Oct 2, 2026 · 54 min
Listen from 15:55
Listen at 15:55
Eisman warns AI debt could turn a market boom into a downturn
Steve Eisman: One Company Could Break The AI Boom
The episode connects weakening AI moats, debt-financed infrastructure, and higher Treasury yields to risks spanning technology, credit, and the broader market.
- 1Anthropic and OpenAI sit at the center of an interconnected AI ecosystem whose failure could trigger a broad recession.
- 2Off-balance-sheet infrastructure debt and higher long-term yields could pressure data centers, corporate borrowers, and heavily indebted AI firms.
- 3Eisman is reducing exposure, monitoring open-weight competition, and favoring disciplined risk management over aggressive macro bets.
Don't miss
Eisman identifies Anthropic and OpenAI as the AI ecosystem’s most important companies to monitor, while distinguishing a recession from true systemic risk.
The brief
Steve Eisman joins Ed Elson for a quarterly market review centered on a provocative question: are leading AI companies amplifying a crisis through spending, regulatory rhetoric, and debt-financed infrastructure?
Anthropic’s losses, potential IPO, and slowing competitive moats become a test of the AI business model, while open-weight models threaten to intensify price competition across the ecosystem.
The risk extends beyond model companies: off-balance-sheet data-center debt, large AI-related issuance, and a possible 5% ten-year Treasury yield could pressure infrastructure economics and corporate borrowing.
Eisman says he has reduced exposure and hedged some AI positions, while watching oil, interest rates, open-weight models, and Anthropic’s filings for signs the narrative is weakening.
He also explains his FICO short, arguing that VantageScore threatens the company’s former mortgage-scoring monopoly, before returning to the lesson of 2008: reduce risk when positions stop allowing comfortable sleep.
What was said on this episode
45 statements · 6 positive · 33 negative · 6 neutral
Open-weight AI models are taking substantial market share.
“The open weight models are taking big market share.”
Listen at 3:00
Major AI companies currently have no durable competitive moats.
“they realize that there are no moats around their business whatsoever”
Listen at 3:07
AI is not currently close to achieving artificial general intelligence.
“there's absolutely no evidence whatsoever that AI is anywhere close to AGI. None. Absolutely none.”
Listen at 3:59
Current AI functions as next-word retrieval rather than genuine thinking.
“what AI is, is like a next word retrieval model. That's what it is. It doesn't think.”
Listen at 4:07
Open-weight models are beginning to capture very large market share.
“the open weight models have really come into their own, and as far as I can tell, are starting to take very large market share.”
Listen at 5:53
Anthropic and OpenAI face increased competition from open-weight models.
“if I was Anthropic and OpenAI, I'd be nervous because I had everything to myself and now I don't.”
Listen at 6:03
Anthropic will experience a slowdown by the fourth quarter.
“I actually think by the fourth quarter we'll definitely see a slowdown.”
Listen at 8:46
AI companies are beginning to engage in a damaging price war.
“you're starting to see signs of a price war breaking out by all the various players, which is very bad.”
Listen at 10:17
The AI business lacks durable competitive advantages.
“this business has no moats. I think is really true.”
Listen at 10:25
Nvidia’s accounts receivable were highly concentrated among five customers.
“70% of their accounts receivable were from 5 accounts.”
Listen at 11:18
Hyperscalers derive 70% of AI revenue from OpenAI and Anthropic.
“70% of their AI revenue is from OpenAI and Anthropic.”
Listen at 11:26
Problems at OpenAI or Anthropic would threaten the entire AI supply chain.
“if there's a problem with those 2 companies, then I think the whole chain is in trouble.”
Listen at 11:46
Market participants will better understand AI pricing within six months.
“at some point within the next 6 months, we'll have a better idea.”
Listen at 13:06
Meta used an off-balance-sheet vehicle for nearly $30 billion of Louisiana data-center debt.
“Meta did a— last year did an almost $30 billion deal where they're building a data center in Louisiana and they created some off-balance sheet vehicle where basically all $30 billion of debt or something like that doesn't show up on their balance sheet.”
Listen at 16:03
AI infrastructure companies are highly concerned about preserving credit ratings.
“these companies are very, very nervous about their credit ratings, and they're trying to do everything they can to preserve them.”
Listen at 18:38
A 5% or higher ten-year Treasury yield harms housing activity.
“it hurts housing”
Listen at 23:27
Higher ten-year yields increase corporate borrowing costs.
“any corporation that's going to raise debt, it's more expensive.”
Listen at 23:48
CoreWeave’s comparable borrowing cost would now be approximately 11%.
“CoreWeave raised debt 6 months ago at 9%. Today it'd probably be 11%.”
Listen at 23:57
Higher interest rates make data-center economics more difficult.
“if rates are 100 basis points higher, it just makes the transaction more difficult.”
Listen at 25:46
Oracle is extremely concerned about its credit rating.
“I think Oracle's petrified of their credit rating.”
Listen at 26:28
Oracle wants to avoid a downgrade to junk status.
“Oracle does not want to get downgraded to junk.”
Listen at 26:37
Oracle has the weakest balance sheet among larger AI companies.
“It's the weakest.”
Listen at 26:51
Oracle and CoreWeave are riskier than Meta, Google, and Microsoft.
“I don't lose sleep over those companies. I worry about Oracle, CoreWeave, and all the other companies at this point.”
Listen at 27:30
AI debt issuance is crowding out Treasury-market borrowing.
“AI is a $500 billion issuance this year, and it's having a crowding out effect of the Treasury markets.”
Listen at 28:44
Without AI borrowing, ten-year Treasury yields would be much lower.
“If there was no AI, None. The tenure would be much lower.”
Listen at 28:53
AI debt issuance competes with Treasury borrowing for long-term capital.
“there's a bit of a competition which didn't exist before.”
Listen at 29:30
AI debt issuance contributes to elevated ten-year Treasury yields.
“the 10-year is as high as it is, because like I said, there's a crowding out effect where AI debt is crowding out Treasury”
Listen at 29:39
The Treasury Department has lost substantial credibility after its buyback effort failed.
“the Treasury Department has lost a lot of credibility”
Listen at 30:11
Treasury markets are highly efficient.
“The Treasury markets are pretty efficient. They're more than pretty efficient, they're really efficient.”
Listen at 30:50
New transformative technologies usually become successful.
“9 times out of 10, the new, new thing becomes usually successful.”
Listen at 33:02
The AI investment story currently has exceptionally high concentration risk.
“the incredible, incredible concentration”
Listen at 33:46
A breakdown in the AI narrative would cause a major stock-market correction.
“if the AI narrative breaks, the stock market will have a huge correction.”
Listen at 35:40
Oil prices and interest rates are currently the only variables that matter.
“the only 2 variables that matter are oil prices and interest rates.”
Listen at 39:14
Investors should avoid making major bets on oil prices or interest rates.
“I think you're supposed to do nothing”
Listen at 40:09
Eisman is not prepared to predict an imminent collapse of the AI story.
“I am not willing to make some major call that, you know, the whole AI story is going to implode.”
Listen at 41:06
FICO raised prices 1,600% over the past five years.
“over the past 5 years, they raised prices— get this— 1,600%.”
Listen at 42:10
Bill Pulte helped create VantageScore as an alternative to FICO.
“Bill Pulte has created, has helped create an alternative score. Called VantageScore.”
Listen at 42:40
VantageScore could reach 50% market share within months.
“it's possible within a few months it could have 50%”
Listen at 43:11
FICO will lose its monopoly in credit scoring.
“you had a monopoly and you're not going to have a monopoly.”
Listen at 43:33
Investors should reduce or hedge positions if necessary for comfort.
“if the result is to sell something or take down some risk or short something just to make yourself feel more comfortable, that's what you should do.”
Listen at 45:22
Trouble at Anthropic or OpenAI would endanger the broader AI ecosystem.
“if one of these 2 companies gets into trouble, everybody's in trouble.”
Listen at 47:07
A major AI-company bankruptcy could quickly push the U.S. economy into recession.
“the U.S. economy would probably either be in or very close to a recession very quickly.”
Listen at 47:43
Failure of Anthropic or OpenAI would not constitute systemic risk.
“That is not though systemic risk. That's just a recession.”
Listen at 47:48
Eisman has no interest in investing in Anthropic at its proposed valuation.
“Zero. I mean, it's like SpaceX. I had no interest in SpaceX”
Listen at 48:32
Approximately 86% of the S&P 500 is exposed to the AI trade.
“the remaining 86% of the S&P to one degree or another is AI-ish.”
Listen at 51:05
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
