"The Big Short" Prototype: Shorting AI May Be the Right Call, but Timing Remains Unknown
N.R. Finch
Steve Eisman, one of the real investors behind *The Big Short*, says AI's biggest risk is not technology but the business model — a price war or margin erosion could force a repricing, yet nobody can call the timing, making an early short potentially devastating.
Why does he say the risk isn't about technology?
Eisman's core argument: markets assume AI will keep growing fast, and capital keeps pouring in on that assumption.
The fragile part, he says, is not whether AI works — it is whether anyone can actually make money from it.
This means → his worry is not "AI fails" but "AI succeeds, yet nobody profits."
What could trigger a repricing?
He flags two catalysts: a price war among leading model providers, or sustained margin deterioration at the top AI companies.
In plain terms = if model companies slash prices to grab users and revenue cannot cover costs, valuations across the entire AI supply chain must be reset.
This reflects a deeper concern: today's high AI valuations rest on the expectation of future earnings, not on earnings already delivered.
If the thesis may be right, why not short AI now?
Eisman himself stresses: the real question is not whether the risk materializes, but when.
This means → even if shorting AI proves correct in the end, betting too early can still produce massive losses — markets can stay at "unreasonable" levels far longer than a short position can survive.
In plain terms = right direction, wrong timing hurts just as much as being wrong. That is the classic short-seller's dilemma.
Content is for reference only, not financial advice.