The H20 chip was designed to comply with previous U.S. restrictions but was swept up in a wider export crackdown in April. Nvidia, warning that the U.S. could lose AI leadership to Chinese firms like Huawei, launched an aggressive lobbying campaign to resume sales. That effort included direct talks between CEO Jensen Huang and President Trump.

The U.S. has now signaled its willingness to approve export licenses for the H20, and Nvidia said it hopes to begin deliveries “soon.” Huang is currently visiting Beijing to update Chinese partners and seek high-level meetings, including with Premier Li Qiang — a sign of Nvidia’s rising role as a diplomatic actor in the U.S.-China tech rivalry.

New Chips, New Rules

In tandem with the H20 license breakthrough, Nvidia unveiled plans to release a new GPU for China based on the RTX Pro 6000 platform. The chip will comply with updated U.S. export rules but is designed to meet Chinese demand for AI and industrial applications. Huang emphasized that “every civil model should run best on the U.S. technology stack.”

Despite Beijing’s efforts to promote domestic AI chips from Huawei, Cambricon, and Biren, Nvidia’s ecosystem remains the dominant choice due to its maturity and developer accessibility. Demand for AI chips surged in China following the release of DeepSeek’s R1 model earlier this year.

Nvidia’s maneuver reflects a broader trend: navigating regulatory pressure while protecting market share in one of the world’s largest tech economies.