Business & markets
Nvidia stops selling H200 chips to China and responds to new remote computing-power rules
According to TechCrunch, Nvidia acting CEO Colette Kress said the company will no longer sell H200 chips to China, ending an arrangement involving shipment through the United States and a 25% tariff. The new rules also restrict overseas data centers from remotely serving Chinese customers, while the company has established a board committee to oversee national security and export-control compliance.
According to TechCrunch, Nvidia acting CEO Colette Kress said Tuesday that the company will no longer sell H200 chips to China. Previously, the chips entered China under an arrangement involving transit through the United States, allowing the U.S. government to impose a 25% tariff.
The report said this arrangement had drawn dissatisfaction from U.S. hardliners on China, who advocate restricting Chinese laboratories’ access to advanced computing power. Co-founder Jensen Huang has long argued that giving up the Chinese market is not beneficial to the U.S. technology industry. After Huang stepped down as CEO last week, Nvidia’s first earnings call made Chinese access to leading AI accelerators a major topic.
The United States issued new rules on Monday banning advanced U.S. chips from entering overseas data centers that remotely serve Chinese customers. Export-control lawyer Dan Fisher-Owens told TechCrunch that data centers suited to training large models that buy Nvidia chips must prove they do not serve Chinese customers, or they may turn to other chip solutions.
Investment bank Jefferies estimates that the relevant rules could reduce Nvidia’s data-center revenue by 44000000000 to 62000000000 dollars next year. Kress said there is a backlog of demand for advanced chips such as Vera Rubin and that the new rules will not affect the company’s total revenue. The report also noted that potential customers in Singapore, Malaysia, Thailand, and Indonesia could also be affected over the long term.