The AI Demand Bubble with Ed Elson

Tech earnings and AI trade

0:00This is an iHeart Podcast. Guaranteed human. Amazon Health AI presents Painful Thoughts. I, um, I can't stop scratching my downtown. Yeah, but I'm not itching to go downtown and tell a receptionist I'm here to talk about my downtown. Some things you'd rather type than say out loud. There's no question too embarrassing for Amazon Health AI. Chat your symptoms and get virtual care 24-7.

0:30Healthcare just got less painful.

0:34Hi, I'm Steph, and I work at UnitedHealthcare. When I think about UnitedHealthcare's members, I think of my own family. My niece was diagnosed with an ultra-rare genetic disorder. I know how hard it is for families. At UnitedHealthcare, we can make it easier. I'm Brian. I care because I know what it's like to not speak the language. It's not easy. Health insurance is complicated. I would say our job is to make everything easy for our UXC members. I work for UnitedHealthcare. I'm committed to care. This is Jacob Goldstein from What's Your Problem?

1:07Running a business is hard enough. Don't make it harder with a dozen apps that don't talk to each other. One for sales, another for inventory, a separate one for accounting. That's software overload. Odoo is the all-in-one platform that replaces them all. CRM, accounting, inventory, e-commerce, HR, fully integrated, easy to use, and built to grow with your business. Thousands have already made the switch. Why not you? Try Odoo for free at odoo.com.

1:38That's odoo.com. Odoo is over the energy. The hunger speaks for you and your plans are not feeding you. What are you waiting for? Look at 7-Eleven and try the sandwich of chicken to live red. Really spicy. Enjoy a intense heat in every bowl full of flavor. Available at 7-Eleven for just $4.99. Visit your 7-Eleven local to Chattanooga and satisfy your wish today. Válido hasta el 25 de agosto de 2026. Tiendas participantes hasta agotar existencias, ingredientes y salsas por separado. Copyright 7-Eleven. Todos los derechos reservados.

2:12Call Zone Media

2:30That's right, we are back with Prof G Markets' Ed Elson, and we're here to talk about the naughty little buggers in the Magnificent Seven, primarily the big four, but really the big three. The Amazons, the Googles, and the Microsofts of the world. Ed, thank you for joining me again. Thank you so much for having me. I'm very, very excited to be here. Yeah. So, tell me, they just reported earnings, and as you well know from the conversation we just had, I got some points to make about it. But why don't you run me through how they did on their earnings?

3:02Absolutely. So, let's just run through what we learned on Microsoft's end. We saw pretty good growth, $90 billion in revenue, up 18%. Amazon, similar story, $200 billion in revenue, up 20%. Meta's revenue increased 28%. I mean, on its surface, this was pretty good. Yeah. And then you just look at the numbers, even the net income, which exploded for many of these companies. But, of course, that's not really what matters here.

3:35What really matters is how is the AI trade actually working out? There are two main questions that we really want to understand when we look at these big tech companies. The first question is, how much revenue did you generate from your AI business, specifically? That's what we really want to know the answer to. And then the second question is, how much of that revenue came from two relatively unstable companies, financially, specifically, OpenAI and Anthropic? Those are the two questions I want to know the answer to.

4:05I know you want to know the answer to. So, we didn't get those answers. They didn't really tell us. And what they did tell us was not very helpful, kind of papering over the reality, which makes you think maybe the reality is something that they don't really want us to see. So, that's the funny thing as well. Like, before we get to the meat of what you and I both know we're going to discuss, there was also these weird massive equity gains. So, the net income part of earnings is now not super useful because in all three cases of Amazon, Google, and Microsoft, they all posted massive equity returns based on the increasing value of OpenAI and their shares of either or both OpenAI and Anthropic, right?

4:50Yes. Yeah, a wrinkle in there that I'll address, but we'll start with Amazon and Google. The answer to that question is yes. Amazon's net income, so their earnings, exploded 245%. That's crazy growth, more than tripled. Google's exploded 298%. Again, crazy growth. How did that happen? What's going so right? What could possibly not? You dig into it.

Unrealized gains skew earnings

5:14Exactly.

Unrealized gains skew earnings

5:15The answer, in Amazon's case, is that 85% of their net income came from unrealized gains in their stakes in Anthropic and OpenAI. So, essentially, they basically looked at that company. They said, okay, I guess it's worth this amount of money. I guess it's worth a trillion dollars. And they marked that up, and then that was reflected in their earnings. And then in Google's case, 87% of their earnings came from their unrealized gains in their stakes in Anthropic and SpaceX. So, those earnings of those companies, you can't really take seriously because they're not reflecting the fundamental underlying health of the business.

5:52They're reflecting the optimism associated with the valuations of these AI startups and these frontier labs. So, that's the first problem is the earnings themselves are not really reliable in Amazon and Google's case. We'll get to Microsoft in a second. And the other problem is that that then funnels down into how we value companies. Specifically, one of the most important valuation metrics in investing is the price-to-earnings multiple. That is how we figure out how expensive or cheap a company really is.

6:26We look at their earnings, then we look at the price of the stock, and we compare if the two match up. Now, as of the end of last week, which is when I did this analysis, Amazon was trading at 19 times earnings. Google was trading at 17 times earnings. The S&P, the overall market, currently trades at around 25 times earnings. So, if you were to look at those multiples on Amazon and Google, you'd say, oh my gosh, these companies are crazy cheap. It is time to buy. But of course, we know that actually, no, those multiples have been massively skewed and compromised,

6:59again, by those unrealized gains, by those paper gains. And just to be clear, that comparison is always based on net income. Yes, it's based on net income is equivalent to earnings. So, we look at their PE multiple, and they're way down. I mean, you look at last year, Amazon was trading throughout the year at around 32 times earnings. Google was trading around 31 times earnings. So, you look at 19 and 17, you go, oh my gosh, what's happened? They're so cheap. They're really cheap right now. That's not the case. So, what we have to do, if you want to actually get an understanding of how to value these companies,

7:32is you need to subtract out the gains that they have registered in OpenAI and Anthropic and SpaceX. You need to create, basically, a new metric. And we did that, and we found that the true multiple for Amazon, when you adjust for that, is actually 30 times earnings, and for Google, it's 31 times earnings. So, no, these stocks are not cheap right now. Now, a lot of people will say, well, are you factoring in the fact that maybe there is some value in those companies, and did you put that in the numerator, blah, blah, blah.

8:04There are a lot of, you know, nitty-gritty things that we need to do. And the answer is, no, we didn't. But that's the kind of analysis that Wall Street needs to be employing right now. And the trouble is, the more that we see these stakes in these little AI startups and these businesses, and the more that it starts to skew the actual earnings of the companies, it makes it a lot more difficult for us to actually value which companies are cheap, which companies are expensive. And, of course, it is in that complexification and in that obscurity that bubbles arise.

8:36This is the problem that we saw in the financial crisis. It's the problem that you and I have addressed when it comes to the data center SPVs. It's not necessarily that the SPVs are illegal. The problem is that they are hard to track. They're opaque. They're not reflected on the balance sheets of these companies, which means that Wall Street and the underwriters and the investors, when they look at the data, they're not looking at data that really tells them the reality of what is going on. So, that is the problem with the earnings. That's the problem with the PE multiple.

Microsoft and OpenAI losses

9:07I will pause there and we will get to Microsoft, which actually didn't have to report an increase in their earnings due to their stake in OpenAI. I thought they had an Anthropic bit that did that, a $3.2 billion gain there. Sorry. Yes, they did do Anthropic. But when you look at, I mean, Anthropic is a smaller investment for Microsoft. Their big investment is, of course, OpenAI, which did not show up in their net income. Their net income rose by about 30%. If we were to look at the OpenAI stake, you would think that it would explode by around

9:40300%, like Google's earnings exploded. That didn't happen due to an interesting gap accounting principle, which I can get into in a moment, if you'd like me to do it now. Why not? Go on. Why not go into it? All right. So, why didn't Microsoft's net income explode? Why wasn't their stake in OpenAI reflected in their earnings? And the answer is that their investment in OpenAI is so significant that because of stupid

10:13accounting laws and gap principles, they are required to report that stake differently. So, instead of reporting the alleged value of OpenAI in their earnings, instead what they have to do, because their stake is so large, and so I guess the law says you need to report this differently, instead what they have to do is they have to report their share in OpenAI's profits. But of course, OpenAI has no profits. So, that's actually why Microsoft had to take a $600 million hit on their earnings to reflect

10:46their share of OpenAI's losses in the second quarter. So, in a funny way, when accounting laws require you to report your AI investments correctly, it actually reveals that these big tech companies are actually losing money from AI. But when the accounting laws are loose and they're flimsy, it allows them to say, oh, I made money on this stuff. So, it's an interesting wrinkle there. And yes, their anthropic stake, they were able to kind of BS that in their earnings,

11:16but they weren't able to do it with OpenAI, which is why they took that $600 million hit. If we just play this out over the next quarter and the following quarter, what we can start to see is that actually these investments are not as sexy as the net incomes on Amazon and Google would make them seem. And the other thing with Microsoft that I found interesting, and I'm a curious little critter, so I went and looked, they mentioned they had annualized revenue, sorry, annualized run rate for AI in Q3 fiscal 26, which was last quarter.

11:47Just for the listeners, their fiscal year begins on July 1st of the year beforehand. So, their fiscal year 26 begun July 1st, 2025. It's annoying. But their last quarter, they said, oh, we're here, 37 billion run rate for AI. And in this quarter, they didn't. Right. They just didn't mention it. And I think what it is, is the, I don't know, Ben from the Ben and Emile show once said this to me, it's like the way to understand the stock market is imagine the thing that would piss you off the most happening. So, there you go.

12:18Because it's so frustrating seeing everyone saying the AI bet paid off, the AI bet paid off, look, number go up so big. Even though they gave us less information this time, they are obfuscating the information.

Run rate metrics examined

12:34Yes, exactly.

Run rate metrics examined

12:35If AI were going as well as a lot of people say it's going, then you would expect these companies to simply report their AI revenues and brag about it. Yeah. But instead, they don't do that. Instead, what they do is they report their compute revenue, which to be fair is growing, but that was a business that existed long before AI. That's not their AI revenue. And when they do report their AI revenues, they usually report it in ARR, their annual revenue run rate. And Amazon actually did that in this quarter and said it had a $25 billion run rate, which

13:09is $2.08 billion. But here's the other thing that people seem to miss as well. You know what also happened in this quarter? Anthropics compute spend accelerated because its revenue accelerated. Like, it's a very obvious, like, between, so this quarter would refer to what is, it's June, May, April, right? Mm-hmm. Yeah, that is, that's when Anthropic, I think, claim they hit $30 billion at annualized run rate. Like, they are spending so much more in compute than their actual revenue. I wouldn't be surprised, and this will get into, if most of this run rate came from Anthropic

13:43and OpenAI spend. Of course. Again, jingle, jingle, the keys are here. Look at the run rate. Look at it. Exactly. And I think just for, just for your audience to understand why run rate is a bullshit metric, I mean, this is sort of like, this is the classic metric that is reported by startups where their, their revenue is very choppy. And so what they'll do is they'll pick their best month, and then they'll just times it by 12 and say, this is how much we are expected to make over the course of the year. I'll just like point out with my business, you know, my podcast generated double the revenue

14:18in June that it generated in May. We had a really good June. Now I could just annualize that number. I could just times it by 12 and say, oh, that's how much money we're going to make this year. But of course, that would be BS, and I'm not going to do that to you. I'm not going to lie to you. But this is what they are resorting to instead of just reporting how much money they actually made off of AI, instead what they do is they pick a random month and they say, oh, well, if we times it by 12, then that will be a good number. And that's what they are starting to report. It's very, um, it's very BS-y.

14:51Yeah, it is. And that's the one that Microsoft declined to report this month, which means likely that, so I, the thing is, I, I'm guessing that they're referring to a month within the quarter, but we actually don't know. We don't know if it's 30 days, 28 days. We don't know if it's four weeks times 50, but I, we truly don't know what'd be like four weeks times 13. It's just, they never disclose this, but I think it's just a failure of the media and analysts on this one, because I feel like if these companies felt any pressure to actually

15:22be honest with their investors, they, I don't know, would do anything other than this. Yes. Well, I think it's actually interesting.

Meta capital expenditure questions

15:30Meta is a, an interesting example of this meta is one of the few companies where the analysts are actually trying to ask management, the tough questions, because the, it is so obvious what the problem is. The, the problem is that their, their, their free cashflow is declining. It's down 91%. And the reason that it's declining is because they're spending just gargantuan amounts of money on this AI infrastructure. Now for the other companies who are doing that, they already have these cloud businesses

16:00like Google cloud, like AWS, like Microsoft Azure. And so wall street kind of understands generally how they are expected to generate a return on building all of those data centers. Wall street can stomach that with meta. It hasn't been outlined whatsoever. You and I talked about this. They later kind of folded and admitted that they were just going to build their own cloud business. That was what we thought was going to happen. But when the analysts asked Mark Zuckerberg point blank, what the actual plan was in terms

16:33of ROI, how are you actually going to generate a return on all of this AI data center spending? He didn't give them an answer. He didn't discuss the cloud plans. He didn't address any of the reports about how meta was an early talk to sell compute to Anthropic, which I know you have covered at length. He basically just avoided the question. And I think that tells you something about Mark Zuckerberg and the extent to which he gives a shit. He doesn't really. He doesn't really care. He can't be fired. He literally cannot be fired.

17:03That's right. Which is in and of itself insane. And that goes back to the governance. Exactly. I mean, it feels like we're in the golden age of grift. Like it feels like we're just in an era where the only reason this is happening is because they know they can get away with it. There's no, it's not like I've read multiple, I would say like five different articles on meta, Google, Microsoft, Amazon, their earnings. And the only reason anyone is slightly critical of meta is because their revenue growth wasn't as good as it was expected to be.

17:34That's the only reason. And it's not, it's not like analysts are point blank saying, Hey, you haven't told us what your AI strategy is. You may have wasted that. They're not being blunt with them. They're just like, Hey, Hey, Mark, when you get around to it, would you mind telling me why you spend like, like 200, $300 billion and you want to spend $145 billion this year? When you, when you got a minute, don't worry. I know you're busy. I actually disagree with you on that.

18:05I think meta is pushing them. I think, I think that this has been recognized by the market by wall street took them a while. I'll, I'll give you that. But I think that in the case of meta, they actually are starting to ask those questions. And I think the problem is that the other companies, because they have those cloud businesses, they're kind of refusing to ask the tough question or not even the tough question. The obvious question of to what extent are you dependent on open AI and anthropic to companies, which are losing gargantuan amounts of money, two of the most unprofitable companies in the history of companies.

18:40And that's literally a fact. You can look up the list of the most unprofitable companies ever. And you'll notice if you do the math, open AI and anthropic should be in that list. So I think that they are not doing that homework, that hard work of digging into the numbers and asking that difficult question when it comes to, especially when it comes to Google and especially when it comes to Amazon. But I do think that they are worried about meta's total lack of interest in laying out how they plan to generate any returns on this whatsoever.

19:20All right, folks, I'm here talking to you again about Quince, one of my favorite clothing brands. We've got a little under two months worth of summer left. So it's worth looking at some of the new lightweight down packable puffers for the fall. I personally am going to be picking up one of their new non-iron cotton Oxford dress shirts, as apparently that's what you wear at the stock exchange, not a t-shirt. And Quince is just a great place to shop for clothing. They've got premium denim and tailored chinos for just $60 with the kind of perfect fit, structure and high quality fabrics you'd expect from a luxury designer brand.

19:51Everything at Quince is priced 50 to 80% less than similar brands. They work directly with ethical factories and cut out the middlemen. So you're always paying for high quality, not brand markup. And I absolutely love their stuff. Upgrade your every day. Download the Quince app for app-exclusive offers or go to quince.com slash better. Get free shipping on your order and 365-day returns. Now available in Canada and the UK too. That's Q-U-I-N-C-E dot com slash better.

20:22Amazon Health AI presents Painful Thoughts. I, um, I can't stop scratching my downtown. Mm-hmm. Yeah, but I'm not itching to go downtown and tell a receptionist I'm here to talk about my downtown. Some things you'd rather type than say out loud. There's no question too embarrassing for Amazon Health AI. Chat your symptoms and get virtual care 24-7. Healthcare just got less painful.

20:52Every small business owner has that one story. The one moment that could have broken them, but remarkably, it didn't. Hi, I'm Kathleen Griffith, co-host of The Unshakeables, along with Stevie Barron, CEO of Chase for Business. We have more incredible stories of overcoming the impossible still to come in Season 3. You'll hear from entrepreneurs who faced make-or-break moments and came out stronger. We dig into how they did it, what they learned, what they'd never do again, and what kept them pushing forward.

21:29These are the stories that remind us that being unshakable isn't about never falling. It's about how you rise, rebuild, and keep going. Whether you're starting up, scaling up, or trying to find your footing again, these stories remind us that resilience lives everywhere. In small businesses shaping communities all across America. Listen to The Unshakeables wherever you get your podcasts and learn more at chase.com slash podcast. J.P. Morgan Chase Bank, N.A., member FDIC, copyright 2026, J.P. Morgan Chase & Company.

22:04Hi, I'm Steph, and I work at UnitedHealthcare. When I think about UnitedHealthcare's members, I think of my own family. My niece was diagnosed with an ultra-rare genetic disorder. I know how hard it is for families at UnitedHealthcare. We can make it easier. I'm Brian. I care because I know what it's like to not speak the language. It's not easy. Health insurance is complicated. I would say our job is to make everything easy for our UXC members. I work for UnitedHealthcare. I'm committed to care.

22:37Well, that's the thing, though. I will push back again on Meta. Sure, they are asking the questions. Yeah. But they're not point-blank saying, hey, you've not shown it. They're not. And even the media. The media is absolutely not. Like, they're saying, oh, there's concerns over AI spend being higher than usual. That's strange. It's this kind of, like, plodding, oh, I guess maybe we should find something out. And the analysts are pushing these questions now.

23:07But as we speak, and in two days this will come out, so he knows it will be down again. But Meta's already recovered 6% today. It recovered a couple percent. It's already on course to recover all of the losses. Yeah. So it's not like anything changed. But as we discussed before we got on here, there are some analysts who have tried to work out or have estimated the actual concentration of revenue for Amazon, Google, and Microsoft for open AI and anthropic spend.

Revenue concentration risks

23:38And as we were just discussing, so Ross Sandler over at Barclays, he estimates that 73% of all Amazon's AI revenues in both 2026 and 2027 will be anthropic and open AI's compute spend. Crazy. Which is completely insane. Crazy number. Insane. It is insane. So I found this a couple days ago. I've been working on it. And I thought, well, it can't be that bad elsewhere. Except I looked at UBS's Stephen Jew. He said this back in, so Barclays is back in March.

24:09Stephens was in June. And he said that 40, sorry, it's 27% of all Google Cloud revenues. So not AI revenues. All Google Cloud revenues for 2026 are estimated to be open AI and anthropic. And then 44% in 2027. That is completely fucking insane. Like that is just, this is, this is to me such a huge scandal. And I mean, I could also mention that, well, that there's also a Wells Fargo report that says that Microsoft's same deal.

24:3973% is anthropic and open AI of their AI revenues. Michael Turin over there. And it's like, I feel like I'm going a little insane because when you read these numbers, you're like, oh, these are from major investment analysts. How are they not calling like, hey, Mr. Fargo, Mr. Fargo, I read these numbers and you should really check these out because we might be in trouble. And I think, and friend of the showcase, Kagawa, really helped me nail this down. I think it's, these people are only capable of thinking two quarters in the future.

25:11Like they can, they can think of 2028, they can think of 2027, but they can only do so in terms of the last two quarters. Number has gone up. Open AI and anthropic are still solvent. So number will keep going up. It isn't a problem that the revenue is centralized because number keep going up. But to me, this is potentially one of the largest corporate scandals in tech history. And maybe I'm being dramatic, but this is, we're talking over $990 billion of CapEx now.

25:44And most of it is for these two companies that can quite literally not afford to pay unless one of those three companies gives them the money. Yes. This is a huge, huge problem. And you were saying that the analysts weren't, and the media weren't doing their job in terms of asking meta questions about how they're going to generate a return. I sort of pushed back on you. On this one, the silence is deafening from the analysts and from the media. There is a very, very obvious concern when it comes to the businesses of the largest and most systemic companies in the world.

26:21The companies on which the portfolios of every American are entirely predicated on, on which the entire stock market relies on. And that is how much of your business depends on two companies, OpenAI and Anthropic. And you are one of the few people who has been doing significant work into uncovering that answer. And you have looked into some of the research that has come out of Wells Fargo, Barclays, UBS, and we should give those analysts credit for crunching those numbers and doing the hard work of figuring it out.

26:54But the numbers are insane. OpenAI and Anthropic, 74% of Microsoft's AI revenue. That is very, very unsustainable. We talk about diversification in markets, the importance of having a diversified portfolio. The same thing goes for business. You want to have a diversified revenue stream. That indicates strong, healthy, sustainable growth. None of these big tech companies have anything close to diversified revenue streams when it comes to AI.

27:27And so you have to ask the very obvious question, what would happen if one of those two companies, OpenAI and Anthropic, went under? And you also have to ask the question, what is the likelihood that one of those two companies, OpenAI and Anthropic, could go under? What is the likelihood that they would not be able to make good on those payments that are propping up the growth of these big tech companies, specifically when it comes to their revenue? And the answer is, it's actually pretty fucking high. Yeah. Because you look at the financials, and they are ugly, as you have talked about at length.

28:01And we can get into, but I want to pass it back to you. No, the thing is, a fun story. A couple of days ago, Amazon, they were meant to give OpenAI another $35 billion if they went public or reached AGI. Well, none of those things happened, of course, but they still gave them the $35 billion a couple of days ago. And that's crazy. That's a really crazy number, especially because that is, well, let's just go and have a look at these numbers from Ross Sandler at Barclays. That is, so $35 billion is roughly, well, let's see, yeah, it's a little bit, it's around how much AI revenue Amazon has made this year.

28:43So $40.1, $9, $8.5, actually, yeah, yeah, it's $3.4 billion less, sorry, more than Amazon is estimated to make on AI this year, in totality, including OpenAI and Anthropic spend. So that's good, just handing themselves money, just handing it to OpenAI and Anthropic so they can hand it back.

Circular financing in big tech

29:07Exactly.

Circular financing in big tech

29:08They already gave $5 billion to Anthropic. They already are on course to give them another $20 billion sometimes this year. Google already gave $10 billion to Anthropic this year, has another $30 billion they owe. So this is not a real, this is not a real economy. It's not real growth. No one's actually doing business here. It's just people handing money to each other and being like, yep, look how profitable we are. Even though that does not even, doesn't seem like it's exploding margins either. It doesn't seem like it's having much of an effect at all.

29:40But what's really concerning is everyone's excited about these companies because of their cloud growth. And it's really beginning to look like most of that revenue growth is coming from this spend. Yes. Which is terrifying. It's almost all of it is coming from those two companies. Earlier you said that them not reporting that their revenue is coming from OpenAI and Anthropic alone, that that is something of a scandal. I think that is just irresponsible.

30:10I think of it less of a scandal. The part that I do think is a scandal is the circular financing aspect of it. It's the fact that those two companies are also the two companies in which big tech invested billions and billions of dollars in order to keep them afloat. And that is the only reason why those companies are surviving. Yes, it's Silicon Valley and venture investors. But a lot of that investment is corporate investment coming from big tech. It's coming from Microsoft. It's coming from Google. It's coming from Amazon. And so the idea that the largest customers, the customers that are driving the majority of these companies' revenue growth are also the companies in which those companies had invested in, that is a very, very obvious conflict of interest, at which point it starts to get into actual scandal territory.

31:00I don't think that calling the circular financing epidemic, I don't think calling that a scandal is overkill. I don't think that's a stretch too far. I think it really is a scandal. And I think it's actually shameful the way Jensen Huang has talked about it, where he was asked this question about the circular dealmaking. And he said, I don't think it's circular at all. He simply denied it. And he didn't come up with an explanation as to why it wasn't circular. He didn't come up with any real excuses. He just said, no, that's wrong. BS. I don't like what you're saying.

31:30And then he just moved on. And that is a real problem. That's when you start to see that actually these executives either they're just blindly ignoring what is happening, or they are actually lying. I mean, I think that it's reasonable to assume that Jensen Huang told a lie when he said that he didn't think that what they were doing was circular. Just to go back to the question of, would open AI go under? Because that's how this goes down.

32:01Yes. And that is also a point that people are now readily discussing. It's no longer fringe economic theory from a crazy person called Ed. It's now something that people want to discuss and are worried about. Yes. I think. But it's like. Yeah. But they discuss it in this way that is like, I don't know, like if a neighborhood restaurant shut down. Like, it's like, oh, you know, Jeff, you know, these things happen.

32:33Business is tough. I wish I was kidding, but I've tried to talk to reporters about this and tried to be like, hey, look, this company is really holding up the earth. Yeah. Like they are quite consequential to the future revenues of multiple companies. And they're like, yeah, but you know, there's other demand. Yeah. And that's the thing when the big thing that this, that the piece I'm working on will be out by the time this goes out says is like, yeah, the other problem with 73% or so of AI revenues for Google, Amazon, and Microsoft being open AI and Anthropic is, and just be clear, this progresses over the years of estimates.

33:13It's not like they say, okay, demand will even out. Now, that suggests that there isn't demand for AI, just writ large, that there is not demand, at least not at the scale that's been promised, and definitely not at the scale that will satiate all of these data centers. It's not possible when you look at these numbers, and this is not me just being like, oh, I worked some stuff out with some articles. These are actual financial analysts. Yeah.

33:44Yeah, these are, yeah. And it's, it's genuinely terrifying, because it's, it's not just how bad it will be, and yes, we'll get back to open AI down in a second. It's the fact that everyone's kind of just looking at it and be like, ah, it's all right. Well, I think, yeah, I think it was really, the actual philosophy was, was revealed in your Bloomberg interview. Um, and, you know, I don't mean this disrespectfully to the interview. I thought it was actually a great interview, and I thought they did a really good job. But, um, I forget her name, but the interviewer said something, said something quite important to you, where she said, I just want to make sure you understand.

34:22I want to make sure we're all on the same page page here. The people who run these companies are very smart people. Yeah. And that was the beginning of the question, which would go on to say something like, are you really sure about all of this? And I think that is the hesitancy that you're seeing when you talk to media outlets, when you go to the New York Times. And I think that's the hesitancy among all of us, which is that to say that this company might not actually work out, which is not an insane thing to say by any stretch of the imagination when you look at their financials and when you look at their numbers.

34:54But to say that it might not work out is to say that the people who are running that company and the people who are investing in that company aren't the smartest people in the world. And today, that is a bold thing to say. It is a bold thing to say that Satya Nadella might not know what he's doing. It's a bold thing to say that Sam Altman might not know what he's doing. It's a bold thing to say that Mark Zuckerberg might not know what he's doing, which is quite crazy to me, considering that just a few years ago, he spent more than $80 billion trying to build the metaverse, which was a total failure, total dead on arrival.

35:32He renamed his company to Meta, because he thought that this was the future. He was so, so wrong about that. So the idea that these people can't be wrong, or that they can't spend money incorrectly, or that sometimes they might actually not be great capital allocators, that is total lunacy. It is a total fantasy. But for whatever reason, people don't want to believe that. People get uncomfortable when you start to suggest that. And that, I think, is why people get uncomfortable when you say the things that you say, because it suggests maybe these guys don't know what they're doing.

36:06Well, it's that. And also, let's just say this right now. I'm looking at the Barclays numbers in front of me. OpenAI dies tomorrow. That is, let's look at it, 28% of Amazon's AI revenue next year. Gone. Goodbye. Just evaporated. And indeed, 27% for the next two years as well. If Anthropic and OpenAI, if one of them dies, well, that's like 20%. Let's see. OpenAI is, well, it looks like about $12.5 billion of Google's 2027 AI revenue.

36:41Just like gone, evaporated. And otherwise, their other AI revenues are pretty flat. It looks like Vertex. Vertex kind of increases, but again, that's through selling Anthropic and OpenAI's models. It's, I don't think people realize that there's a great chart that's coming in this article as well. How much of Azure's revenue as well is just OpenAI's spend? And it's, let's see. Let's take a look at this chart of Azure revenue as well. 62%, let's see. Yeah, it's about 72%, 68% to 72% in fiscal year 2026, and as much as 74% in 2027.

37:17If these companies die, they lose most of their AI revenue, evaporated. And that's not including the amount of money that comes through Azure AI renting out their models. Yeah. Like, and people are just like, yeah, you know, if that happens, there'll be more demand. I have this guy who's a very well-meaning guy. He's on Twitter. He's a, he's a semiconductible analyst. He always responds and say, look, all useful compute will be bought. All useful compute will be bought. It will always be bought. And I'm like, yeah, but if OpenAI dies, it's the largest consumer of AI compute, what happens to that?

37:48Well, it'll get bought. Yeah. And it's just insane to me. It is genuinely bonkers to me that no one, like, I guess, I guess what it is taking a step back is looking at what you said as well, is it's, they don't want to follow the argument to the logical endpoint because it's all of these smart people were stupid. That's right. And also everyone was wrong, wrong in a way that is going to be very difficult to back out of. That's right. And also they will lie to just, let's be clear.

38:22Yeah. That's, that's the problem. Now. So I'm in total agreement with you on all this.

Valuation multiples and anxiety

38:28I want to take things to their logical conclusion further. And I'm going to say some, but I don't know if you're going to be that excited about what I'm going to say. That's fine because my, when I think about this, what would actually happen if AI were to basically just die? If open AI actually died and if anthropic actually died, and I think there's good reason to believe that, which we can get into. We can get into their financials and just rehash them because they are very important. They're systemic to the ecosystem. But I sometimes like to think, okay, what would a world look like where the AI thing just didn't work out?

39:02And I have to say, a lot of that anxiety actually is priced in right now when it comes to big tech. And I just want to point you to the price to earnings multiples. Yes. The price to earnings multiples of big tech at the beginning of 2023, I went back and looked at this. This was right before the AI trade took off. If you look at the price to earnings multiples, you look at Microsoft, it was trading at 26 times earnings. Amazon was trading at 80 times earnings.

39:33Meta was trading at 31 times earnings. If you look at the PE multiples today, and by the way, I looked at this as of last week, at the end of last week, right after they reported their earnings. So they could have changed by now. I'm sure they have. Microsoft's multiple is lower than it was. It's at 25. Not that much lower, but it is lower. Amazon's is significantly lower. And this is the real PE, the one that I adjusted, not factoring in their investment gains on their anthropic and opening high stakes. It's trading at 30 times earnings.

40:05That's lower. And Meta's is significantly lower. It's trading at 20 times earnings compared to 31 before the AI boom. So I think one thing that we should recognize is that in some places, in some areas of the market, I think there are people on Wall Street who are actually quite anxious. I think there are a lot of people on Wall Street, Ed, who are reading your newsletter. I truly believe that. And I think that that perspective, the bearish perspective, and the acknowledgement of how this could all implode, I don't think that it is completely driving the market.

40:41I think there's more optimism than there is pessimism. But I do think that it is somewhat being reflected. And that is to say, if this all were to blow up, I don't think these stocks wouldn't crash. They would certainly crash. But I do think that it wouldn't be as horrible as it might have been, say, in 2008, when the housing bubble was truly elusive to almost everyone. That is my conjecture, and I'm sure you might disagree with me on it. But I do think we should recognize the extent to which there is some anxiety that is beginning to be reflected in some asset prices.

41:29Amazon Health AI presents Painful Thoughts. Why did I search the internet for answers to my cold sore problem? Now I'm stuck down a rabbit hole filled with images of alarmingly graphic sores in various stages of ooze. I can clear my search history, but I can never unsee that. Don't go down the rabbit hole. Amazon Health AI gets you the right care, fast. Healthcare just got less painful. Every small business owner has that one story, the one moment that could have broken them, but remarkably, it didn't.

42:11Hi, I'm Kathleen Griffith, co-host of The Unshakeables, along with Stevie Barron, CEO of Chase for Business. We have more incredible stories of overcoming the impossible still to come in season three. You'll hear from entrepreneurs who faced make-or-break moments and came out stronger. We dig into how they did it, what they learned, what they'd never do again, and what kept them pushing forward. These are the stories that remind us that being unshakable isn't about never falling. It's about how you rise, rebuild, and keep going.

42:43Whether you're starting up, scaling up, or trying to find your footing again, these stories remind us that resilience lives everywhere, in small businesses shaping communities all across America. Listen to The Unshakeables, wherever you get your podcasts, and learn more at chase.com slash podcast. JPMorgan Chase Bank, N.A., member FDIC, copyright 2026, JPMorgan Chase & Company. My name is Jessa. One of the things I love the most about working for UnitedHealthcare is that everybody matters.

43:18Every moment matters. There's a person behind every problem. I care because it's what I was put on this earth to do. I'm Ben, and I work at UnitedHealthcare. I am just one piece of a larger puzzle, but every piece matters. It's more than just work. We want to make the healthcare system better for everyone. I care because I want to make a difference. That's what committed to care means to me.

43:42Hey there, I'm Kat. I'm a college athlete. I was diagnosed with ankylosing spondylitis, which caused back pain and stiffness every day. With Cosentix, I'm able to stay active. Cosentix Secukinumab is prescribed for adults with active psoriatic arthritis, ankylosing spondylitis, or non-radiographic axial spondyloarthritis. Available in both IV infusion and self-injection. Don't use if allergic to Cosentix. Get checked for TB before starting. Increased risk of infections and lowered ability to fight them may occur.

44:13Like TB or other serious bacterial, fungal, or viral infections. Some were fatal. Tell your doctor if you have an infection or symptoms like fevers, sweats, chills, muscle aches, or cough. Had a vaccine or planned to. Or if IBD symptoms develop or worsen. Serious allergic reactions and severe eczema-like skin reactions may occur. Learn more at 1-844-COSENTIX or cosentix.com. Move and feel better. Ask a rheumatologist about Cosentix.

44:44So, one pushback, and I think it's a critical one, is that even though the price through earnings and all that reflects a little bit lower and it's showing some anxiety, it still is a price which factors in the idea that these companies will keep growing. Yes.

Cloud growth without OpenAI

45:01And I think that that is a critical problem because I'm not sure, at least in the case of their cloud businesses, whether these companies are growing anymore when you remove OpenAI and Anthropic. Right. And I don't think it, like, I'm not just talking, like, I don't think Microsoft, Google, Amazon, and Meta, though I think Meta could die in 10 years. I've thought that for a while. I think that they're just, it's falling apart. But I don't think they're dying. I think their stocks are going to get their arses beat. But the problem is going to be not just that AI didn't work.

45:33That's what's going to be the first problem. The second one is when they go, right, well, OpenAI isn't paying you anymore, and suddenly your revenues are down. Yes. And your growth has slowed. Hey, Google Cloud has stopped growing. It's stopped growing pretty much at all. It's, in fact, looks kind of flat. What's the deal with that? And it's because Anthropic and OpenAI's compute spend was propping up the growth. So while Wall Street might be saying, oh, well, if AI doesn't work out, I think what they mean is that they think AI won't work out and be the next big thing, but there will still be big revenue-generating businesses from this, which is not the case.

46:17I went and looked at the Wells Fargo estimates from Turin, and what's interesting for Microsoft is that in fiscal year 2027, so next year, they estimate the Microsoft 365 co-pilot will make $10 billion a year in annual revenue, which is pathetic. Yeah. That's like a, what, like a third of intelligent cloud? It's, I mean, it's a little bit less than they estimate the Azure AI will make, which is predominantly made up of selling OpenAI's models.

46:50I mean, they said one of their bragging points on that earnings call was that they have 30 million co-pilot, paying co-pilot users. Oh, yeah. But what they don't really talk about is the fact that that's part of a larger enterprise bundle, which only makes up about 8% of their total Microsoft 365 subscribers. So basically what they're telling us, they're trying to brag, but ultimately I read that and I go, oh, so you only converted 8% of your user base. That was the share of your users who were down to try this AI product that you say is going to change the world.

47:26So to that point, I don't think that the strength that they are projecting in their consumer AI business is anywhere near what they'd like us to believe. Well, on top of that, the Wells Fargo estimates, estimates, estimates, someone will get me on that one in the comments. They only estimate that they made 3.85 billion for the entire fiscal year 2026 from 365 co-pilot. You know, the thing with 30 million paid seats. They're discounting this to Timbuktu.

47:56Like this is, this is being, they're discounting it to hell. And I mean, pretty much anything, like, like pretty much anything that Microsoft sells has discounts, especially when it comes to Microsoft 365. But again, the hogs lapped it up. They love their slop. They love being given a new big number so that they can interpret it. When they, again, not really sure what these analysts were doing, because I don't know, again, I think I've said this already. Like, if I was working at Wells Fargo and I had these numbers, I would be calling my boss's boss.

48:29I'd be like, hey guys, hi, we have a problem. Because like I said earlier, it's not just that Microsoft, Google, and Amazon have pathetic AI revenues outside of OpenAI and Anthropic. It's that if they have pathetic revenues, if they, as the largest holders of AI compute, as with the ones with the best pricing power, with the best teams, ostensibly, with the most experience building AI data centers and providing AI compute, if they can barely scrape together what?

49:03Like, let's see, so it's like 73, 74% of their revenues. So like they have this meager little shrivel of their AI revenues is from diverse revenue. If that's the case, then there is not demand for AI compute. It's not that there's only a little. It means there's none. It means there's actually, like, there is a couple billion. Because again, when you look at the other providers, CoreWeave, who are their biggest customers? Anthropic, Meta, and OpenAI. Who's Iron's biggest customer? Microsoft, Microsoft, Nebius, Microsoft.

49:35What's Microsoft doing with AI? Not very much. Not actually very much at all. They are providing compute to OpenAI and conning people into buying Copilot. Yeah. Like, it's just, it's astonishing because this is bigger than them. It's bigger than them. It's the 190 gigawatts of planned data center capacity. It's all of these data centers being built everywhere. I, like, I'm stuttering because it's like, the ramifications of this are everything I've kind of feared. Which is that AI demand does not exist at scale.

50:09And most of these data centers don't go paid for. And that's even if Anthropic and OpenAI, like, succeed in surviving past 2027. Well, I think you said something important earlier about the growth. And that is, if this AI thing doesn't work out, it's not as if Microsoft and Amazon and Google and Meta won't have a business anymore. They had a business before AI, and the business was pretty big and doing pretty well.

50:40What it does mean is that these companies won't actually be growing anymore. They won't be growth companies. And I think this gets to the heart of what AI has been all about. And that is that the big tech companies in late 2022, when ChatGPT was released to the world, they suddenly realized, oh, my gosh, innovation is happening without us. There are other companies on the scene. There are other technologies, and we don't have a part in it.

51:11And so they decided, okay, what are we going to do? We're going to do whatever we can to take whatever semblance of technology exists in that company and then bolt it onto our own companies. We're basically going to inject that optimism into our own faces as if it were Botox to convince the market that we are, in fact, young again, that we are growing, that we're hot, that we're innovative, and that you should come and put your money in us. Because we're not just a company that's going to continue to generate cash. We're a company.

51:41We're a generational company. We're a company that's going to significantly increase profits at an exponential rate over 10, 20, 30 years.

Mature companies seeking growth

51:50That is the story that they wanted to tell the markets. And they wanted to tell it so desperately that they were willing to actually go and invest in those companies like OpenAI and Anthropic and also SpaceX, which we haven't mentioned that much, but it actually plays a role in all this. And then have those companies come back and then buy our chips, give us back the money, and then we're going to use that and use that as a story to tell the world that actually we're young again, that we are growing. And convince everyone by proxy to spend money on AI compute to compete with us with the companies that we invest in.

52:28We will obliterate our balance sheets in the process. And that's where things are going to get really ugly. The companies that really start to lose cash, and we're starting to see it. Amazon, negative free cash flow. Google, negative free cash flow as of this quarter. But that's really what they're doing here. They want to be young again. But if you look at these companies, these companies are old. These are mature companies. Their heyday is behind them. But they don't want to believe that.

52:59They want to live forever. They are the Brian Johnsons of the world of technology. And so I think what's going to happen if this thing doesn't work out, and I think there's a high likelihood that it won't because of the financials of OpenAI and Anthropic and how dependent the whole world has become on them. I think what will happen is the companies won't go under, the big tech companies. I think what will happen is Wall Street will look at them and they'll say, you're not a growth company. You're a mature company. And that means that we're going to assign a mature multiple to your earnings.

53:34We're not going to have the days of... I think that'll look like something in the range of 15 to 20 times earnings. So that's why, by the way, Meta I'm not so worried about because Meta is actually in that range. I think if you look at the company, I think Wall Street is looking at it and saying, this is a company whose growth prospects are unlikely at best. So I'm not so worried about Meta. The companies I'm more worried about is a company like Google trading above 30 times earnings. By the way, a company that's gotten way out over its skis from a multiple perspective, which is really surprising to me, is Apple.

54:07And I think that's because people have actually gotten nervous about the AI trade. And so they said, okay, Amazon still has cash on the balance sheet. Apple's trading at nearly 40 times earnings. The way that they have been able to dupe the market into thinking that they are a growth company when they haven't come out with a new product in 10 years, and the ones that they've tried to come out with and tried to sell have completely flopped. They canceled the Apple car. The Vision Pro was a total disaster. I mean, that's very clearly not a growth company.

54:37But I think that's the question. The funny thing is with Apple as well is that they've managed to keep going by just not blinking on the AI trade, except the AI trade fucked them over because of the increasing cost of RAM. Right. Because of everyone buying into... I mean, that's the other thing. It's like the increasing... This is the other... So actually, this is a good bridge from what you were just saying. The other problem with this as well is, yeah, you're saying these companies will be assigned growth, like the multiples associated with old, boring companies, right?

55:09Boring, old, mature. But again, I don't think that factors in, A, the fact they're not growing, but B, the fact that they have hundreds of billions of assets now. They're just burdened with debt, endless debt. Microsoft has been kind of smart in the amount of debt it's taken on. It's nowhere near as encumbered, but who knows at this point? Yes. Google's done a massive equity sale. They plan to do a massive equity sale. The more on tilt they get in order to pursue this Botox future of being young again,

55:41the more dangerous... You've got to write that up. The more dangerous their... The larger the threat to their entire existence becomes. So I will agree with you. If they continue down this road of, I'm not just going to pay for the Botox with the cash that I already have in my bank account, I'm going to take on significant amounts of debt, significant amounts of leverage and margin in order to pursue that dream, because I'm so obsessed with pursuing it, and I need to do it because I have to be young again. If they take that to its end, to its logical conclusion, as you say,

56:13then, yeah, we can start talking about, will these companies implode? And I think Oracle is an example of a company where they have done that. They have genuinely done that, and its prospects are genuinely terrifying. We're starting to see it with some of these other companies. I was shocked that Google was willing to come out with a quarter of negative free cash flows. I was shocked that Amazon was willing to do that. So it does appear that they're starting to lose their marbles on this front. They're so obsessed with the mission.

56:44They're so obsessed with the AI future that they are actually acting irresponsibly. But I don't think that we're quite there yet. I think the signs show that they're going to continue to do this. I mean, if they continue to load up their balance sheets with debt, who knows what the future will actually look like. But so far, they're not quite in a place where they're fully going to collapse and implode, would be my read of the situation at present. I fully agree, by the way. I don't actually think the big three collapse.

57:16I think Meta is already on the down spiral. I think their product is collapsing. And the things that I continually hear from that company suggest that the culture is screwed. But nevertheless, for the short term, at least, the next five years, I think that they will be fine. They're not going to have a fun time of it when this ends. Again, when it comes to logical endpoints and the way that people don't want to discuss them, Oracle is now spending all of its revenue, just more than its revenue on AI CapEx just for OpenAI.

57:50And I mean, the New York Times just fucking ripped me off, by the way. Like, there is a sentence in there about fallen angels that is like bordering on identical to my piece. Fucking New York Times, man. If you're listening from the Times, go read the Oracle piece. Go read my fallen angel bit from the OpenAI bubble. You tell me if I'm crazy. But nevertheless, people are having this conversation like, wow, Oracle, they're kind of unprofitable. That's not good. Wow, that's not good. But then they stop short of saying the thing that they don't want to say, much like everyone else. And I want to talk about this bit, which is, yeah, if Oracle doesn't work out the AI trade, it dies.

Oracle and credit default swaps

58:26Yes. Its revenue has been flat for 15 years, adjusted for inflation. It has, the crazy thing is, I wrote about this a couple weeks ago, the crazy thing about Oracle is that they actually would be on the decline already. But whenever they slow down, they buy more companies. They just acquire companies. They're like, they're really strange. Yeah. But nevertheless, no one wants to talk about the fact that Oracle will die. And honestly, seeing these numbers, yeah, Oracle will die. It is a genuinely, it is a genuine possibility.

58:58It's not overhype, it's not doomerism, it could genuinely happen. I think we should give credit to one group of people who actually do seem to be pricing in that possibility, and that is the credit default swap traders. Oh, yeah, the CDS people are. Whichever ones have driven those CDSs up to where they are, I don't know where they are exactly right now. And that's just, for the listeners, that's just betting on whether Oracle will die. That's betting on it dying. Specifically betting on whether it will default on its debts. Yes, whether it's the insurance that you pay on the possibility that it will default.

59:33And the price of that insurance is going up. People are beginning to realize in that market specifically that the possibility is real, that it could actually happen. And I mean, I think there's a question as to, again, whether they're too late on that. Weren't the signs there before? Couldn't we have started to see this coming, not today, but 6 to 12 to 18 months ago? And I think that would be the contention of your readers and your listeners who have been hearing about this from you for a very long time.

1:00:06And then suddenly everyone decides, oh, maybe it's a problem. Maybe we'll start to write some articles on this. Maybe we'll start to see those spreads widening on the CDS swaps. But we should give them their props. You know, some investors seem to be acknowledging the situation. Yeah.

Closing thoughts and newsletters

1:00:25Well, you know what, Ed? I think that's actually a good spot to end it. People can find you on Prof G Markets. What are you working on at the moment? Because you should write up that Botox thing. That's an incredible idea. It's a banger. Please run with it. Okay. What else are you working on? That'll be my next newsletter. Let's see. I'm working on a newsletter where we're talking about what happened in South Korea last week. That'll go out on Tuesday of this week. So that's tomorrow. I don't know when this episode will come out. But if you want to come follow my work, go to Prof G Markets.

1:00:55We're on YouTube. I cover the markets every single day of the week, or at least Monday through Friday. And we talk about everything. We have people on from all walks of life, from different sides of the market. We try to have bulls on. We try to have bears on. And we'll certainly be having Ed Zitron on in the future as well. Because we love him. And we love his analysis. We love the Eds. All right. You'll find me on the monologue this week as well. Haven't worked out what it's going to be. It'll probably be about this. Thank you as ever for listening. Subscribe to the newsletter. Premium as well. It's my main source, along with the podcast.

1:01:27I love you all. Thank you for listening. Thank you for listening to Better Offline. The editor and composer of the Better Offline theme song is Matt Ossowski. You can check out more of his music and audio projects at mattosowski.com. M-A-T-T-O-S-O-W-S-K-I.com. You can email me at ez at betteroffline.com or visit betteroffline.com to find more podcast

1:01:57links and, of course, my newsletter. I also really recommend you go to chat.wheresyoured.at to visit the Discord and go to r slash betteroffline to check out our Reddit. Thank you so much for listening. Better Offline is a production of Cool Zone Media. For more from Cool Zone Media, visit our website, coolzonemedia.com, or check us out on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

1:02:39This summer, Prime Video is the place for steamy romances, addictive love stories, and the book-to-screen favorites you've already read twice. The Love Hypothesis, L, The Devil's Mouth, Sterling Point, The Last Sunrise, and Off Campus. The Slow Burns, The Second Chances, chemistry you can actually feel through the screen. It's all here. Obsession is in session.

1:03:09Get ready to obsess on Prime. The Colonels cooked up a new $10 bucket of the day just for you. Monday, 10 wings for $10. Tuesday, 8-piece fried chicken for $10. Wednesday, 8 tenders for $10. Thursday, 10 wings for $10. Friday, 24 nuggets for... Ooh, you guessed it, didn't you? $10. The $10 bucket of the day deal. Every weekday, only at KFC, it's finger-lickin' good. Prices and participation vary.

1:03:40Well, supplies last. Not available on third-party ordering platforms. Tax extra. This is Jacob Goldstein from What's Your Problem? Running a business is hard enough. Don't make it harder with a dozen apps that don't talk to each other. One for sales, another for inventory, a separate one for accounting. That's software overload. Odoo is the all-in-one platform that replaces them all. CRM, accounting, inventory, e-commerce, HR. Fully integrated, easy to use, and built to grow with your business. Thousands have already made the switch.

1:04:11Why not you? Try Odoo for free at odoo.com. That's odoo.com. What it sounds like when you road trip to the iHeartRadio Music Festival in the all-new Hyundai Palisade Hybrid. Wow, this is really nice inside. The Palisade Hybrid has more luxury than ever. Whoa, this sound system is awesome. Turn it up. More tech than ever. Sweet. You got the available built-in dash cam? And more range than ever because the Palisade Hybrid gets up to 600-plus miles of range on a full tank.

1:04:42So do we ever have to stop for gas? Visit HyundaiUSA.com or call 562-314-4603 for more details.

1:04:52This is an iHeart Podcast. Guaranteed human.