The potential expansion follows a recent announcement by former U.S. President Donald Trump that the U.S. government would permit Nvidia to export H200 processors—its second-fastest AI chips—to China, subject to a 25% export fee.
Sources said interest from Chinese companies has been “overwhelming,” prompting Nvidia to explore ways to scale output. The company, however, emphasized that licensed sales to authorized Chinese customers would not affect its ability to supply the U.S. market. “We are managing our supply chain to ensure that licensed sales of the H200 to authorized customers in China will have no impact on our ability to supply customers in the United States,” an Nvidia spokesperson told Reuters.
Major Chinese firms, including Alibaba and ByteDance, have reportedly approached Nvidia this week with inquiries and indications of substantial orders. Despite the enthusiasm from Chinese companies, uncertainties remain: the Chinese government has not yet authorized any H200 purchases. Emergency meetings were convened by officials on Wednesday to evaluate whether the chips could be imported, according to multiple sources.
Currently, only limited quantities of H200 chips are being produced, as Nvidia prioritizes manufacturing its most advanced Blackwell series and the upcoming Rubin line. The H200, first deployed last year, is the fastest chip in Nvidia’s previous Hopper generation and is fabricated by Taiwan Semiconductor Manufacturing Company (TSMC) using a 4nm process.
The high demand in China is largely driven by the H200’s superior performance relative to local alternatives. It is approximately six times more powerful than Nvidia’s H20 chip—a scaled-down version developed specifically for the Chinese market in late 2023. Analysts say the H200 significantly outpaces domestic accelerators. “Its compute performance is approximately 2–3 times that of the most advanced domestically produced chips,” said Nori Chiou, investment director at White Oak Capital Partners. “Many cloud service providers and enterprise clients are aggressively placing large orders and lobbying for regulatory relaxations, because local production cannot meet their needs.”
The issue is particularly sensitive amid China’s ongoing efforts to boost its domestic AI chip industry. Some proposals discussed during the emergency meetings reportedly include requiring H200 purchases to be bundled with a certain proportion of Chinese-made chips, a move aimed at balancing foreign imports with local development.
For Nvidia, expanding H200 production presents its own challenges. The company is not only transitioning to the Rubin series but also competing with global tech giants, including Alphabet’s Google, for limited advanced chipmaking capacity at TSMC. How Nvidia navigates these pressures—and whether China ultimately approves large-scale H200 imports—could have significant implications for the global AI chip market.