China Approves Purchases of Nvidia’s H200 Chip, Easing Tension With U.S.

Beijing has given the green light to Alibaba and others, but is keeping in place limits intended to encourage purchases of local chips.

SINGAPORE—China has approved purchases of Nvidia’s popular H200 artificial-intelligence chip for the first time, giving authorization to several of Nvidia’s Chinese customers, people familiar with the matter said Wednesday.

The long-awaited move came during a trip to China by Jensen Huang, chief executive of the American chip giant. (Flie photo/Photo by JUSTIN SULLIVAN / GETTY IMAGES NORTH AMERICA / AFP) (AFP)

The long-awaited move came during a trip to China by Jensen Huang, chief executive of the American chip giant. The Trump administration said late last year that Nvidia could sell the AI chip to Chinese companies, but it had been uncertain whether Beijing would allow the sales to go through.

The first approval covers several hundred thousand H200 chips, which would be worth around $10 billion, the people said. Major technology companies including Alibaba and ByteDance received the initial batch of approvals, and authorities are expected to greenlight more imports in coming weeks, they said.

Beijing’s decision is another sign of its recent rapprochement with Washington on some issues ahead of a visit by President Trump to China scheduled for April. The two superpowers reached a trade truce in October.

Companies that wish to buy the American chips recently submitted documents to Chinese authorities to explain how they plan to use them. Chinese officials have told the companies that any purchases should be for uses deemed necessary, such as advanced AI research and development, The Wall Street Journal has reported.

Some companies have also discussed with officials their plans to purchase domestic chips, the people said. Officials have required that companies use homegrown chips for some AI training tasks and most AI workloads, involving inference—when AI taps its training to produce output such as a chatbot’s answers.

Beijing wants to help the country’s best AI developers build new models and applications quickly, and Nvidia chips would advance that goal. On the other hand, Beijing has worked for years to cultivate a self-reliant semiconductor industry. That has encountered resistance, especially from private-sector tech companies that are accustomed to using Nvidia’s products and software tools. Chinese chip makers also haven’t been able to match the best chips the U.S. produces.

The latest policy reflects those competing goals: Beijing is allowing some Nvidia AI chip imports while imposing limits to encourage Chinese companies to buy local chips.

Huang, the Nvidia CEO, regularly visits China around this time of the year, ahead of Lunar New Year celebrations, to meet local employees and business partners.

Huang on a visit to Beijing last year.

Since Friday, Huang has visited Shanghai, Beijing and Shenzhen, three Chinese cities where Nvidia has offices. He has been seen touring a grocery market and delivering kumquats—a citrus fruit that represents prosperity in Chinese culture—to local employees at a company party.

However, he hasn’t met with any senior Chinese officials, according to people familiar with his trip. Last year, Huang visited China at least three times and met with Vice Premier He Lifeng.

The sales of the H200 mark the breaking of a logjam that began last April, when the U.S. initially banned sales to China of H20 chips, a less-powerful cousin of the H200 that Nvidia designed for the Chinese market. Washington later reversed course on the H20 chips but Beijing then stepped in and told companies not to buy them. It flagged cybersecurity issues, which Nvidia said didn’t exist.

Following an October meeting between Trump and Chinese leader Xi Jinping in South Korea, Washington cleared the way for the H200, which is more powerful than any of the chips Nvidia had previously been allowed to sell in China. The U.S. government required Nvidia to make sure there was sufficient supply of AI chips in the U.S., and its customers have to demonstrate adequate security procedures before chips are shipped.

Huang said in October that the company’s share in China’s market for the graphics processing units that are used for AI had fallen to zero from 95% while it waited for the two governments’ actions.

Chinese competitors have raced to fill the void. Companies are using technology such as bundling a large number of less-powerful chips to boost computing power. AI developers have managed to develop decent AI models on homegrown chips, fueling recent rallies in Chinese semiconductor and AI stocks.

Some AI researchers, however, are more pessimistic and have warned that limited access to chips including Nvidia’s could widen the AI gap between China and the U.S.

Huang plans to travel later to Taiwan, also part of his annual routine, where he expects to speak to suppliers about manufacturing more H200 chips to feed Chinese demand, people familiar with his trip said.

Write to Raffaele Huang at raffaele.huang@wsj.com