A loosening of Beijing’s own restrictions on Nvidia’s H200 chips, even at a small scale, signals that Chinese authorities are prioritising domestic AI competitiveness over previous concerns about reliance on US hardware, a shift that could be read as bullish for Nvidia’s addressable market in China if it extends further. The fact that the bulk of an estimated large H200 stockpile is being kept in Hong Kong rather than moved onto the mainland, reportedly due to insufficient local capacity to support the chips, highlights the logistical and infrastructure constraints still shaping how China deploys advanced compute even when policy allows it. For Nvidia and the broader semiconductor supply chain, any signal of renewed Chinese demand access matters given how central China policy has been to swings in chip stocks this year. The move also sits against the backdrop of the US-China tech rivalry, where export controls and countermeasures on both sides continue to shape which chips can move where, and how quickly.
—
China is loosening restrictions on Nvidia’s H200 chips to help its tech giants keep pace in the AI race, the FT reports.
Summary:
- Beijing is permitting small shipments of Nvidia H200 processors to leading Chinese tech companies, according to the Financial Times
- ByteDance and Tencent have each received around 10,000 H200 chips in recent weeks
- The move is aimed at helping Chinese tech groups catch up with US AI rivals
- Beijing is directing companies to keep the majority of their H200 stock in Hong Kong, with total holdings estimated at around 500,000 units
- Hong Kong reportedly lacks the capacity to support full deployment of that stockpile
China is easing restrictions on Nvidia’s H200 chips as the country’s leading technology groups push to close the gap with US rivals in the artificial intelligence race, according to the Financial Times.
Beijing is now permitting small shipments of the advanced processors to reach top domestic tech firms, with ByteDance and Tencent each having received approximately 10,000 H200 units in recent weeks. The move marks a shift in approach from authorities who have previously restricted access to advanced US chips as part of a broader push toward technological self-sufficiency.
Despite the relaxation, Beijing is instructing companies to keep the majority of their H200 holdings in Hong Kong rather than moving them onto the mainland. Total H200 stock held in this way is estimated at around 500,000 units, a substantial stockpile that nonetheless faces a practical constraint: Hong Kong reportedly lacks the data centre and computing infrastructure capacity needed to put that volume of chips to full use.
The arrangement points to a policy that is easing in principle but still cautious in practice, allowing Chinese firms limited access to cutting-edge US hardware while keeping the bulk of that capacity offshore, at least for now. The gap between the scale of the stockpile and the infrastructure available to deploy it suggests any meaningful uplift to Chinese AI development from these chips may be gradual rather than immediate, contingent on further build-out of local capacity or a broader loosening of where the chips can be used.
The development comes amid an intensifying competition between US and Chinese firms over access to and control of the most advanced AI chips, with export controls, stockpiling strategies and questions over deployment capacity all shaping how quickly either side can translate hardware access into AI capability gains.
This article was written by Eamonn Sheridan at investinglive.com.