Nvidia Weighs Expanding H200 AI Chip Production as China Demand Surges

Nvidia is evaluating whether to expand production capacity for its powerful H200 artificial intelligence chips after demand from Chinese customers significantly exceeded current supply, according to multiple sources familiar with the discussions. The move follows a policy shift by U.S. President Donald Trump, who announced that the U.S. government would allow Nvidia to export H200 processors to China while imposing a 25% fee on such sales.

The H200, Nvidia’s second-fastest AI chip from its Hopper generation, has attracted strong interest from major Chinese technology companies. Firms such as Alibaba and ByteDance have reportedly contacted Nvidia to discuss large purchase orders, underscoring pent-up demand for high-performance AI hardware in China. Sources noted that Nvidia is leaning toward adding new production capacity to meet this demand, although discussions remain private.

Nvidia emphasized that any increase in supply aimed at licensed Chinese customers would not disrupt deliveries to U.S. clients. The company said it is actively managing its supply chain to balance global demand, especially as it prioritizes production of its newer Blackwell chips and the upcoming Rubin architecture. As a result, current H200 output remains limited.

Uncertainty persists on the Chinese side, as government approval is still required before imports can proceed. Chinese officials reportedly held emergency meetings to assess whether to allow H200 shipments, weighing the benefits of access to advanced computing power against the potential impact on China’s domestic AI chip industry. One proposal discussed would require H200 purchases to be bundled with a certain proportion of locally produced chips, reflecting Beijing’s push to support homegrown semiconductor development.

Manufactured by TSMC using its advanced 4nm process, the H200 is widely seen as the most powerful AI chip currently accessible to Chinese companies. It is estimated to deliver around six times the performance of Nvidia’s H20, a downgraded chip designed specifically for the Chinese market. Analysts say this performance gap explains the rush by cloud service providers and enterprises to lobby for relaxed restrictions.

For Nvidia, expanding H200 production presents challenges, including limited advanced manufacturing capacity at TSMC and competition from other major players such as Google. Still, with Chinese AI demand outpacing local supply, the company’s decision on capacity expansion could have significant implications for the global AI chip market.