Nvidia CEO Visits China After Trump Admin Restricts AI Chip Sales

Jensen Huang arrived in Beijing at the invitation of the China Council for the Promotion of International Trade, according to Chinese state-owned television.

Nvidia CEO Jensen Huang delivers a keynote address at the Consumer Electronics Show (CES) in Las Vegas, Nev., on Jan. 6, 2025. Patrick T. Fallon/AFP via Getty Images

The CEO of tech giant Nvidia is in Beijing this week to shore up regional trade ties following the unveiling of stringent new export requirements on one of its key chips for the Chinese market.

Jensen Huang arrived in Beijing at the invitation of the China Council for the Promotion of International Trade, according to Chinese state-owned television, where he expressed hope that the company could continue to cooperate with China.

Huang’s visit comes just days after the Trump administration announced new export controls would be applied to Nvidia’s H20 AI chip, the only one that the company offered in China without a special license.

The H20 chip—a watered-down version of Nvidia’s H100 chip—was created in 2022 to comply with the Biden administration’s restrictions on technology exports to China at the time.

Nvidia has said that it expects to take a roughly $5.5 billion hit from the new restrictions, due to covering charges related to inventory and purchase commitments.

The company’s AI chips have been a key focus of U.S. export controls as officials have moved to keep the most advanced chips from being sold to China, as the United States tries to ensure national security and keep ahead in the race to AI dominance.

While the H20 has a reduced core count, which makes it lower performing, it is still able to deliver some key AI capabilities, including in the domain of inference, the process by which a trained AI model draws conclusions from data it hasn’t encountered before.

That has led to scrutiny that the chip could be used to build supercomputers if gathered in great enough numbers.

To that end, a report released this week by the Select Committee on the Chinese Communist Party (CCP) alleged that Chinese AI company DeepSeek was able to build its well-known chatbot in part by using 30,000 H20.

What’s more, the select committee said, DeepSeek had ordered thousands more chips, which it would now be ineligible to purchase.

Select Committee Chair John Moolenaar (R-Mich.) and Ranking Member Raja Krishnamoorthi (D-Ill.) likewise sent a letter this week to Huang suggesting that DeepSeek’s access to H20 processors—as well as some export-controlled chips—was in part due to smuggling from nearby nations like Singapore.

According to the report, China’s share in generating sales for the company has plummeted from more than 25 percent to less than 15 percent since 2021.

However, there is the possibility that regional sales to Southeast Asia are diverted to China or otherwise sold on a secondary market.

As such, Moolenaar and Krishnamoorthi demanded that Nvidia divulge the information of any entity that had purchased more than 500 AI-related chips since 2020, as well as copies of communications the company had with DeepSeek.

Though Nvidia’s future in China appears to be dwindling, the company has announced major initiatives to pivot some of its most critical manufacturing capabilities to U.S. soil.

Huang announced on April 14 that Nvidia will produce its artificial intelligence supercomputers in the United States for the first time, and had commissioned more than 1 million square feet of manufacturing space to build and test its products in Arizona and Texas.

The company’s investment in new U.S. infrastructure, Huang said, would come out to more than half a trillion dollars over the next four years, including help from international partners such as Taiwan Semiconductor Manufacturing Company.