Market Cracks Under AI Debt Burden | The Weekly Wrap

CoreWeave

Across the shows: 6 speakers in other episodes, 0 bullish and 7 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBearish

  CoreWeave's heavy debt and 15% bond yield signal fixed-income doubts about AI data-center financing.

  CoreWeave, a neocloud that builds AI-only data centers, has about $35B of debt plus $10B of capital lease obligations against a $43B market cap, and it competes with hyperscalers that have much larger balance sheets. Its debt was issued at a 9% coupon, which Eisman calls very high. The debt has since fallen below 90 cents on the dollar, implying a yield near 15%, which he reads as fixed-income markets quickly losing confidence. CoreWeave fell 15% on Wednesday and Nebius fell 10%, and he sees this as investors becoming afraid of AI debt levels, with the AI complex going down together.

  “This is not a company with a great balance sheet to say the least. When it raised its debt, the coupon it had to pay was 9% which is very high. However, it looks like the fixed income markets are getting really nervous about CoreWeave.”at 6:09

Anthropic

Across the shows: 18 speakers in other episodes, 9 bullish and 29 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBearish

  Anthropic's annualized revenue disclosure is a fundraising tool, not a reliable measure of performance.

  Eisman says he has no idea what annualized revenue means in the figures that Anthropic and OpenAI both publish. He thinks the data point is merely designed to help them raise money regardless of the underlying business. The argument is thin and applies equally to both labs, and it rests on his doubts about the metric rather than on Anthropic's own numbers.

  “To me this annualized revenue data point that both Anthropic and OpenAI employ is merely designed to help them raise money regardless.”at 7:10

OpenAI

Across the shows: 14 speakers in other episodes, 4 bullish and 13 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBearish

  OpenAI's annualized revenue claims look unreliable; a $20B drop from its prior figure is a bad sign.

  The Financial Times reported that OpenAI claims $50B in annualized revenue as of the end of September. Eisman says he cannot tell what that means, since it could be one day times 365, one week times 52 or one month times 12. He sees the metric as designed to help raise money. Previously OpenAI had claimed $70B in annualized revenue, so $20B has disappeared, which he calls not a good sign.

  “The most important part of the story was that previously OpenAI had claimed that it had $70,000,000,000 in annualized revenue. I have no idea how 20,000,000,000 evaporated but that is not a good sign.”at 7:19

The Progressive

• Steve Eisman · The Real Eisman PlaybookUncertain

  AI shopping agents could erode Progressive's brand moat, though AI isn't yet the cause of its slump.

  Eisman argues that AI agents do not care about brands and only weigh credit ratings and price, which could seriously impair the franchise value of personal lines insurers such as Progressive and GEICO. But Progressive's September report showed new premium volume falling for the first time since November 2023, and he thinks the likely culprit is soft pricing that is leading Progressive to pull back. He doubts agentic AI is yet big enough to matter materially, though he expects investors to speculate on it. Progressive fell 4% on Wednesday.

  “AgenTic AI does not care about brands. It cares about credit ratings and price. ... Leaving aside credit ratings, AgenTik AI only cares about price and this could seriously impair the franchise value of personal lines insurers.”at 11:07

Wells Fargo & Company

• Steve Eisman · The Real Eisman PlaybookBearish

  Wells Fargo is not the large bank to buy; its revenue has stalled and it keeps missing.

  Wells Fargo beat on earnings with EPS of $1.87 against $1.66, up 13%, but missed on revenue, which rose only 2% year over year. Eisman says Wells has not grown revenue in multiple quarters and seems to miss every other quarter or so. With returns on tangible common equity of 12% to 16% and a valuation of 1.2 to 1.8 times tangible book, he says it is not the large bank he would look at if he were investing in one.

  “Revenue missed. Total revenue was up only 2% versus last year. In fact, Wells has not grown revenue in multiple quarters. ... If you were interested investing in a large bank this is not the one I would look at.”at 20:48

Citigroup

Across the shows: 1 speaker in other episodes, 0 bullish and 1 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBullish

  Citi continues to deliver good results under Jane Fraser, with earnings and returns holding up.

  Citi reported EPS of $2.85 against $1.86, up 53% year over year. While higher rates hurt Goldman's fixed income results, Citi did fine. Its return on tangible common equity was 11.5%, below the 13% of the prior two quarters, but Eisman still calls it pretty good for Citi. The support for the view is solid results rather than a strong valuation case, since Citi trades in the lower valuation group.

  “Under the leadership of Jane Fraser Citi continues to report good results.”at 21:02

Banks

• Steve Eisman · The Real Eisman Playbook

  AI agents moving cash to money market funds could impair bank net interest margins; banks will defend.

  Checking and savings deposits are a cheap funding source and the most important driver of bank net interest margins, and many customers are lazy about moving cash to money market funds. Investors fear agentic AI will sweep excess cash out of banks and brokers into higher-yielding funds, which would seriously impair net interest margins. Eisman says banks will try to protect their franchises, and JPMorgan has already announced it will not allow outside agents access to customer accounts, so the fight has begun. He calls this the long-term risk for banks.

  “Now let's think about Agentic AI. What investors are worried about is the idea that consumers will use Agentic AI to sweep their excess cash out of banks and brokers and into nice paying money market funds. If this were to happen it would seriously impair the net interest margins of banks.”at 10:16

• Steve Eisman · The Real Eisman Playbook

  Bank stocks are priced for a strong investment banking cycle, leaving little margin for error.

  Eisman values banks on return on tangible common equity. Goldman, JPMorgan and Morgan Stanley trade at 2.6 to 3.4 times tangible book with returns above 20%, while Wells Fargo, Bank of America and Citi trade at 1.2 to 1.8 times with 12% to 16% returns. Buying banks at these levels requires assuming the investment banking cycle stays very strong, which is driven by trading volume, AI financing needs and M&A. Some froth has left since July but not much, so he sees little margin for error and says it is not a great entry point.

  “To invest in banks at these valuation levels requires you assuming that the investment banking cycle remains very strong. ... There is just not much margin for error at these valuation levels. To me, it's just not that great an entry point.”at 23:02

Telecom & connectivity

• Steve Eisman · The Real Eisman Playbook

  Starlink threat to US carriers is probably overblown; Starlink still lacks a good terrestrial network.

  AT&T, Verizon and T-Mobile fell 9% to 13% after SpaceX's Starlink, which Eisman says provides service that is much worse than telecom providers', bought 14 MHz of spectrum. That improves Starlink's US service but is still not good enough. Building a terrestrial network would be unbelievably expensive, so the cheapest route is a deal with one of the big three carriers, which Eisman believes have no interest in helping. He concludes the telepocalypse is probably overblown, though the narrative is powerful and will not go away.

  “And none of the big three believe me have any interest in helping Starlink. So the telepocalypse is probably overblown.”at 4:45

Europe's economy

• Steve Eisman · The Real Eisman Playbook

  France's debt at 120% of GDP and no political will to cut spending signal dangerous fiscal risk.

  Investors are losing confidence in France's ability to manage its deficit. The spread between French and German 10-year bonds is 140 basis points, a multi-year high. Debt is 120% of GDP, a dangerous level, and no consensus or desire exists among French politicians to cut expenditures. Marine Le Pen, who may become the next president, advocates cutting the retirement age from 62 to 60, which would worsen the fiscal problem.

  “Investors are losing confidence in France's ability to manage its deficit. The spread between the French ten year and German ten year bonds is at 140 basis points, a multi year high.”at 7:38

US deficit & government debt

• Steve Eisman · The Real Eisman Playbook

  US debt at 120% of GDP is a lesser problem than France's because Treasuries anchor the global system.

  US debt is also about 120% of GDP, which is a problem but a lesser one than France's. The dollar is the world's reserve currency, and the global financial system functions on Treasuries. The repo market, a multi-trillion-dollar market for banks, uses only Treasuries, and large institutions park money in them. Until a large, safe and liquid alternative exists, the US has the luxury of time, though not endless time. No one needs to own French sovereign bonds, but many large institutions need to own US Treasuries.

  “Well, of course it's a problem but it's a lesser problem. Why? The dollar is the reserve currency of the world. More importantly, the entire global financial system functions on treasuries.”at 8:17

Private credit

• Steve Eisman · The Real Eisman Playbook

  Private credit is overexposed to software as AI disrupts it, possibly a harbinger of a credit cycle.

  Private credit is exposed to software, and software is being hit by AI. Eisman sees definite problems from overexposure to software in private credit. He suggests these software issues may be harbingers of an approaching credit cycle. Bank credit data remains benign, but private credit is much more opaque than bank reporting, which makes economic weakness harder to foresee. Private credit is one of the places where bad outcomes could emerge.

  “Yes, there are definite problems in private credit with respect to overexposure to software. And perhaps these software issues are harbingers of an approaching credit cycle.”at 17:25

US growth & recession risk

• Steve Eisman · The Real Eisman Playbook

  Bank credit is benign, so banks aren't flagging a recession; the risk sits in private credit or AI debt.

  Non-accruing consumer and commercial loans at JPMorgan, Wells Fargo, Citi and Bank of America show no deterioration. JPMorgan's non-accruals were $9.4B, down 7% year over year and flat quarter over quarter. Bank of America's were $5.7B, up 3% year over year and down 4% quarter over quarter. Before the GFC there were large increases in consumer non-accruals, which are absent now. Banks are well capitalized and are a concurrent indicator, so they are not leading the economy this time. Recession timing will depend on AI and private credit problems, which are more opaque than bank data.

  “Again, in my view the banks are not leading indicators this time. In the past I have said, as the banks go, so goes the economy. Not this time.”at 18:07

AI: overall outlook

• Steve Eisman · The Real Eisman Playbook

  The US economy's future hinges on AI's success, and AI debt could trigger a recession.

  Eisman argues that the potential issues in AI and in private credit will determine whether and when a recession arrives, and that banks are no longer the leading indicator. He points to the AI debt stress shown by CoreWeave's bonds and the sell-off across the AI complex as evidence that AI-related debt could be where the next problem emerges. For that reason he says the entire future of the US economy hinges on whether AI succeeds or fails.

  “Frankly I'm starting to think that the entire future of The US economy hinges on the success or failure of AI. That's why I talk about it so much.”at 18:44