Aswath Damodaran (NYU) | Private Credit Is the Biggest Loser When AI Corrects | #15

He argues that entrepreneurs and their funders systematically overestimate their odds of winning, so each individual story can look internally consistent even though the combined expectations for the whole space exceed what the market can support; the correction comes later when the real market size becomes clearer and some firms “drop off the map.”

He says picking AI winners is inherently a low-odds game where being wrong most of the time is normal, and the investor’s job is to assess which teams are least likely to overreach and most willing to admit mistakes—while accepting frequent misses.

He emphasizes that attaching multiples and leaning on comparables is pricing, not valuation, and that real valuation requires confronting uncertainty about business economics—starting with what AI will do (tools vs replacement), because that determines the total addressable market and the entire cash-flow path.

He argues debates about GPU depreciation are largely irrelevant to valuing young AI businesses, and reflect “lazy” multiple-based thinking; he says the real work is understanding unit economics—what is produced, what it costs, and how that becomes cash flows.

He says the biggest damage from an AI “delusion” unwind is likely in private credit financing data centers and similar buildouts, because lenders have limited upside, and lending based on equity narratives or market caps rather than current cash-flow capacity is historically dangerous.