Invezz

Michael Burry is shorting Nvidia and Micron, but won’t touch this AI stock

Michael Burry is shorting Nvidia and Micron, but won’t touch this AI stock
Devesh Kumar
Aug 14, 2026, 04:28 AM

powered by

Invezz
QQQ puts

Buy QQQ puts. Burry is leaning into a Nasdaq-heavy risk-off bet (he exited SOXX puts and added QQQ puts ~6% of portfolio). The news shows AI supply/demand debate is unresolved, and CoreWeave’s surge proves fundamentals alone won’t cap downside—AI can rip on execution beats, then roll over when financing/compute economics reprice. Key risk: a sustained “AI demand beats supply” regime that keeps Nasdaq bid and compresses put volatility.

Key Risk: Nasdaq keeps rallying because AI demand stays stronger than supply, making QQQ puts lose value fast.

Micron short

Sell Micron (MU) short. Burry increased his Micron short as price approached ~$1,000, and the article frames the core bearish bet as capital costs and eventual excess compute visibility (2028). Even if conditions are tight now, the thesis is that memory margins eventually mean-revert when the AI buildout matures and financing costs bite. Key risk: DRAM/NAND tightness persists through 2027 longer than expected, keeping margins elevated and forcing shorts to cover.

Key Risk: Memory supply stays tight longer than the thesis, keeping MU margins high and squeezing the short.

  • Burry keeps Nvidia and Palantir puts while raising his bearish Micron bet.
  • CoreWeave’s meme-stock profile makes it too dangerous for Burry to short.
  • Wall Street still sees tight AI supply despite Burry’s bearish AI thesis.

Michael Burry is becoming more aggressive against the AI trade, but one company he has criticised remains off limits: CoreWeave.

The “Big Short” investor has retained put options on Nvidia and Palantir, increased his Micron short and replaced losing semiconductor ETF puts with a larger bearish position in the Nasdaq-heavy QQQ.

Yet Burry says CoreWeave is too dangerous to short because its limited float, retail following and volatility can overwhelm fundamentals.

His decision is tactical rather than bullish. CoreWeave may be one of the AI companies Burry distrusts most, but it is also the one he sees as most capable of punishing bearish traders.

Burry says his short book now rivals 2020

Burry said his current bearish exposure resembles, and may be even more aggressive in some respects than, his positioning around the 2020 market crash.

His latest portfolio update shows QQQ puts representing about 6% of the portfolio after he exited losing SOXX puts.

An outright SOXX short remains his largest bearish position at roughly 7%, while he retained Nvidia and Palantir puts and increased his Micron short as the memory stock approached $1,000.

Burry has also increased cash to about 12% as partial preparation for a broader decline.

His concern stretches beyond valuations. Burry argues that AI companies increasingly finance one another and recycle capital through the same ecosystem, while growing debt introduces a real cost of capital.

He has identified 2028 as a possible point when excess compute capacity becomes more visible.

CoreWeave breaks Burry’s own bearish rule

CoreWeave would appear to fit Burry’s thesis perfectly.

Earlier this year, he criticised its debt-funded GPU spending, rapid depreciation and customer concentration, comparing the AI cloud provider unfavourably with infrastructure businesses from the dot-com era.

But when asked whether CoreWeave was an attractive short, Burry focused instead on trading risk.

“CRWV has more of a meme vibe,” he said, according to Stocktwits, adding that he would rather avoid shorting the leading candidate for “memesville.”

This week demonstrated the danger.

CoreWeave shares surged more than 19% on Wednesday after second-quarter revenue more than doubled to $2.58 billion.

Revenue backlog reached $104.2 billion, while the company raised its 2026 revenue outlook to $12.4 billion-$13.2 billion.

Bernstein analyst Madison Rezaei, previously critical of CoreWeave’s execution, acknowledged after the results that “in this quarter, they delivered.” She nevertheless retained an Underperform view, with debt and longer-term AI-compute economics still concerns.

The contradiction captures Burry’s problem: questionable fundamentals do not guarantee a falling stock.

Wall Street is still challenging Burry’s AI thesis

Burry is betting against a market where many analysts still see demand, not excess capacity, as the dominant issue.

Brad Neuman, director of market strategy at Alger, told Business Insider that investors should be “more worried about supply than demand,” pointing to constraints around data centres and grid connections.

Micron offers another sharp disagreement. Mizuho analyst Vijay Rakesh reiterated an Outperform rating and $1,375 target this week, arguing that tight DRAM and NAND conditions could persist through 2027 and support unusually high margins.

Nvidia also retains strong analyst support. Bank of America’s Vivek Arya described Nvidia’s new third-party AI infrastructure financing platform as structurally bullish because it shifts much of the capital burden away from Nvidia while reinforcing its CUDA ecosystem.