How OpenAI Could Cause a US Recession | TCAF 264
Josh BrownMichael BatnickSteve Eisman
- Eisman agrees that an OpenAI failure could trigger a US recession, while saying it would not match the scale of 2008. This is hypothetical: Josh says, “I’m not making that prediction.” Eisman points to concentrated exposure, including five customers accounting for 70% of Nvidia’s accounts receivable and OpenAI representing roughly half of Oracle’s backlog. A host’s separate price-war scenario imagines Oracle going bankrupt.
- The hosts discuss how cheaper open-weight models could pressure AI labs’ prices. Michael says they cost one-tenth as much and compares choosing a model to choosing a car; Josh says enterprise habits may offer Claude some protection, but he would switch if a cheaper model performed almost as well. Eisman says the end of “tokenmaxxing” is starting to affect the story.
- A financing squeeze could strengthen Amazon, Alphabet and Microsoft while limiting independent infrastructure providers. Josh cites CoreWeave’s roughly 15% debt yield and argues corporate and consumer AI demand will persist. Eisman agrees hyperscalers are in a stronger position, while warning that the market depends heavily on two companies.
- The hosts frame AI as a common driver of stocks and Treasury yields. Josh links weaker AI sentiment to the 10-year yield’s fall from 5.4% to 5.1%; Michael agrees the move is about AI, not renewed confidence in government finances. Josh estimates that adding AI infrastructure beneficiaries brings the S&P exposure to about 65%.
- AI’s eventual success does not guarantee today’s leading labs will remain the winners. Eisman invokes first-generation internet companies whose foundations enabled later winners. His “conspiracy theory” is that Anthropic should list before third-quarter results; when asked whether its IPO would follow SpaceX’s path, he says he is unsure.
- Insurance funding and government ownership raise questions about who bears AI investment risk. Michael says policyholders might have received 3% interest where investors would have demanded 12% if they understood the exposure. Eisman says the answer to policyholder risk matters, opposes government equity because it could make companies seem “too big to fail,” and questions whether executives would face personal consequences.
- Housing remains a separate economic weakness that rate cuts alone may not resolve. Michael cites an estimate that total monthly ownership costs are about 50% above five years ago, including insurance, utilities and taxes. Eisman says prices may fall gradually; lower rates would help, but there is “still a long way to go.”
1. OpenAI’s spending commitments create a recession channel
Eisman distinguishes an industrial failure from a banking collapse. GM’s bankruptcy destroys jobs and suppliers; a JPMorgan failure threatens access to money itself. “If JPMorgan collapses, the Earth will burn up.” An AI collapse could cause substantial economic damage while leaving banks functioning.
Josh’s OpenAI scenario is explicitly conditional: “I’m not making that prediction,” but he asks whether its bankruptcy tomorrow could immediately cause a US recession. Eisman agrees. In a separate hypothetical, a host imagines a full-scale price war turning Anthropic’s sales growth negative, ending investor funding and forcing investment cuts. That host’s scenario concludes Oracle would go bankrupt because roughly half its backlog comes from OpenAI.
The connecting mechanism is customer concentration. Eisman cites five customers accounting for 70% of Nvidia’s accounts receivable, which he describes as a proxy for revenue, despite recent quarterly revenue growth above 100%. A host says OpenAI and Anthropic account for 70% of hyperscalers’ AI revenue, with AI representing 25–30% of total cloud revenue. Oracle is more exposed still: roughly half its backlog comes from OpenAI.
2. Cheaper models threaten the labs’ pricing power
Josh and Michael describe a lack of durable model moats: users can move from Claude to Gemini to GPT. Michael says open-weight models may be worse, but cost one-tenth as much: “you don’t need a Mercedes-Benz every time you go on a trip.” He suggests calls to regulate AI could help incumbents limit competition from open-weight models.
Josh says his company’s Claude Enterprise subscription has displaced his previous ChatGPT use because employees get accustomed to the tool already being paid for. He sees that as a possible enterprise moat, but says he would switch if an open-weight model performed almost as well at one-tenth the price. He has heard rumors that Claude prices are being lowered.
Michael describes “tokenmaxxing” as companies ordering developers to use AI around the clock, which became too expensive; he says Uber exhausted its AI budget in three months. Eisman says the end of tokenmaxxing is starting to affect the story. The hosts connect the move from per-seat SaaS to usage-based pricing with greater cost scrutiny: employees choosing cheaper models could pressure OpenAI and Anthropic’s prices.
Eisman questions OpenAI’s reported annualized revenue moving from $70 billion a month earlier to $50 billion at September-end, but says he does not think the figures are a valid comparison. For Anthropic, he cites $16.8 billion and 43% three-month growth versus 140% in Q2, qualifying, “I could be a bit wrong.” His measurement objection remains: “I have absolutely no idea what annualized sales mean.”
3. Financing stress could hand capacity growth to hyperscalers
CoreWeave is the financing warning rather than Eisman’s chosen systemic trigger. Josh cites a roughly 15% yield and a 60% price decline; Eisman says the company is spending $640 million on debt repayment, a serious issue given its cash flow. Josh says abandoning a half-built project is not viable; Eisman replies that it could be done, but the company would go bankrupt. He adds, “I don’t think CoreWeave is important enough to shake up the entire AI industry.”
Josh relays reports from visits to Wall Street banks that AI budgets and related hiring should be larger in 2027 than in 2025 or 2026. He argues that if CoreWeave and Nebius cannot borrow enough to build, Amazon, Alphabet and Microsoft inherit more opportunity. Eisman agrees that hyperscalers’ position has strengthened.
The hosts discuss how political opposition could slow data center construction. Josh says this would mean fewer computing resources purchased for projects that do not proceed; Eisman adds that demand for existing computing resources would increase. Josh argues that corporate and consumer AI demand will persist. Eisman allows that more companies could emerge to secure computing resources in three to four years. For now, he says, the market is so concentrated that “we can only hope these two companies do well.”
4. The hosts frame AI sentiment as a driver of equities and Treasuries
After high CPI and PPI figures, interest rates rose, but the cited 10-year Treasury yield fell from 5.4% to 5.1%. Josh attributes the reversal to weakening sentiment around AI; Michael agrees that the move reflects AI trading, not greater confidence in government finances. Josh connects weaker AI sentiment with less bond selling associated with the trade.
The hosts’ sharp illustration of the preceding market move is Western Digital: down 54%, yet still up 100% for the year. That offers little comfort to peak buyers. Josh also points to Lennar’s rebound and strength in other rate-sensitive stocks as signs of how quickly market leadership can change.
Eisman resists making valuation the central explanation. “If you tell me how the AI story will turn out, I’ll tell you how the stock market will turn out.” Josh says technology makes up 40% of the S&P, and that Google and Amazon bring the share to about 50%; adding investment banks and infrastructure beneficiaries such as GE Vernova, Quanta, Eaton and Rockwell brings it to about 65%. Josh concludes that “it’s all just one trade. It’s AI.”
5. Insurance funding complicates the question of accountability
The hosts discuss reporting about insurance money financing AI construction and ask what that could mean for policyholders. In a related discussion of Mark Walter’s insurance company and sports-team holdings, Michael says policyholders would not have received money at 3% if they had understood the exposure; the cost could instead have been 12%. The transcript does not fully clarify the mechanics of that comparison.
Asked about systemic danger to policyholders, Eisman answers, “I’m not worried. The important thing is finding the answer.” He separately rejects government equity investment in AI companies because it would create an expectation that the recipients were “too big to fail.”
Eisman supports holding an AI company responsible if it hacks someone or commits wrongdoing. His concern is whether executives who knew about risks would face personal consequences. Invoking the lack of executive punishment after 2008, he asks, “Who would sue me?” Josh, not Eisman, says he expects regulations to come only after a crisis.
6. A winning technology can leave its pioneers behind
Eisman recalls Henry Blodget correctly predicting the internet’s enormous impact while first-generation companies failed to capture much of the eventual value. AOL, Netscape, Yahoo and Excite are his examples; Amazon is the exception he identifies. Second-generation businesses prospered on foundations built by the first. “Maybe that will happen again this time.”
His self-described “conspiracy theory” is that Anthropic needs to list before third-quarter earnings, as the end of tokenmaxxing makes its performance look worse. He thinks OpenAI is in a similar situation. The hosts offer SpaceX as an example of dire IPO predictions giving way to typical technology-stock trading; asked whether Anthropic would follow that pattern, Eisman says he is unsure.
7. Housing weakness makes the AI dependence more consequential
Josh describes housing as the worst market since the financial crisis, while distinguishing the circumstances. Existing owners have been stuck because of high rates; Eisman says a wave of new homes benefited builders including Lennar, Pulte and KB. He says that situation is over as listings return and housing-related stocks fall.
Affordability involves more than mortgage rates. Michael says insurance, utilities and taxes have risen while house prices have resisted adjustment. He cites an estimate that total monthly homeownership costs are 50% higher than five years ago. Josh says the cost of a loan relative to income may be at its most difficult level in his lifetime.
Michael points to a roughly 70 million-person age cohort supporting underlying housing demand, but says he cannot afford to buy a home. He expects prices eventually to fall, though gradually over several years. Eisman says the process will be slow; lower rates would help, but the other ownership costs remain a burden.
Eisman rejects the romance of predicting catastrophe: his own experience brought “tremendous irritation and anxiety.” An AI collapse could produce a severe recession, he says, but the 40-to-1 leverage institutions used in 2008 is absent today. “It probably won’t be on that scale.”
Full transcript
You ask what an LLM is? It’s just a slightly more sophisticated word search system. Let’s predict the next word. It’s pretty good. I’m not saying an LLM is a thinking system, but it’s not like they think, right? Then why are you saying you’ll kill us? That sounds a bit strange.
It seems like they should have perceptual ability, but LLMs don’t. I don’t even think about it. It’s simply searching for words or creating software, and software can also be considered a type of word. The reason I think that is because Dario Amodei and Sam Altman are probably terrified by now.
They got scared because they thought tokenmaxxing was over. For those who don’t know what tokenmaxxing is, it means software developers or companies received orders from their firms to use AI 24 hours a day. It cost too much. As a result, prices skyrocketed, and Uber used up its entire AI budget in just 3 months. So, was it this year or last year? I don’t remember very well.
It was this year. It was spring.
Hmm. Tokenmaxxing is over. Open-weight models have come out that cost only one-tenth as much. They’re not as good as the existing models, but you don’t need a Mercedes-Benz every time you go on a trip, right? So we’re losing market share. There’s clear evidence that a price war is starting, and they were terrified. So they’re driving us in a direction like, “If you don’t regulate us, the world will end.” They think that if they’re regulated, they can prevent open-weight models from taking too much market share by manipulating the regulations.
So you think you can create a duopoly that way? However, the fundamental problem with the entire AI industry is that there is no moat. There isn’t a single one.
That’s right. There really isn’t a single one. A model might be good for a day, but people quickly switch to another model. It’s different from Google, isn’t it? As you know, in the past, everyone used Google forever. It was to the point where I couldn’t even imagine the concept of not using Google. You can use Claude now, but if you don’t like it, you can switch to Gemini, and if you don’t like that either, you can switch to GPT. There is no moat either. The fact that people spend trillions of dollars annually on services without a moat shows how easily people can switch to other services. I’m not even surprised. It’s just like that.
That’s right. I always thought that the best service would be the one that remembers the most information, or the one that can recall the most user information every time it’s used. But they’re all pretty similar.
That can happen. Which one would be the best? What’s the most suitable LLM for your work?
It’s not for my own purposes; it’s just for research. Everything is fine. Everything is good.
Is there anything in particular you prefer?
Our company uses Claude Enterprise as the standard.
Yes. As a result, I now do all the work I used to do in ChatGPT in Claude, since we’re paying the cost anyway. You just get used to it. I think enterprise LLMs are a kind of clever way of building a moat. If all the employees at Fortune 500 companies get used to using Claude, there’s no reason to use anything else, right? Of course, if it were much cheaper, I’d use something else. If an open-weight model performs almost the same at one-tenth the price, I’ll use the open-weight model.
Yes, that’s the point.
It’s not that I’m saying I won’t use Claude, but that Claude will have to lower its price. I’ve already heard rumors that prices are being lowered.
Yes. The founders of these AI research institutes are coming out and saying, “Please regulate us.” He even did a skit on Saturday Night Live. Now this phenomenon has become mainstream.
That’s right. Since they don’t want competition, it means that as the scale grows, they can effectively set the rules. It’s a lie that some people claim LLMs will destroy the Earth.
Yes. It’s a very simple lie.
1. Steve Eisman Joins Live
Great. John, then let’s start the broadcast. Great. Are you ready? I’ll begin. This podcast is provided by VanEck. We are living in an era of rising debt, deglobalization, and large-scale physical construction to support the AI economy. This means an increase in demand for energy, raw materials, and infrastructure. Historically, investment in real assets has been a very solid foundation. VanEck holds ETFs specialized in these real assets: RAAX and VanEck physical asset ETFs. RAAX is actively managed and adjusts investment weightings for gold, energy, infrastructure, and natural resources based on the current macroeconomic environment rather than past market conditions. RAAX is the best choice for real asset investment. For more information, please visit vaneck.com/raaxcompound. Thank you, John. Now, everyone. Welcome to Compound and Friends.
Welcome to Compound and Friends. We are one of the best podcasts in the United States. I’m very happy to announce that today’s guest runs the best investment podcast. While listening to the broadcast yesterday, I found out that some episodes are paid. It looks like I’ll have to subscribe to the paid episodes.
By the way, the name of our podcast is The Real Eisman Playbook. I wanted to let you know in advance.
Well, I don’t think I’ll need much help. You have a huge number of YouTube subscribers, you know.
That’s really good.
Thank you.
Were you surprised? I didn’t expect to succeed this quickly after the broadcast started. That’s a really big broadcast.
I was completely surprised.
Yes, but that’s cool.
Yes, it’s really cool.
Steve is an investor and the host of the weekly podcast Real Estate Expert Playbook, which started in 2025. Steve became famous for short-selling the subprime mortgage market while running FrontPoint Partners just before the 2008 financial crisis, a deal detailed in Michael Lewis’s book, The Big Short. He later founded Emrys Partners in 2012 and joined Neuberger Berman as a managing director and portfolio manager in 2014.
Ladies and gentlemen, Miami, everyone! Please give a big round of applause to Steve Eisman.
You already like me.
It’s really good. The University of Miami is a really good school. It’s nice that there’s a swimming pool in the middle of the school.
That’s right. So my son looked around the campus and said, “My son isn’t stupid.”
I’m not a fool. I remember when my daughter went to the University of Miami; she really enjoyed both her academic studies and the time she spent there. I told my wife, Valerie, who’s also from the University of Pennsylvania, “We looked like fools.”
Yes.
I was shivering in the cold while attending school in the middle of Philadelphia, but I could have gone to Miami. What did your daughter major in at Miami?
My wife?
No, no, no. My daughter majored in art history.
Buy art.
2. CoreWeave Debt and AI Concentration Risk
All right. I’d like to ask what you consider to be the most important question in the market right now. We’ve been planning to have you on for quite some time, and it seems we made the request at just the right time. Will the financing mechanisms for AI products collapse?
There was a problem with CoreWeave yesterday. CoreWeave is one of the companies called a neocloud, and it’s basically a data center that operates rather than being self-owned. It isn’t owned by a hyperscaler, nor does it generate cash flow from advertising revenue like Alphabet, for example. It seems that the current yield has reached 15% due to the surge in debt. As you can see from chart 5, the price has dropped by 60%. The situation isn’t good.
It’s really bad.
So what you’re seeing right now is the phenomenon where CoreWeave’s spread and yield are skyrocketing. It can be seen as a symbolic example representing all the companies that have to issue debt to finance these kinds of projects. You can’t just build half of these projects and leave them abandoned.
Of course, you could do that, but you’d go bankrupt.
That’s not a good idea. I’d like to hear the opinion of someone who experienced the financial crisis firsthand. If you were someone who predicted and prepared for the financial crisis, you might have bet everything. I think the current situation is somewhat different from back then, but I’m curious to hear your thoughts on the fluctuations in bond yields and the discussions about whether these projects can continue.
Compared to the financial crisis, there’s a big difference. For example, let’s assume General Motors goes bankrupt. What happens? Everyone is laid off, some of GM’s suppliers go bankrupt, and many people lose their jobs.
That’s right.
A recession will almost certainly hit that region. If bankruptcies spread, a nationwide recession could occur. If JPMorgan collapses, the Earth will burn up.
That’s right. It’s the end of everything. If you can’t do it, everything is over.
The reason the government had to provide bailout funds to everyone is that a situation where people can’t withdraw money from banks must not occur.
That’s right. This is in a similar context to the GM situation.
Yes. It’s a very big incident.
Hmm. What would happen if we couldn’t extract the chips from the data center? That would be a big problem for you.
Yes, but it’s not the end of the world.
Of course, I’ll cover the story of AI in more detail later, but let’s assume, for example, that OpenAI goes bankrupt tomorrow. I’m not making that prediction, but if that happens, I think the U.S. economy will immediately fall into a recession. Seems suspicious, isn’t it?
Yes.
How did it become such a systemically important company? This is different from the financial crisis.
Because the banks are fine.
How did it become such a systemically important company? OpenAI is a privately held company.
I understand that the company’s valuation in the private equity market is around $1 trillion to $1.5 trillion.
How did it become a systemically important company in just 3 years, one that could plunge the entire U.S. economy into a recession tomorrow?
Both OpenAI and Anthropic have promised to invest in any company capable of adopting their technology.
That’s right.
If you want to think about the problems with the AI story, simply put, the problem is market concentration. Two giant corporations, namely Nvidia, are leading the way. Nvidia’s revenue increased by more than 100% in the recent quarter. That’s surprising. It’s encouraging that the largest company on Earth has increased its revenue by more than 100%.
However, the 10-Q report states that 70% of Nvidia’s accounts receivable—a proxy for revenue—came from 5 customers. Why is that a problem? You don’t want that kind of customer concentration. What would happen if even one of those customers had a problem? You’d lose a significant portion of the entire business.
Of course. I think I’m asking a bit of a stupid question. I was just curious.
Let me give you an example from when I was at Oppenheimer in the 1990s. There was a very talented software analyst. I remember a conversation I had with her. Oppenheimer wanted to take a rapidly growing software company public, but it only had 2 clients. She said, “No.” The reason was that if even one customer left, the company would be finished.
Nvidia has 5 customers, but virtually every company at that scale is a customer.
No, they’ll probably be hyperscalers. Hyperscalers have hundreds of millions of customers. But 70% of hyperscalers’ AI revenue comes from Anthropic and OpenAI, and that figure accounts for about 25% to 30% of total cloud revenue.
Oracle is in a much more serious situation because half of its order backlog consists of OpenAI projects. You can see it in the overall scale of the investment. OpenAI said to Oracle, “We’ll use your cloud. We’ll use the computing resources you provide, so build it for that. And for your reference, we need 10 times more resources. Build all these data centers.” All of this construction happened because Anthropic and OpenAI demanded it like crazy.
3. Chip Selloff and OpenAI's Missing $20 Billion
Yesterday, we learned a few things. First, high inflation figures emerged again.
Yes. Interest rates have risen. They went up again this morning.
Yes. But the interesting point, as I’ll talk about later, is that interest rates have fallen significantly. That shows where the pressure is coming from: people are afraid.
Yesterday felt like the moment a bubble burst. CoreWeave’s debt has increased astronomically.
Yes, that’s right. They’re currently spending $640 million just on debt repayment, which is a very serious problem given their cash flow situation.
$2.1 trillion poured out of the market over the past 7 days. Did yesterday feel like the moment the bubble burst, or was it an opportunity to buy again? Looking back later, I’m going to say, “You stupid bastards!”
Before I answer, look at how $2.1 trillion vanished from the stock market in 7 days. You can see right on the screen how serious the situation is. This doesn’t happen in a normal market. It’s on a massive scale. Don’t you think people are too zealous? It seems like too many people want to be like Steve Eisman, the hero of the next big short-selling prediction. Honestly, it feels like a kind of disease.
It reminds me of something I said on CNBC. It was such a good point that I’ll say it again. It was the day Ray Dalio, whom I respect, predicted that the world would end because of the fiscal deficit. Was it 2015? He did it in 2015, and he’s doing it again now. 1937? I was like that back then.
Melissa Lee asked for my thoughts on the 5 p.m. show, and I said, “The problem with these people is that they all want to be like me. But I have news for them: Steve Eisman’s spot has already been taken.”
Yes.
The funny thing is, people want to be known as the person who predicted the next disaster. They want glory. But you know what? I’ve already enjoyed that honor. I don’t need it again.
Honestly, I don’t really know. I don’t think CoreWeave is important enough to shake up the entire AI industry. I think Anthropic and OpenAI are much more important.
To share some interesting data, OpenAI announced last week that its annual revenue as of the end of September was $50 billion. A month ago, they said it was $70 billion. Where did the $20 billion go? Has it disappeared? It’s gone.
But I don’t think that’s a valid comparison. I have absolutely no idea what annualized sales mean. No one will know. Today, an article about Anthropic came out in the Financial Times. Regarding the revenue, I could be a bit wrong, but I checked the figures. They say that sales in the sixth quarter increased by 43%, recording $16.8 billion.
It increased by 43% in 3 months, but that’s lower than the 140% growth rate in the second quarter. Of course, we all know that kind of growth rate isn’t sustainable.
Yes, that’s right. But the important thing is that the impact started to appear as the tokenmaxxing came to an end.
As companies reviewed their budgets and started saying, “We can’t spend money at this rate on AI,” the technology was so good that, thanks to tokenmaxxing, they were making more money than they could have continued to earn.
Yes, that’s right.
The interesting point is that under these interest rate conditions, financing for new data centers will become difficult, and proving that the financing is feasible will also become much more difficult. On the demand side, there may be some changes as existing methods change, but supply constraints remain severe in many places. If supply is limited and demand continues to increase, interesting dynamics may form in the market.
That’s possible. Part of the current problem is that there’s strong nationwide opposition to data centers. It’s difficult to quantify exactly what might be unpopular. I’m going to make a bit of a crude joke: maybe the L.A. Dodgers?
Yes, yes, that’s good. But as you know, the backlash is so severe that it looks like the pace of data center construction will slow down significantly. This will be good news for people who sell computing services because the demand for data is decreasing. If data centers disappear, it will have a positive impact on the market because fewer computing resources will be purchased.
Because if we can’t build a data center, there’ll be nowhere to put computing resources. But demand for existing computing resources will actually increase.
That’s highly likely.
Yes. Actually, I wanted to ask you about that, because I think it’s an important part of this story. I don’t think demand for AI itself will change.
The people in front of me visited all the investment banks on Wall Street over the course of 2 days. They say that in 2027, those banks will invest much more budget in AI and AI-related projects, and hire more AI-related personnel, than in 2026 or 2025. At the corporate level, demand for AI won’t change.
Another thing that won’t change is consumer demand. I think most people primarily use ChatGPT, and there’s also Claude. Looking at it from the agent perspective, there’s AI like Gemini, and virtually all Google searches are now being performed by AI. This technology won’t disappear either. Anyone using this technology will never say, “I should stop now. I didn’t really need this or make it part of my life. I’m just going to go back to 2022.” I don’t think that at all.
Bond prices or the difficulty of issuing debt won’t significantly change the demand side. What changes is who builds the capacity to meet that demand. If companies like CoreWeave or Nebius can’t supply as much computing capacity as they want by issuing debt, the ball will be in the hyperscalers’ court again.
Amazon, Alphabet, and Microsoft are in a much stronger position than they were a year ago. Back then, the cost of issuing debt might have been lower, or the demand for bonds might have been greater. What do you think of this idea from an investment perspective? It seems the position of hyperscalers will be further strengthened.
I agree that the position of hyperscalers has been strengthened. The point I find most concerning is that the business is too concentrated in a small number of companies. Anthropic and OpenAI say that even if price competition occurs, everything will eventually fall into place. The situation may change in 3 to 4 years, because by then many other companies will emerge to secure computing resources. But the market is so concentrated right now that we can only hope these 2 companies do well.
The pricing method is shifting from SaaS, or software as a service, which has been the sales method for enterprise software for the past 15 years, to a usage-based model. It used to be a system where costs were paid based on the number of employees. For example, if a company had 1,000 employees, Salesforce received $80 per employee. We’re now transitioning to usage-based pricing plans.
This approach appears more sustainable than existing service sales methods. Of course, prices have gone up, and it seems companies have gained more control—or maybe not. In any case, companies seem to be accepting these changes, even though it’s more expensive. As people become more sensitive to costs, employees will use cheaper models much more frequently, which will put pressure on Anthropic and OpenAI to reduce prices.
Yes, that’s a pessimistic outlook. Great.
Do you think this is the beginning of the phenomenon we’re witnessing?
4. AI Price War Bear Case and Oracle
I believe there’s a possibility that this could be the start.
What is the propagation mechanism by which a problem at OpenAI leads to various events, resulting in an overall economic recession?
Let’s assume a full-scale price war breaks out. I don’t know if it will actually happen, but let’s assume it does. Anthropic’s sales growth turns negative. No one will invest, and eventually investment will have to be reduced. Oracle will go completely bankrupt because half of its order backlog came from OpenAI.
For your information, Oracle’s stock price has completely crashed.
That’s right.
For those who don’t know, around the third quarter of last year, before Oracle’s earnings announcement, the stock price was $230. But when Oracle announced its earnings, they said their order backlog had increased from $138 billion to $455 billion. I don’t remember the exact figure, but it increased by $300 billion in 3 months. Everyone went wild, and the stock price jumped from $230 to $330 in just 2 days.
It’s the first time I’ve seen a stock price go up by 100 points. Those were the stock prices of the largest companies in the market at the time. It was a tremendous upward trend.
I think it was Gil Luria from Davidson; someone published a report saying that all $300 billion was due to OpenAI. Everyone was in complete panic. That was the peak. The stock price fell from $330 to $230, and now it has dropped to $130.
Don’t forget that people can’t trust companies that make them lose money. You bet your future on a company that’s pouring money into it like crazy?
What did you say?
The stock market has been really exciting over the past few days.
I agree. As I mentioned yesterday, with the inflation figures coming out high, it seemed like people were thinking, “Oh no, this is the end.” It was really serious.
It was even more serious this morning. There are no signs of improvement. Producer prices came out today. Yesterday was the Consumer Price Index, and today is the Producer Price Index. This isn’t good either. Memory-related stocks plummeted yesterday.
Let me tell you a fun fact. Western Digital fell by as much as 54%.
54%!
Even so, it has risen 100% since the beginning of the year.
Right. This shows how sharp the rise was. I’ve never heard of a stock price doubling from the beginning of the year even though it was cut in half. I can’t believe it.
That’s right.
The rate of return skyrocketed yesterday, but that doesn’t help those who bought at the peak.
Yes. The fact that returns have skyrocketed doesn’t help. Although Anthropic is not a listed company, a blockchain-based company called Hyperliquid is massively increasing its trading volume. $17.6 million worth of Anthropic stock was traded over the past 24 hours. This system is a kind of market and works quite well. It worked very well during the SpaceX IPO, too. This is Anthropic traded on Hyperliquid.
Ah, I see.
Anthropic had poor earnings yesterday. Of course, that would have been the case. Sales figures were released after the market closed, and they completely plummeted.
It plummeted by over 70% in over-the-counter trading. People went crazy. Twitter turned pitch black last night. People were in a complete panic, and it was the same this morning.
I wasn’t flustered at all.
You were very calm. I saw it. Interestingly, Anthropic’s stock price went up about 5.5% this morning. Returns are falling significantly, and the stocks surging today are the worst stocks in the market. For example, Lennar seems to be the best-performing stock in the S&P 500 today. It’s a housing-related stock, so it’s everything even slightly related to interest rates. It’s fascinating how quickly market trends and the atmosphere change.
Who would care about interest rates last week?
Interest rates probably wouldn’t matter at all.
Interest rates probably wouldn’t matter at all. That’s what I thought. But suddenly, the market started to panic. Ironically, that was the bottom. It’s really interesting that things like this happen in the stock market. Once interest rates peaked, did the stock market need some bad news to hit bottom? Anthropic stock broke through its low point, and the S&P 500 surged in just one day, didn’t it? And equal weight is taking out the lows.
Well, this is the best game on Earth. Are you surprised that interest rates dropped so quickly?
I was actually quite surprised. The inflation figures were very bad. It seems the state of panic was severe enough for people to buy government bonds. But what we learned is exactly this: I fully agree with the questions or arguments of the past few months, and with your statement that the previous concerns about the deficit were nothing but bluff.
I did say that.
When?
Hmm, with us. Ah, the last time. That’s a good expression. But he spoke passionately.
It was very popular. The market is like a difficult-to-unravel ball of yarn. You can’t always know why it moves for one reason or another, but on some days you can identify the factors driving the trade. And yesterday we found out. Is it because of the financial irresponsibility of governments around the world? Or could it be because of someone?
No, it’s because of AI. AI is leading the trading. As soon as even the slightest sign of weakness or sluggishness appeared in that trade, the 10-year Treasury yield plummeted from 5.4 to 5.1.
Yes. It’s all because of AI. That doesn’t mean we’ve suddenly become more reassured about financial matters.
I don’t think so. People thought that bond selling, which was being conducted to boost AI trading, would slow down as investor sentiment weakened. So the government bond market breathed a sigh of relief and rose.
Yes.
Yes.
5. CoreWeave, Insurers and Mark Walter
That was an interesting part. Over the past year, the CEO of CoreWeave has sold a massive amount of stock. Show me Chart 5A, Daniel. The CEO of CoreWeave sold $682 million worth of stock. There’s a chart showing the price, volume, and selling position.
To make an excuse for him, he’s not the kind of person who suddenly says, “Ah, I don’t like this stock anymore.” He has always sold stocks. All of this was planned in advance, and a significant portion of it is likely to pay taxes on the shares he received from shareholders.
Nevertheless, if a company is under such thorough scrutiny, eventually it will say to its shareholders, “Fine, I won’t sell my shares anymore. There’s no point in doing this any longer. I trust this company.” One might expect him to say that, but he didn’t. And yesterday, I’m not sure if it’s right to say “adding insult to injury,” but Steve, I’d like to hear your thoughts on this issue. The Financial Times published an article stating that CoreWeave is aiming for unlimited insurance money to fund the construction of AI. That’s the main point of the article. Striking that pose, he said, “I believe now is the right time to make an investment proposal by leveraging my insurance contract.” Imagine saying that.
The point is that Mark Walter has to sell the LA Dodgers back to his insurance company. Actually, he bought the Dodgers, didn’t he? And you have to invest the proceeds in CoreWeave.
If that happens, 2 potential scandals will become entangled into 1 big story. He has a lot of money, so he’s just saying he’ll give it a try.
Steve, this is one of the market areas where people anticipate the next generation of short selling. It’s where people plant flags.
I completely agree. Life insurance is risky. Let’s go back to Mark Walter’s story. We just covered that on the podcast. He bought the LA Dodgers, the LA Lakers, and many other sports teams, but I didn’t disclose that. He just bought them because he wanted to. It means they want to own the LA Dodgers. The Dodgers are the same. But I want them to buy me.
So he lent money to another company through an insurance company, and he ended up owning that company as well.
That’s right. Or they’re trying to cover up the fact that they intend to use that money to personally seize control of the management. Anyway, it worked out well. That went really well. The price went up. The Lakers, too. He got his hands on $2 billion in an instant. I don’t know if he said it himself or if others said it, but I’ve seen something like that. I think Matt Levine wrote about this topic. Someone called this a crime, but they called it a victimless crime because the value has gone up.
The rebuttal was, “Wait a minute. You can’t do that.”
I suppose so. Of course it’s illegal. But if Grandma had known she was investing in the contract, she wouldn’t have bought it.
That’s right. Then you wouldn’t have been able to receive the money at 3% interest. You would have had to pay 12% interest. But let me tell you a fun fact: Mark Walter’s insurance company is headquartered in Delaware, and there’s an insurance regulatory agency in Delaware. Okay, here’s the question. What did the current Delaware Insurance Commissioner do before becoming an insurance commissioner? What was his job? What kind of work did he do for a living? Try to guess.
Did he work in finance?
No.
Hmm, try to guess. Was it a podcaster?
He was a sheriff.
Ah, now I understand. And he’s an insurance supervisor.
Yes, I understand perfectly now. I have no ill feelings toward the sheriff. They protect us. However, when it comes to insurance, associating sheriffs with insurance is a bit strange. Great. Okay, here’s the question. Many people writing on Substack are mentioning that insurance companies are investing in AI data centers. My question is: How dangerous is this for the policyholder? How will these investments affect insurance policyholders? Will the claim made by those writing about this issue—that there’s greater systemic risk to ordinary people than to speculators—be proven true? What do people think about this part? Aren’t you particularly worried?
They’re literally controlling policy.
I’m not worried. The important thing is finding the answer.
6. AI Regulation: Will Anyone Be Held Liable?
Yes. Great. Steve, before we wrap up, there are two more things I’d like to ask you. You just mentioned the political backlash. I’d like to ask about that, not just in relation to data center construction, but overall.
I heard an episode you shared with someone who recently gave a speech at the UN about the dangers of AI. The White House has released 4 plans for AI regulation. It sounds like everyone is fending for themselves, but there’s talk of establishing a committee composed of AI industry professionals. The committee’s guidelines say that responsibility must be taken if a problem arises.
Imagine that something you invented goes out of control and causes damage. I agree with that point, but who’s responsible? Should investors who invested in such a company pay out, or should the executive who created something truly harmful be held personally responsible? I’d like to hear your opinion on whether regulations like these will be effective. And if you have any other suggestions, which ones would be good?
In my opinion, if you run an AI company and that company hacks someone or commits wrongdoing, the company must be held responsible. Of course, you have to take responsibility.
But the more interesting question is this: I’m the CEO of that company, and I knew this would happen, but I didn’t care, and then it happened. Who would sue me? What happened in 2008? That’s the point I made next. No punishment was handed down to the executives who committed the worst fraud in history.
I was on a podcast where I interviewed a colleague. She often talks about corporate fraud. We talked about fraud and discussed how the CEO must take responsibility. She asked me, almost pleading, “Steve, what can I do?” I said, “I can’t do anything, because I won’t do anything anyway.” That was my answer.
That’s right. Unfortunately, that’s probably the case, at least for now. No one went to jail during the 2008 financial crisis, so there’s no way anyone would go to jail for hacking, right? Can current regulations really prevent anything? I’m a bit skeptical. I think regulations will only be put in place after a crisis hits.
It’s always like that.
What do you think about the government investing equity in selected successful companies? You must like that.
I think it would be terrible.
You don’t like the government investing equity in AI research institutes or projects?
No, because then everyone would think the company is too big to fail. I don’t want a bailout. I don’t want government intervention.
7. Hyperscalers, Anthropic IPO and SpaceX
All right. Looking at the current structure of the hyperscalers, the top 7 companies, known on Wall Street as the “Mag 7,” excluding Amazon, Alphabet, Apple, and Tesla, are all jumping into this business. Meta has built its own data center, but it could also lease computing resources to outside companies if it wanted to.
Do you think these companies’ market capitalizations will reach $10 trillion within the next 5 to 10 years, or have they already reached their natural limits of scale?
I don’t think there’s a natural limit to scale.
All right. But I think the important question is whether this will unfold like the growth of the internet. When I was at Oppenheimer, Henry Blodget had an office right across from mine.
That’s right. He went out and said that the internet would generate tremendous wealth and conquer the world. He was 100% right. But he was wrong about the first generation of internet companies. The second generation went from strength to strength, using the foundation established by the first generation as a stepping stone. Maybe that will happen again this time.
Companies like AOL, Netscape, Yahoo, and Excite ultimately failed. Of the first-generation companies that come to mind, Amazon is the only one that’s done relatively well. It’s not that the current situation is better or worse. In the past, though, there were too many companies that didn’t even qualify to go public.
That’s true. There are 2 companies that have taken over everything right now.
Yes, that’s right.
Do you think Anthropic will go public this year?
It really should.
Yes, I agree.
I think they need to list right now because of Token Max. Their performance is going to start looking really bad. They have to grab the opportunity while the iron is hot. My conspiracy theory is that Anthropic needs to go public before the third-quarter earnings announcement.
All right. Why does OpenAI seem like it isn’t in such a hurry?
I think they’re in a similar situation, and that’s how it should be.
I don’t understand. All right. Do you think Anthropic’s IPO will go well? Aside from the fact that its performance isn’t particularly good, do you think the IPO will be well received?
Let me put it this way. SpaceX was the biggest IPO we’d ever seen. There was a lot of pessimism on Twitter, where pessimism is prevalent. People predicted it would hurt index funds and that money would flow out of other stocks to cover the size of the IPO. There was talk that shareholders would suffer huge losses from insider selling, and that it would start at $135 and plummet to $50.
With insider selling allowed, people could sell. Many said the stock would be a huge disaster, a disaster for everyone. But that didn’t happen. After listing, the stock price fell, rose again, and fell again, but it’s moving like a typical tech stock.
A company like Anthropic is of a similar size. Do you think it will have a similar reaction before gradually adapting to the market? I’m not sure.
Great. Let me ask you this: would you buy it if it went public? You’d never buy it, right? Then let me ask you something you might be able to answer.
I’ll give it a try.
Earnings increased by 30% this year, but investors are saying that growth can’t be sustained. The price-to-earnings ratio we’re willing to pay has also dropped sharply. What do you think about this situation? Is it a bull market or a bear market, or are you not bothered at all? I’m curious what you think.
I don’t know much about valuation. It’s sensitive in that respect.
All right. But what is the market telling you?
That’s purely a story. There’s only one important story right now, and that’s AI. If you tell me how the AI story will turn out, I’ll tell you how the stock market will turn out. That’s all.
Many people say the stock market isn’t the economy, but right now, I think that’s true to some extent. I believe that companies that haven’t even been listed yet have the ability to cause an economic recession. I believe the market performance of listed companies stems from derivatives of AI trading. And the bond market has come to realize, as Michael said yesterday, how important the ability to issue bonds is to yields.
I think the current economy has become one of those situations where it’s like the stock market. It won’t last forever, but right now, that’s the case. I think that’s true now more than ever.
Well, in that case, how important is it for this market to hold near its high, or for a catastrophic event not to occur? It feels like feedback, but I don’t think so. Let’s get back to my original point. If you look at the S&P, tech stocks make up 40%, and companies like Google and Amazon account for 50%. If you add investment banks, GE Vernova, Quanta, Eaton, Rockwell, and others, it amounts to 65% of the S&P. What about industrial stocks? In the end, it’s all just one trade. It’s AI.
There’s no need to discuss this further, right? If demand holds and someone is willing to support construction funding, it’ll be fine.
8. Housing Market: Worst Since the GFC
It’ll be fine. There’s another area I know well: the housing market.
Yes.
We’re currently in the worst housing market situation since the global financial crisis. Clearly, it’s not exactly the same situation, but it’s definitely bad. I can’t believe nobody cares, even though one-third of the economy is at recession levels. Nobody cares. Everything is running smoothly thanks to AI, right? That’s a huge relief. Thanks to AI, the stock market, and the way the current economy is structured, everything is running smoothly.
2022 was a really strange year for homebuilders. It’s really strange that housing construction stocks are booming when the housing market is paralyzed. Everyone who owns a home has been completely stuck. Interest rates were so high that I couldn’t sell my house, and there were neither buyers nor sellers. So what was left?
A massive number of new homes poured out. All the construction companies, including Lennar, Pulte, and KB, enjoyed a boom. They poured out all sorts of benefits like a frenzy.
But that situation is over now. It seems like people are gradually getting used to this reality: “This is how interest rates are supposed to be. This is what mortgage loans are really like.” But listings will definitely start appearing on the market, and housing-related stocks are plummeting. Lennar is falling day by day. There was a huge rebound today, which is fortunate. I hope it lasts a long time.
What do you think about the housing market?
I’ll think about it for a moment. The housing market is stuck because of interest rates. Housing prices have probably never been this cheap in our lifetime.
It wasn’t just that it was unaffordable. Considering the cost of the loan relative to income, it has probably never been this difficult to manage.
You’re right. It’s a disaster. Mortgage interest rates aren’t the only problem. Of course, interest rates are a problem, but housing insurance premiums, utilities, and taxes have also skyrocketed. Even so, housing prices didn’t go down. According to an estimate I received, if you include all the monthly costs of owning a home, home prices today are 50% more expensive than they were 5 years ago.
Yes, that’s right. I think so too. As a landlord, I believe housing prices will go down.
I think it’ll end that way. Yes, that will happen. It’s already started going down. Does it go down when the economy is in a recession, or does it go down because more is built? If supply decreases, supply in the market will increase.
For the first time in a long while, the number of home foreclosures is increasing quite significantly. Supply is increasing. However, demand is still huge. There are as many as 70 million people in my age group, and while not everyone needs a home, many people do. But I can’t afford to buy one.
Yes.
If the price doesn’t go down, I can’t buy it.
Yes.
You’ll be able to buy it when the price goes down. So I think housing prices will go down.
Do you think that will happen?
Eventually, housing prices will fall, but the process is so gradual that it seems it’ll take several years. Could a decline in the housing market actually revitalize the economy?
It’ll proceed slowly.
Do you think that’s impossible with interest rate cuts alone?
Dropping interest rates would be helpful, but there’s still a long way to go. As I mentioned, the burden of insurance premiums, utility bills, and taxes has become so heavy that it’s difficult to handle.
9. Who Will Call the Next Big Short?
It’s really hard to handle. Lastly, you said that everyone tries to be like Steve Eisman, but I don’t want to be like Steve Eisman.
You’ve already done enough.
There are more people predicting crises now. It’s not necessarily about betting money on predictions, but about using your reputation. You spend your time writing very negative research reports or Substack posts, or retweeting all sorts of bad news on Twitter. I think that if something bad happens, you’ll eventually look for the person who said the right thing before it happened. And that person will be regarded as, “This is the very person who foresaw that situation,” whether they were right or wrong. There are hundreds of people like that. There really are a lot.
Do you think that will happen after a crisis? Will we end up finding the so-called “person who spoke the truth”? Or will people say, “What you did will be difficult to replicate in the next crisis”? Because he didn’t just tweet. That’s the difference.
Yes, I think that’s right.
But what do you think when you see so many people competing to say the most negative things possible about what will happen next? Shouldn’t you be happy that you were wrong?
Yes. That’s exactly what I want to tell you. They were wrong, and I’m glad they were wrong. Look, there will be an adjustment someday. I don’t know what kind of adjustment that will be.
Is the lesson “Don’t make wishes carelessly”? Are you asking if it was good for you to become what you are after this situation?
It wasn’t good at all. To be honest, I was plagued by tremendous irritation and anxiety. Many people probably got angry at me. “You ruined the world. It’s all your fault.” I heard that a lot, and I was overwhelmed with anxiety. As you know, I never want to go through that again after hearing that everything is my fault.
Now, as I mentioned, whatever happens, it won’t be a serious situation like a financial crisis. You mentioned that institutions used 40-to-1 leverage in the past, but that’s not the case now.
However, if AI really collapses, could it lead to a severe economic recession?
Of course. However, it probably won’t be on that scale.
Well, I think that’s a good way to wrap things up. To those reading all these writings, comments, and opinions, I believe that although you’re not the ultimate authority, the perspective of someone who has gone through a much more serious situation will be helpful.
I don’t think it necessarily needs to be resolved that way.
Thank you so much for saying that, and thank you sincerely for appearing on the show today.
Steve Eisman, everyone. Shall we wrap up by taking some time to talk about our individual resumes?
How was it?
Yes, that’s good.
Everyone, on behalf of Compound and our friends, thank you for becoming the impromptu live studio audience today. I hope you had a pleasant time. Steve, Michael, and I all sincerely wish you the best of luck with your graduation from Miami and the start of your future careers.
Please give a round of applause.
Thank you.