Nvidia is shifting focus away from China chip sales

By Britney Nguyen

The chip maker has reportedly stopped production of its H200s for China at TSMC, and is focusing instead on Vera Rubin

Nvidia reportedly stopped production of its H200 chips for China.

Nvidia's artificial-intelligence chips have been a point of tension between the U.S. and China, and the company's latest move could signal that it's giving up the fight to sell them in China for now.

The chip maker (NVDA) has stopped production of the H200 chips that were approved by the U.S. for sale to China, according to the Financial Times.

The FT reported that the manufacturing capacity for the H200 at Taiwan Semiconductor Manufacturing Co. (TW:2330) has been reallocated to the upcoming Vera Rubin chips, citing unnamed people described as familiar with the matter.

However, the two chip platforms don't use the same technologies and are built on different process nodes, meaning the production of one chip would not necessarily impact supply of the other. While the H200 uses an advanced chip-packaging technology called CoWoS-S, Vera Rubin uses CoWoS-L, which is suited for higher-bandwidth graphics processing units. Additionally, the H200 uses a later generation of high-bandwidth memory, while Vera Rubin will use the new HBM4 technology.

The decision to stop production of the H200 chips comes as Nvidia is reportedly worried that it will continue to face hurdles from both governments that will hamper its ability to sell to Chinese customers. Following months of uncertainty over its China business, President Donald Trump granted Nvidia approval to sell its H200 chips in December, under the condition that it would give the U.S. government a 25% cut of sales. Previously, Nvidia sold a less powerful version of the chip in China called the H20 that the Trump administration banned it from selling last April.

See more: Nvidia's earnings report is just a prelude to what should really matter for investors

Despite U.S. approval to sell the more powerful chips that U.S. tech companies like OpenAI and Meta Platforms (META) have used to train their AI models, Nvidia's H200 sales have also been limited by the Chinese government, which is pushing to build a self-reliant chip industry.

On the chip maker's earnings call last week, Nvidia CFO Colette Kress said the company has "yet to generate any revenue" from China despite having U.S. approval for some shipments. She added that she wasn't sure if any imports would be allowed by China.

In addition, Kress issued a sort of warning that the company's Chinese competitors are gaining an edge following a recent string of initial public offerings by Chinese chip companies that "have the potential to disrupt the structure of the global AI industry over the long term." She said that Nvidia will continue working with the governments of both countries.

Read: 'China's Nvidia' shows that the global chip race is heating up as it basks in post-IPO glow

Nvidia declined a request for comment from MarketWatch.

Meanwhile, Nvidia CEO Jensen Huang said earlier this week that the company's $30 billion investment in OpenAI's $110 billion funding round in late February "might be the last time" the chip company invests in OpenAI, as he expects the AI startup to go public soon. Huang said the $100 billion investment that Nvidia announced with OpenAI in September is likely "not in the cards," during a talk at the Morgan Stanley Technology, Media & Telecom Conference, CNBC reported.

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

03-05-26 1016ET