Michael Burry Sells Nvidia Options While Maintaining Bearish Stance on AI Stocks

Michael Burry, renowned for his prescient bets against the U.S. housing market before the 2008 financial crisis, has divested his December 2026 put options on Nvidia. This decision is part of a larger strategy to trim his portfolio, which also included exiting near-term puts on Palantir. Despite these moves, Burry's overall bearish outlook on AI stocks remains intact.

On September 9, Burry's adjustments appear strategic, akin to a chess player repositioning pieces rather than a shift in conviction. He continues to hold long-dated puts on Palantir and the Nasdaq-100 (QQQ), with short positions comprising over 21% of his portfolio, excluding put options. By choosing not to roll over his near-term options into later contracts, Burry aims to free up capital, suggesting he anticipates an “interesting market this fall.”

His critique of AI investment centers on two main concerns: the sustainability of capital expenditures in AI and the financing structures that support massive GPU purchases. Burry has labeled some of these financial arrangements as “fugazi,” indicating skepticism about their legitimacy. He also warns of the rapid obsolescence of semiconductors, arguing that today’s advanced hardware could quickly become outdated.

In response to Burry's bearish stance, Nvidia CEO Jensen Huang has countered by highlighting the rising rental rates of GPUs, which surged 22% month-over-month. Huang frames Nvidia's chips as durable assets that generate ongoing revenue, contrasting Burry's view of them as speculative investments. Huang's defense includes emphasizing the productive lifespan of GPUs, which the company claims can remain in service for four to six years.

The ongoing debate between Burry and Huang underscores significant implications for investors in AI infrastructure. The $105 billion credit support linked to Nvidia's projects raises concerns about risk concentration. If AI spending slows, Nvidia could face challenges both in revenue and credit exposure, making GPU utilization and rental economics critical metrics to monitor.