CoreWeave Is Imploding — And Taking AI Stocks Down With It

Could the current AI boom be built on a foundation of unsustainable debt and collapsing prices? The conversation between Scott Galloway and Paul Kudrowskiy centers on the precarious state of “neoclouds”—specialized AI cloud providers—with a specific focus on CoreWeave. Kudrowskiy argues that CoreWeave is currently trapped in a “debt death spiral,” where the rising cost of servicing its massive liabilities is colliding with a staggering deflation in the price of AI tokens. While CoreWeave’s leadership claims the market is acting irrationally, Kudrowskiy asserts that the math is simple and brutal: when token prices drop thirteen-fold annually, a company must grow its sales by 1,300% a year just to stand still.

Galloway and Kudrowskiy delve into the systemic risk this poses to the broader AI trade. Kudrowskiy rejects the notion that NVIDIA will step in to bail out CoreWeave, despite NVIDIA’s significant investment and partnership. He posits that NVIDIA is a chip manufacturer, not the Federal Reserve, and has little incentive to restructure a failing cloud provider. Galloway expresses concern that a default at CoreWeave could serve as the first domino, shaking investor confidence in the entire AI sector and triggering a wider correction.

The discussion expands to the nature of the “AI bubble,” with Kudrowskiy arguing that the most dangerous valuations aren’t in the public markets, but in the private ones. He suggests that private markets have effectively become the new public markets, hiding nonsensical valuations that dwarf the peak of the dot-com era. Galloway adds his own skepticism regarding companies like Anthropic, noting that he is less interested in “earnings before bad things” and more concerned with extreme customer concentration and the potential plateauing of the AI coding frenzy.

Ultimately, Kudrowskiy predicts a significant correction, potentially occurring by the first half of next year. He views the current landscape as a “refinancing problem” similar to the lead-up to the 2008 Global Financial Crisis, where a wall of debt maturing between 2028 and 2032 will be impossible to refinance at current prices. Together, they paint a picture of an industry that has over-leveraged itself on the hope of infinite growth, now facing the reality of rising interest rates and diminishing returns on compute power.

Surprising Insights

• The “Stand Still” Growth Rate: Because of the rapid deflation of token prices, neoclouds may need to grow their sales by 1,300% annually just to maintain their current financial position.
• Private Market Bubble: The AI bubble isn’t primarily visible in public stock prices but is concentrated in private valuations (like OpenAI and Anthropic) that exceed the peak NASDAQ valuation of 2000.
• The 90-Day Double: To keep up with the current trend of falling token costs and rising debt, some AI infrastructure companies effectively need to double their token sales every 90 days.
• Refinancing Wall: There is a massive “wall” of debt obligations coming due between 2028 and 2032 that the market believes cannot be refinanced at current business valuations.
• NVIDIA’s Limits: Despite the interdependence of the AI ecosystem, there is little evidence or logical reason to believe NVIDIA would act as a backstop for failing neoclouds.
• Coding Plateau: The initial surge in AI adoption was heavily driven by coding; a shift to other marketplaces is required for continued growth, but this transition is not guaranteed.

Practical Takeaways

• Watch the Private Markets: Be wary of the “halo effect” where public AI stocks look stable while the underlying private infrastructure companies are heavily distressed.
• Analyze Customer Concentration: When evaluating AI companies, prioritize looking at how many customers provide the bulk of the revenue; extreme concentration is a major red flag.
• Monitor Bond Spreads: Keep an eye on the spreads of AI-related bonds over treasuries; “blowing out” spreads are often a more accurate signal of distress than equity prices.
• Question the “Irrational Market” Narrative: When executives claim the market “misunderstands” their company during a price crash, scrutinize the debt-to-revenue ratio and interest payment costs.
• Diversify Away from “Pure Play” AI: Given the risk of a systemic correction in the neocloud sector, avoid over-concentration in companies that rely solely on the current AI CAPEX frenzy.
• Prepare for a 2025 Correction: Based on the refinancing cycle and valuation gaps, maintain a defensive posture regarding AI-heavy portfolios through the first half of next year.