NVIDIA Resumes China AI Chip Production Amid $1 Trillion Revenue Forecast

NVIDIA CEO Jensen Huang confirmed Tuesday that the company is restarting manufacturing of its H20 chip for the Chinese market, after receiving export licenses from the U.S. government. The announcement came during the GTC 2026 conference, where Huang also projected over $1 trillion in revenue from Blackwell and Vera Rubin chip orders through 2027 — more than double last year's $500 billion estimate. Despite the bullish outlook, NVDA stock dipped 0.70% to close at $181.96.

The H200, built on NVIDIA's older Hopper architecture and engineered to comply with U.S. export regulations, had been paused due to tightening restrictions on both sides of the Pacific. With licensing now secured, Huang confirmed production has been ramping for several weeks. "Our supply chain is getting fired up," he stated. The restart follows a December 2025 framework from the Trump administration permitting H200 exports to approved Chinese customers, provided NVIDIA remits 25% of those revenues to the U.S. government.

The move marks a significant pivot for the world's most valuable publicly traded company, currently valued at approximately $4.5 trillion, after absorbing a $5.5 billion financial blow from prior export restrictions and Chinese regulatory scrutiny. Chinese authorities had accused NVIDIA of anti-monopoly violations in September 2025 and barred foreign AI chips from state-funded data centers later that year.

Beyond China, NVIDIA introduced its next-generation Vera Rubin chips at GTC 2026, boasting 3.5 times faster training and 5 times faster inference performance than Blackwell. The company also debuted the NVIDIA Groq 3 Language Processing Unit following its $20 billion acquisition of Groq, with shipments expected in the second half of 2026. Additionally, NVIDIA announced robotaxi partnerships with Hyundai, Nissan, BYD, and Geely, targeting 18 million autonomous vehicles annually, with Nvidia-powered self-driving Ubers anticipated to launch by 2027.