Beijing/San Francisco, December 22, 2025 – Nvidia is set to begin shipping its powerful H200 artificial intelligence chips to China by mid-February 2026, marking a significant easing of US export controls on advanced semiconductors. Sources familiar with the matter told Reuters that the company has notified Chinese clients of plans to deliver 5,000 to 10,000 chip modules—equivalent to 40,000 to 80,000 individual H200 chips—from existing stock ahead of the Lunar New Year.
The H200, part of Nvidia’s previous-generation Hopper lineup, is approximately six times more capable than the downgraded H20 chip previously available to Chinese buyers. Major firms like Alibaba Group and ByteDance have shown strong interest, driven by demand for advanced AI training and inference capabilities.
However, shipments remain contingent on approval from Chinese authorities, who have yet to greenlight purchases. Beijing is weighing options, including requiring bundling with domestic chips to bolster local manufacturers like Huawei.
US-China Tech Tensions: A Primer for New Audiences The US and China have been locked in a technology rivalry since 2022, when the Biden administration imposed strict export controls on advanced AI chips and semiconductor equipment. Washington cited national security concerns, aiming to prevent China from using cutting-edge technology for military advancements or supercomputing. These restrictions forced Nvidia to develop compliant, less powerful versions like the H20 for the Chinese market. China responded by accelerating domestic chip development and occasionally restricting purchases of US products over security fears.
Policy Shift Under Trump Administration The latest development reflects a reversal under President Donald Trump, who recently approved H200 sales with a 25% fee. This move prioritizes trade and US corporate interests while addressing concerns that strict bans were boosting Chinese competitors. Nvidia plans to ramp up production in 2026, potentially opening new orders in the second quarter.