going. And this is ultimately the problem. This is an industry run on bad data. Good data destroys bad companies and there's a ton of bad companies to be destroyed. To me, playing around with annualized revenue and all those sorts of things. And I think there's that one rumored case of multiplying the July 31 by 365 days and saying, there you go, rational analysis. I mean, that just tells me that companies like OpenAI, Anthropic just don't have confidence to stand by the reality of their business models, which is about a red flag as you can get. And none of this is a secret is the thing that gets me. And if you were to obviously talk about the key jingling thing, but why or I guess how do venture capitalists and financial analysts just indulge this metric? So I actually asked a mate of mine in fixed income, good mate of mine in bonds about this because I was confused too, and he gave me a nicer answer than I expected, which was people try and be optimists and people try and be optimistic about stuff. Problem is that optimism isn't. Optimism is very short term because people conflate optimism with blind faith. And also it sells people. One of the most annoying criticisms of mine, other than all the others, is the cynicism sells. Skepticism sells. No, complete and utter fabrication from the most egregious lie of them all is that cynicism or skepticism sells. The crap I get in my emails every day is insane. No, what sells for like a podcast or a major, major radio network or various columns you may get is talking the company's books is saying, oh, this will be good. Number will go up. Especially because it's much more difficult to look at the world as this complex and wrong thing and say hundreds of billions of dollars got invested in something for no reason or under bad ideas, or that the real numbers are actually bad and that we shouldn't believe the companies. It's much harder to say that than just look at all the money and the consensus and here all the pigs oinking the same way and saying it's much easy to just oink with them. It's much easier to find an intellectual way of agreeing with everyone than it is to find a rational, intellectual way of disagreeing with them. And weirdly enough, our society also punishes skeptics more. People always say, oh, firebrands. Oh, they get more attention. No, actually, there's way more money in being. No. If I suddenly did like a how I left the left thing and went pro AI, I'd make so much more money. Crazy. I'd make holy crazy. They Sam Altman. We'd have me on a chair like a wedding. Like, they'd love it. But that's the thing. There is so much more money on going on CNBC and being like, number. Look at this number. Wait, you like this number? Wait until you see the next number. Wow, did you hear the leading financial press to this day still uses annualized run rates and never defines them Never all the people at Bloomberg, but the reporting around this being a dog's asshole. It's absolutely ridiculous. The amount of times that they said run rate and just go a period, a shorter period, annualized, it's like what does that mean? Nothing. But again, what gets clicks? Positivity. What gets attention? Positivity. Even now, even as everything's falling apart, what's getting the most attention? Positivity. Except when they are all starting to say the same thing as me because it's becoming a little irrational to seem positive. And people are going to be able to sell positivity for a little bit longer, maybe even as much as a year, maybe less. Probably less. But nevertheless the point I'm making is that they won't be able to sell it much longer if more numbers like this come out, if more underlying details like I going to say it here, I'm not going to say what these numbers are. I have the books of a major data center company and major private one and they losing hundreds of millions of dollars. Hundreds of millions. This is most of the AI industry and when you tell people hey, this is what's going on, they say yeah, but they're the fastest growing companies of all time. Look, I used the computer, made me a little computer program when I did chatgpt. What's wrong with you? Data centers are so big. Oh, you're going to say that the smartest people in the room are wrong. They know that's what the title of the Enron book is. No, these are the different, it's different this time. And those are the most dangerous words in any hype cycle. You ever hear someone say it's different this time? Imagine it's instead of a guy on tv. Imagine it's your boyfriend after he cheated on you. You going to trust him? You're going to trust that boy? Or did he have sex with someone else? I don't know. I don't trust anyone who tries to sell me on something based on a vague promise of the future. That's not how reality works. And people will rationalize this until the very end. They will hear every warning light going off at once and they will to the end say that AI will be the biggest, most hugest thing until enough real numbers and I think it's going to be that S1 from Anthropic and OpenAI. If OpenAI goes public and it's an if now, it's not a when. The more truth in numbers that comes out, the quicker this ends. We should say obviously that a part of this sell off is also being fueled by the worries around inflation and how it might keep up interest rates higher for longer and. But that also itself does lead back into the AI bubble of it all as well. The higher interest rates make funding hundreds of billions, if not trillions of dollars of data center pretty much impossible. And I suppose, do you think that's kind of being factored in much here? Because I'd expect a much larger reaction if investors were predicting debt defaults. I mean, obviously Corweave's interest has gone up, but is this whole situation just like holding a wolf by the ears? They'll get bitten if they let go, but the wolf is only going to get angrier the longer you hold on. I like that. That's actually a great way of putting him because I've said this for a while. Actually, let me start with the first thing. No, hyperscalers are not defaulting on their debts. Core Weave, Oracle, everyone thinks. My big call is that OpenAI dies. It's Oracle. Oracle's dying. Oracle will die. You heard it here first. Unless someone else said it. I didn't read it first. But anyway, the Wolf by the Earth thing is true because it's what I've said for a while, which is hyperscalers are doing this. The kindly thing people say is, oh, it's an existential thing. The risk of doing this, of not doing this is higher than that. Mark Zuckerberg was on a podcast called Cheeky Pint with Patrick Collison, founder of Stripe. It's a bar built inside. I swear I'm going somewhere with this. A bar bill inside of Stripe hq. Mark Zuckerberg and Alexander Wong, gremlins, Gremlin Duo goes on there and Mark Zuckerberg says that he did not know what he agreed to. 100 to $200 billion of capex without knowing why, just he decided to do it. And then Alexander Wong added the at a off site for Meta in a in it seemed like recently the CFO of Meta, Susan Lee, I think it is, came up to her, came up to him repeatedly and was like, hey, hey, hey, if this model doesn't work out, if this model flops, it's really bad financially. Very, very bad financially was the words he used. So all of that's to say is that right now, hyperscalers, they have the wolf because while they keep spending, they can keep saying there's a payoff in the future. Perhaps the markets get pissy with them. But Satya Nadella can go on a podcast and go, yes, Actually, we will see biggest AI numbers you have ever seen. Sundar Pichai can do the same thing. They can do this boring slop and the analysts and the journalists will go yay number without ever being given an AI revenue number. But the. And then the markets will say, well, they're investing in the future. Now there is a point at which that becomes untenable. But the biggest thing is, is they know that if they stop, the markets will go, great. They'll get a stock bump out, but they'll go, brilliant. Love it. You're not spending as much money. Great. Why did you do that? Why did you spend all that money? Just love the fact you're not spending it. Awesome. Why though? What was the plan with that? And why did you stop? Also, that's the kind of devil's deal here. Stopping will give them a stock bump. It will make their financials better. But they will both be asked why did you do that? But also why did you stop? And they will most likely slow rather than stop. But even slowing, they're going to get these questions and it's because they have no hypergrowth ideas outside of this. They don't have a next iPhone. Well, I mean we're not talking about Apple here. So they don't have the next Google search. They don't have a next Facebook. They don't have a next Microsoft 365. They don't have a new thing other than this. And AI is obviously not growing their other things bigger in a way that matters because otherwise they'd be telling us that it was in definitive terms. Again, fake numbers. Microsoft 2/4 ago said 37, $38 billion of AI run rate, whatever the hell that means. No real numbers though. None of those for me, thanks. The fake numbers will only help so far. But the moment that they stop showing complete conviction, they're dropping the wolf. It's eating them. And it won't just. And it's not going to be a short problem either because what it will be is they'll stop doing the capex. The cash flow will improve, they'll have a good quarter, maybe another one. Then the market's going to say, where's your growth? Where's your growth? And the answer was, by the way, that the growth was coming from anthropic and OpenAI's compute spend. And anthropic and OpenAI's compute Spend was funded by venture capital. And how, by the way, was anthropic and revenues compute spend financed other than that? Well, it was from revenues. How those Revenues came in from AI startups predominantly spending their money by subsidizing, using venture capital money. Everything chains together, real nasty like here. And it's funny, when this collapses you are going to see every major media outlet say stuff which they're already kind of doing. That I've been saying for years. Jefferies, Bloomberg reported on this the other day. Jefferies said that the most likely outcome, the most likely long term outcome for the AI bubble was that there would be large capital destruction which is not again, not really what you want to hear. There was also, forgive me for not. It was Alex Altman over at Barclays who said this as well. He was saying that there is, I think he said $250 billion worth of revenue but there's trillions of dollars of expected capex. And that outside of, I'm paraphrasing here, outside of whimsical numbers, there was no clear explanation of how it be paid. And this is the thing, I'd have said that a year ago personally. Oh wait, I did. But on top of all of that, nothing has changed. It's got bigger, it's got uglier, it's been allowed to go on longer. But the fundamental economic problems are what they were in 2025, which is that the AI labs cannot afford their businesses. Their businesses are unsustainable. That running AI data center compute is clearly not profitable. If you were scaling, perhaps it is if you just built one. But no one just builds one. Everyone builds multiples and they're all so big. And we also do not yet have a fully operational gigawatt data center, so we don't even know if that's possible to make a profit after five or six years. But it's the same problems. It's the same. I wrote about core weave in March 2025. I said the same things. This is an unstable company full of debt. It's the same things. And it's going to be one of these things where I think of the big short, where it's the two guys are celebrating because their shorts paying off and Brad Pitt turns to them and says, I think it's unemployment drops. Unemployment goes up 1%, 30,000 people die. I will not be happy. I will be vindicated. I will. And that will feel good. I feel proud. But watching this happen is going to be nasty. It's going to be nasty for everyone involved. And the people that believe the hardest are the ones that get burned the most. Just finally, while I've got you, if I can, I wanted to get your Reaction to the apparent self replicating AI worms that are coming for all of us. Because on the one hand we need billions of dollars of data center for AI that we've talked about, but now it's also going to be installing itself on my smart light bulbs or something. Yeah, it's funny, it's funny how every, every time something goes wrong for anthropic, they're like, hey, my disastrous son is destroying you. So his, their argument there was that they were able to in a limited setting, get a model to copy some it its weights onto an IoT device. But also it was unclear how likely this was and they said within a few months open source models could do this. Not really sure what women meant to do with this warning. Doesn't really seem like anthropic acid thing. But I should also be clear, it would be like point 9 tokens per second which is the slowest of the slowest of the slow. Yeah, so that's three quarters of a word ish every second. So that could do something. But we talking about generating software code because barely going to get like a terminal string out in the space of a minute. What's the plan there? It's not great. If it grew it would be bad. But also it's like, what do you like? Why are you warning us about this? Is this so that we don't fund Chinese models? It kind of seems like this is more about China than anthropic. Are you going to, perhaps you'll build a security suite that magically deals with this stuff. I don't know. Also feels like something that you could very easily check with these things. But their argument is that it's these cheap Chinese devices that could do it, but the speed at which these LLMs would operate on these devices would be so slow that their argument is that just over time it could be a problem. It really suggests what could happen as a result. Which means that they don't really know. Also feels like just one of those problems that they came up with and then they kind of brought into the world in a way that you kind of wonder why they told everyone. I don't know. I guess it's good they did, but it's just so frustrating because it's like mate, you got bigger problems than this. You got bigger problems than a Chinese thermostat potentially doing cybercrime over the course of several months, you, your company's running out of money and I don't know, it's one of those things where I'm like the people are anthropic. I no longer think have any idea what they're doing. I think that every day they come into work someone says, yeah, we're destroying society. What's your likelihood? I'm going to go with 10%, guys. Like 25% over here mate. Great, let's go back. Let's bring that number up. Let's destroy humanity. But also only we can do it. No one else. No Chinese destroying, destroying society. This is an American job. We already have the CIA, like what the heck. Anyway, that's a joke. Please don't kill me. I don't know, I, I actually but before we go, I know this is a complete hard term, but there was an inflation thing that just occurred to me just to close us off. So completely different question. I'm sorry but there is a unique problem with AI and inflation right now, which is governments generally raise interest rates to stop businesses borrowing so much. They say, okay, we're going to make them higher and higher and higher and that will stop businesses borrowing so much. So people get laid off. By the way, it sucks really bad. But it also means that they will lower the prices on the goods because they sell things. Problem with this is manifold, which is first of all, hyperscalers. Data center providers will borrow at whatever Volta's borrowing at 11% like they don't care. These deals will get done because they're not particularly sensitive because they don't care about whether it gets done. They just want the money in the door now so they can start a nebulous project and start paying themselves salaries. These are evil people. Hyperscalers also don't care. They'll pay 7%. They don't care. Sundar Pichai don't care. Microsoft doesn't even really have debt, like other than the Activision stuff. Meta doesn't care. Look at Mark Zuckerberg's the way he dresses. He doesn't care about anything. And so you can't really control it with interest rates. But the other problem is, is that there is an inflationary issue here which is ram. Because RAM is the reason the cost of a lot of goods go up. I haven't done the numbers here, so I don't know what chunk of it, but basically everything that has a lot of things have RAM in it that you wouldn't think like your fridge probably has ram, your computer does. But maybe your weird stream deck you have, maybe you have a webcam that will have RAM on it. All of the prices of every single thing with ram in it. TVs most televisions are going up. So raising interest rates would theoretically hit that. Except the reason that those are going up and the price of RAM is going up is a data center, CapEx. So you can't bring that to heel with interest rates despite it being an actual inflationary force. But because it's not like interest rates going up are going to make Micron or Samsung or SK Hynix lower their prices. They're doing that because they're a triopoly. They can set the prices however they want. And they also have tons, billions and billions of billions of business from these large companies who don't care about interest rates. So it's not something where you can put the screws on them. So there is no bringing this under control unless you bring the companies under control. And they're not going to do that. So inflation is going to keep rising. Oh, and also we have a war in Iran and a war in Ukraine that's also affecting fuel prices. I don't think interest rates are going to help on those either. I don't actually think that that's going to change anything. The other problem is, is that consumers are currently the most indebted they've been, I think, in history would not surprise me. So raising those interest rates is actually going to make the problem. Well, maybe it will reduce consumer spending, maybe. But people are already putting stuff on Klarna and Affirm or Wonga. Is that one of them? That sounded like one of them anyway. But they're putting them on credit cards. They're putting them on credit cards that are going to get more expensive. This is going to keep happening and it's going to get worse because raising interest rates raises interest rates on the credit cards and the personal loans they might be taking, but the fuel costs are still going to go up. So they need to take off more on more debt to pay for their car to get to work, you know, to make the money. That's not going as far. This is something that you can't. The actual thing the government could do would be to end the AI bubble. They can't do that. They won't do that. No one wants to do that. But that would probably bring down prices or end the war in Iran. And I don't get the sense that Trump is sitting around going like it's time to sort things with the aip. We're going to fix this, we're going to talk to them. No, no. He lit the other day, Donald Trump said something like, along the lines, like it always, it'll work out. It always does. One of the best. One of the best things that you can say in a bad situation is just, it'll be fine. But that, my friend, is the AI bubble. That is the you talk to anyone. How will OpenAI work it out? It work out. It always does. It's just like Uber. It's just like Amazon. It's just that you. Why would you possibly worry about this? I worry about it because I can do mathematics. Well, Erzitron, thanks for taking the time. Thanks for having me. If you enjoyed today's episode and you want to hear more of the Tech Report, please consider liking and subscribing. Also, you can get episodes of the Tech Report wherever you get your podcasts.