A retreat signal? "Big short sellers" liquidate Nvidia and Palantir put options.
"Big Short" Michael Burry announced he was scaling back his risk by closing out near-term put options on Nvidia and Palantir, but his overall bearish position remains intact.
According to a recent report from StockTwits , Burry posted on Substack on Wednesday that he "tightened his risk exposure" in September, reducing every position in his portfolio and currently holds some cash, "happy to hold and wait and see how the market performs this fall." He wrote, "This fall is going to be an interesting market."
The most noteworthy part of this operation is that Burry closed out his put options on Nvidia and Palantir expiring in December 2026 without rolling them over to contracts with longer expiration dates.
Closed out near-term options, but the bearish framework remains unchanged.
In his article, Burry clearly explained the reason for closing the position: to avoid the rapid time decay of contracts nearing expiration. He stated that this move was part of "reducing overall portfolio exposure," rather than a fundamental shift in his bearish stance.
According to reports, Burry still holds 2027 put options on Palantir and Invesco QQQ Trust (QQQ) and continues to short multiple technology, semiconductor and AI-related stocks.
In terms of short positions, Oracle (ORCL), Palantir (PLTR), and Nebius Group (NBIS) remain his three largest short positions, followed by Nvidia (NVDA) and the iShares Semiconductor ETF (SOXX). Nvidia ranks fourth in his short portfolio.

Long positions: Primarily in consumer and healthcare sectors, with limited allocation to technology.
On the long side, Burry also reduced his holdings, but the order of his positions remained unchanged.
His largest long positions are, in order: Lululemon Athletica (LULU) – which he called a “fat pitch” last week – followed by Molina Healthcare (MOH) and MercadoLibre (MELI).
The top seventeen long positions include Temple & Webster, Zoetis, Sprouts Farmers Market, JD.com, Birkenstock, Adobe, HCA Healthcare, Fiserv, Flutter Entertainment, Build-A-Bear Workshop, Veeva Systems, Fannie Mae, Freddie Mac, and PayPal. Burry did not disclose the specific percentage of each position in the portfolio.

Jensen Huang used data to refute Burry's depreciation logic.
Burry's core argument for shorting Nvidia lies in his belief that tech companies are overly optimistic about the lifespan of Nvidia GPUs.
Many tech companies assign a six-year lifespan to Nvidia GPUs and depreciate them accordingly—meaning they only charge one-sixth of the cost each year, thus inflating their reported profits. Burry argues that the actual lifespan of these chips should be two to three years, and that the six-year depreciation assumption artificially inflates corporate earnings.
However, real-world data is challenging this assessment. Nvidia's Ampere A100 GPU, launched in mid-2020, is still widely used. It has been reported that cloud computing company CoreWeave just signed a contract last month to lease these chips for AI processing, extending the contract until 2029—nine years after the A100's initial release.
More noteworthy is the price trend. Financial market platform Ornn Exchange published an article on its X platform this week, noting that rental rates for the newer generation Hopper H100 chip have increased by 22% month-over-month .
Jensen Huang then retweeted the post with his own comment. The Motley Fool quoted him as saying that this data "clearly refutes Burry's claim that Nvidia GPU lifespan is overestimated." Huang's implication is that these chips are not becoming obsolete prematurely; in fact, demand continues to grow.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.