Eisman: AI Trade Nearing Inflection Point, Has Significantly Hedged Positions
nashnova research
Steve Eisman, the real-life 'Big Short' investor, says the AI trade is approaching a tipping point. He has shorted against the box on many positions to cut risk — but won't go net short, because AI is too deeply tied to the broader market to isolate.
What has Eisman done — and what hasn't he?
Eisman has shorted against the box on many holdings — holding both long and short positions on the same stock to lock in gains and slash overall portfolio risk.
But he is not net short on any single name. This means → he sees the inflection coming but refuses to bet on the exact timing of a break.
In plain terms = he bought insurance on the house, but didn't bet that it will collapse — a very different posture from his 2008 subprime wager.
Why is he zeroing in on Anthropic?
Eisman flagged that Anthropic's latest IPO filing (S-1) omitted Q3 revenue. He believes the number would show a clear deceleration from Q2.
The key metric: Anthropic's revenue-to-equipment-expense ratio fell from 2.08× in Q2 to 1.7× in Q3 — a notable shift in efficiency trajectory. This means → every dollar spent on compute is generating less revenue than before.
He called parts of the S-1 "extremely negative" — too few recent financial metrics for outsiders to assess the company's true health. This reflects a broader concern: frontier-model companies may be weaponizing public fear of AI to build regulatory moats while masking deteriorating fundamentals.
What did Google's earnings expose?
Alphabet shares dropped sharply after its earnings report, with core advertising revenue showing a clear pullback.
The cause is straightforward: AI search queries are displacing traditional searches, but AI-mode monetization rates are lower. Chief Business Officer Philipp Schindler acknowledged "lower monetization" in AI-mode queries despite strong volume growth.
In plain terms = more users are searching via AI, but each AI search earns less than the old kind — traffic is rising, revenue per query is shrinking.
If he sees the cracks, why not just go short?
Eisman's answer: AI is too deeply embedded in the broader market. Tech cannot be separated from equities at large or the real economy — shorting AI is logically "impossible."
This means → going short on AI effectively means going short on the entire market, an exposure far beyond any single-sector thesis.
His words: "Too many people are on the other side. We're still reading tea leaves. I think we're getting closer." This reflects a structural paradox: fundamental signals are cracking, but positioning is locked by AI's high correlation with the broader market — how and when that deadlock breaks is the key test ahead.
市场有风险，内容仅供研究参考，不构成投资建议。