Burry Warns AI Bubble May Burst Sooner Than Expected, Shifts to Leveraged Put Options
nashnova research
Michael Burry, famed for his subprime short, disclosed in his latest investor letter that he is converting AI-related short positions into put options, arguing the AI bubble may burst earlier than previously expected — and that depressed volatility has made puts "relatively cheap."
Why switch from shorting to buying puts?
The core logic is compressing the timeline: Burry believes the burst window has moved forward, and a shorter horizon makes leverage more attractive.
This means → a straight short only captures the price decline, while a put option — paying a fixed premium for the right to profit from every dollar below the strike — delivers far more bang per dollar of capital.
He also notes that VIX and other volatility gauges have narrowed to unusual lows, pushing put prices down. In plain terms = the calmer the market, the cheaper the "insurance" — Burry is stocking up while it's on sale.
Which names, which strikes?
Micron: short converted to puts at a roughly $500 strike, expiring June next year.
Nebius: short converted to June-expiry puts with a strike in the low two digits.
Semiconductor ETF SOXX: short converted to puts expiring September 2027, strike just above $400.
Palantir: short and put positions consolidated and expanded into September 2027 puts, strike just above $100.
What evidence does he cite for an earlier burst?
Burry references a research report from asset manager Ares Management, which flags AI's reliance on unproven revenue and harsh legal-agreement terms.
The report's key line: "It takes just one quarter of AI revenue falling short of capex expectations for a handful of boards to decide the highest-conviction bet has shifted — and the legal documents already leave room for that decision."
In plain terms = big tech is spending massively on AI infrastructure, betting future revenue will justify the cost. If even one quarter's numbers disappoint, the confidence to keep spending could collapse fast — and the contracts already have exit ramps built in.
What is the cyclical risk in memory chips?
Burry also cites Jason Chen, CEO of Acer, who told Taiwanese media that memory-chip cyclicality will return as Chinese capacity keeps expanding.
Chen's quote: "How can there be a sustained shortage? Chinese capacity keeps growing — there is no shortage. Contract prices are fluctuating at high levels."
This reflects a crack in the AI-hardware demand narrative — if chip supply is no longer tight, the "demand exceeds supply" story propping up valuations at Micron and peers loses its footing.
Is the market proving him wrong so far?
In May, Burry compared the current market to "the final months of the 1999–2000 bubble" — yet the Nasdaq Composite has since hit fresh all-time highs.
Micron currently sits about 16% below its record; Palantir about 10% below — nowhere near a crash.
This means → Burry's expiry dates cluster between June next year and September 2027. He is betting the inflection point for AI trades arrives within that window — and actual revenue data before those expirations will be the make-or-break test.
市场有风险，内容仅供研究参考，不构成投资建议。