How OpenAI Could Cause a US Recession With Steve Eisman

PodcastThe Compound and FriendsSteve EismanOct 16

The discussion weighs AI's growth against its vulnerabilities: high enterprise costs, uncertain revenue figures, limited customer lock-in, and reliance on a small group of buyers and infrastructure builders. Steve Eisman warns that an OpenAI bankruptcy or an AI price war could damage the economy, while also questioning whether regulation would prevent misconduct. Other speakers point to financing and construction constraints, concentrated market exposure, and mounting pressure on housing affordability.

Artificial Intelligence AI Infrastructure Recession Risk Market Valuations Risk Management Real Estate

Preview · 17 more points in the app

Read the Full Summary

Key Takeaways

The White House AI proposal described by the host would make companies liable when their inventions cause harm, but Steve Eisman questions whether executives would face consequences and argues that meaningful rules may emerge only after a crisis. He also opposes government stakes in AI firms, warning that public investment could encourage a too-big-to-fail expectation.

Enterprise AI use can be costly: Steve Eisman says companies ordered workers to use AI without regard to cost, and Uber reportedly exhausted its AI budget in three months. At the same time, a speaker says investment banks expect to increase AI-related budgets and hiring in 2027, suggesting that financing constraints may affect who builds capacity more than whether enterprise demand exists.

AI infrastructure expansion faces both construction and financing constraints: local opposition could slow data-center builds, while CoreWeave's widening spreads and yields reflect the debt needed to fund projects. A speaker says that if CoreWeave and Nebius cannot raise debt, hyperscalers such as Amazon, Alphabet, and Microsoft could benefit, though a price war could limit that advantage; the host also cited a report that CoreWeave was targeting insurance capital.

GEO-POLITICS

AI Liability — The host describes a White House AI proposal that would create a commission of industry participants and hold companies liable when their inventions cause harm. The speakers discuss whether executives who knowingly fail to prevent harm should also face consequences; Eisman says companies would be liable but questions whether executives would.

“And what it sort of sounded like was it's every man for himself, but there'll be a commission…”—

Crisis-Driven Regulation — Steve Eisman argues that meaningful AI rules may emerge only after a crisis. He doubts regulation will prevent misconduct, citing the lack of consequences for executives after 2008.

“I think we have to have a crisis. And then in the aftermath, maybe there'll be some rules. That's a 100% the way I think that…”—

TECHNOLOGY

Enterprise AI Costs — Steve Eisman says AI costs rose after companies told workers to use the technology without regard to cost. Uber reportedly spent its entire AI budget in three months, illustrating the expense of intensive enterprise use.

“The hell with the cost. And then cost, the price went up, and Uber went through its entire AI budget in three months.”—

Weak Customer Lock-In — Steve Eisman argues that AI models lack durable customer lock-in because users can readily switch providers. He contrasts this with the lasting loyalty he associates with Google and says easy switching could make it harder for AI providers to defend market share.

“One model is good one day and then people switch to the next. It's not like Google. Everybody was on Google forever…”—

OpenAI Revenue Figures — The speakers question OpenAI's reported annualized revenue figures: the company cited $50 billion at the end of September, after saying a month earlier that the figure was $70 billion. They note the figures may not be comparable and that annualized revenue is unclear, limiting conclusions about the apparent change.

“I honestly don't know. I don't think that core weave is important enough to sync the AI story. I think Anthropic and OpenAI are important enough…”—

ECONOMY

AI Collapse Risk — Steve Eisman says an OpenAI bankruptcy could push the US economy into recession immediately, but distinguishes that risk from a financial-system crisis on the scale of the Great Financial Crisis. He says the banks are fine and later says a bad recession could occur if AI falls apart. He also notes that OpenAI and Anthropic made commitments to many counterparties.

“I think The US economy would go into a recession immediately. You do? I do. It's not the same thing as a great financial crisis. The banks are fine.”—

AI Price War Fallout — In a hypothetical full-blown AI price war, Steve Eisman says Anthropic's revenue growth could turn negative, funding could dry up, and the company could have to cut commitments. He also says Oracle could be badly affected because half its backlog is from OpenAI.

“Anthropics revenue growth turns negative. Nobody's going to fund them. They got to cut back on their commitments…”—

MARKETS

CoreWeave Debt Risk — The host describes CoreWeave's widening spreads and yields as emblematic of AI infrastructure companies that need to issue debt to fund projects. The speakers say these projects cannot simply be left half-built without serious consequences.

“So what you're looking at here are spreads and yields blowing out on Core Weave, just sort of like as emblematic of all of these companies that need to sell…”—

NVIDIA Customer Risk — Steve Eisman points to customer concentration as a vulnerability in NVIDIA's growth: despite revenue increasing by more than 100% in its latest quarter, 70% of accounts receivable came from five customers. He says the company could lose substantial business if one of those customers went away.

“So NVIDIA grew revenue this most recent quarter by over a 100%. That's an astonishing statement. The largest company on planet earth grew revenue by over a 100%…”—

Western Digital Drawdown — The host says Western Digital was in a 54% drawdown while still up 100% year to date, illustrating the scale of its preceding run-up and subsequent decline. The speakers do not identify a cause for the drawdown.

“Yesterday you had the memory stocks, which were the belle of the ball, killed. Fun fact, Western Digital, it's in a 54% drawdown. 54%? 54…”—

FORECASTS

Compute Market Correction — Steve Eisman agrees that hyperscalers' positions have strengthened, but argues that the AI compute market remains vulnerable because demand is concentrated in Anthropic and OpenAI. He says a price war could cause the market to correct, especially before other companies can take up the compute.

“The hyperscalers position just strengthens. I agree that the hyperscaler position has strengthened…”—

Steve Eisman And AI Funding

After introducing Eisman, the hosts ask whether AI financing can fail; he distinguishes a company bankruptcy from a systemic banking crisis but warns that OpenAI failing could trigger a U.S. recession.

AI Revenue Concentration

Eisman argues that AI infrastructure depends heavily on a few buyers, with NVIDIA, hyperscalers, and Oracle exposed to a small number of AI lab customers.

Token Spending And Growth

The panel examines market losses and slowing AI revenue growth, with Eisman suggesting the end of aggressive enterprise token spending is beginning to pressure AI labs.

Data Center Constraints

They discuss local opposition, financing, and construction limits on data centers, while Eisman emphasizes that continued demand does not protect AI companies from a price war.

AI Price War Risks

Eisman lays out how a price war could hurt AI lab growth, undermine OpenAI-linked cloud commitments, and expose Oracle to a sharp reversal in its backlog expectations.

Rates And AI Market Swings

The hosts trace volatile AI-related trading and falling yields to concerns that debt-funded infrastructure spending may slow, rather than to a sudden improvement in fiscal conditions.

Insurance Risks And Oversight

The discussion turns to insurance-linked funding and Mark Walter’s transactions, then considers whether policyholders face risks and whether regulators can prevent harmful AI conduct.

AI Giants And IPO Timing

Eisman opposes government stakes in AI companies, compares the industry’s growth path to the internet’s early years, and says Anthropic may need to go public before growth weakens.

AI Trade And Housing

Eisman says AI dominates market performance and warns of a possible recession if the sector falters; the conversation then focuses on housing affordability and pressures pushing home prices lower.

Crisis Calls And Recession

Eisman rejects the lure of being celebrated for predicting calamity, recalls the personal strain after the financial crisis, and says an AI downturn could cause a bad—but different—recession.