Burry Buries NVIDIA as Jensen Huang Hands Out GPUs Like Halloween Candy 🎃

NVIDIA unveiled a new AI compute framework on July 2, 2026, designed to help cloud providers deploy advanced hardware through revenue-sharing and credit-support agreements. The model lets the company earn from both hardware sales and cloud usage generated by the supported capacity. Early participants include Sharon AI, which plans to deploy up to 40,000 Grace Blackwell GB300 GPUs, and Firmus, which is developing a campus in Indonesia capable of supporting approximately 170,000 GPUs and 360 megawatts of power capacity. NVIDIA said the approach is intended to lower infrastructure barriers for startups, enterprises, model developers, and regional AI operators.

The launch arrives as Michael Burry disclosed a direct short position on NVIDIA at approximately $198.09 per share, according to filings reported by Business Insider. The investor, best known for predicting the 2008 housing market collapse, also established bearish positions against $TSLA, $CAT, $AMAT, and the iShares Semiconductor ETF (SOXX), describing the semiconductor index as "a pure form of overvaluation" that is "rarely seen." His thesis centers on what he views as excessive enthusiasm surrounding artificial intelligence, including the scale of investments in data centers, chips, and AI infrastructure.

NVIDIA has framed the new strategy as a response to the shift from model training to always-on token production, stating that "AI is shifting from model training to always-on token production, and that shift demands a new business model." The company is partnering with AI clouds to deploy large-scale, multi-tenant AI factories, a step intended to accelerate construction of infrastructure capable of serving inference workloads and token-intensive applications. The revenue-and-credit structure is designed to address the capital intensity that has slowed capacity expansion across the sector.

Burry's disclosure marks one of the most high-profile direct shorts against NVIDIA since the company's recent AI-driven rally, and his comments frame semiconductor valuations as disconnected from underlying fundamentals. Other investors have echoed concerns about the pace of infrastructure spending, though NVIDIA continues to report growing demand for its GPUs and networking products. The combination of an aggressive expansion push from NVIDIA and an aggressive short from Burry has sharpened the debate over how sustainable current AI-driven valuations can be.

The new framework keeps NVIDIA competing for both hardware revenue and downstream compute margins, a dual exposure that some analysts say could reshape how cloud AI capacity is financed. Burry's positions against $NVDA, $TSLA, $CAT, $AMAT, and $SOXX remain live in his latest 13F-related disclosure, and market participants are watching for any follow-up filings that signal changes in the size or composition of those bets.