Aswath Damodaran, the NYU Stern professor known as the “dean of valuation,” took aim this week at the lenders financing the AI infrastructure boom.

He singled out CoreWeave Inc. (NASDAQ:CRWV) as a case study in what he says is broken credit discipline.

Speaking on the Fixed + Floating credit podcast, Damodaran said lenders to AI data center companies have “lost the script” by pricing debt off equity narratives rather than current cash flows.

“Who are these lunatics who are lending money to the data centers?” Damodaran said. “People assume these private credit guys must be smart guys. No, they’re not. They’re sheep.”

The Asymmetric Trade

Damodaran’s core argument is that private credit, not equity, is the biggest loser if the AI build-out re-prices.

If the story plays out, equity holders capture the upside while lenders just get their coupon back. If it doesn’t, lenders absorb the losses. That unfair payoff structure, he said, is not being properly priced into current AI infrastructure spreads.

He drew a parallel to shale oil companies that borrowed at $120 a barrel and then struggled when prices fell to $60.

The Numbers Behind The Quote

CoreWeave reported long-term debt of roughly $22.65 billion as of the first quarter of 2026, according to its Q1 earnings report. Net interest expense climbed to $536 million in the quarter, up from $264 million a year earlier, while the GAAP net loss widened to $740 million from $315 million.

The shift in CoreWeave’s borrowing costs shows how fast lenders bought into the AI story.

The company’s 2023 debt priced at roughly 15%, the kind of rate reserved for risky equipment loans.

Three years later, in March 2026, it landed an $8.5 billion facility at an implied cost of less than 6%, with an investment-grade rating from three agencies.

The new debt is backed by long-term offtake agreements from Meta Platforms Inc. (NASDAQ:META) and Microsoft Corp. (NASDAQ:MSFT), which is what gave it the investment-grade rating.

That’s also what Damodaran means by lending against narrative.

The contracts are valuable today, but if AI demand slows or hyperscalers renegotiate, lenders are still on the hook while equity holders capture whatever upside remains.

The Polymarket Read

Prediction markets are showing some of the same anxiety. The AI Bubble Burst by Dec. 31, 2026 contract is trading at roughly 19% on Polymarket, with $2.9 million in total volume. The market spiked toward 30% in May before drifting lower into June.

Damodaran also flagged the recent SpaceX (NASDAQ:SPCX) IPO, with its $26 trillion total addressable market claim he called “fiction,” that will likely shape the pricing for OpenAI and Anthropic when those companies eventually go public.

“You can’t make interest payments with potential and promise,” Damodaran said. “You got to make it with cash flows.”

Image: Shutterstock