Nvidia’s Jensen Huang turns on the charm in Beijing amid US-China tech rivalry 

BEIJING – On a high-profile trip to China, with billions in potential sales to the lucrative Chinese semiconductor market at stake, Nvidia co-founder Jensen Huang ditched his iconic black leather jacket for a traditional Tang suit. 

At the opening of

a trade exhibition in Beijing

on July 16, the Taiwan-born American spoke a few sentences of halting Mandarin, before adding to laughter: “I’m going to do the rest of my speech in English now. So I don’t torture you for the rest of the day.” 

Describing artificial intelligence (AI) models from Chinese tech players like DeepSeek, Alibaba and Baidu as “world-class”, he said: “The heroes of China’s super-fast innovation are your researchers, developers and entrepreneurs. More than 1.5 million developers in China build on Nvidia today to bring their innovations to life.”

Nvidia is at the heart of a global AI boom, and last week became the first public company to

hit US$4 trillion (S$5.14 trillion) in stock market value

.

Mr Huang, who is among the American tech executives who are trying to court Chinese customers while not falling foul of Washington’s national security concerns, has come prepared on his third China visit of the year.

On July 15, Nvidia announced that sales

of its H20 chip to China would resume

, with Washington’s nod. The AI chip, a less powerful version of the firm’s flagship H100, had been banned in April when Washington tightened export controls to China. 

But analysts said the move to approve the China-specific chip is merely a reprieve which may not last, with technology at the heart of US-China competition and both countries still in the middle of

a 90-day trade truce

since May 12.

The H20 ban was estimated to have cost Nvidia US$10.5 billion across the period from January to July. The policy reversal came after Mr Huang lobbied US President Donald Trump at the White House last week. 

On his visit, Mr Huang also unveiled a new graphics processing unit for the China market called the RTX Pro, which he said is “fully compliant” with US export restrictions and would be designed for smart factories and robot training purposes.

“It takes 200 different technology companies to build one of our AI computers... it is not possible without a sophisticated supply chain,” he told president Ren Hongbin of the China Council for the Promotion of International Trade. “That’s why I’m here – to celebrate the miracle of the (Chinese) supply chain.”

The council, a government-affiliated trade body, is the organiser of the third China International Supply Chain Expo being held in Beijing from July 16 to 20, which Nvidia is taking part in for the first time. Mr Huang was among the speakers at the opening ceremony. 

Even as he received the VIP treatment in China, Nvidia has remained caught in the cross hairs of Washington and Beijing.

Mr Huang has argued that continuing to sell to China means that the US is not ceding the Chinese market to domestic players such as Huawei. He has also said that US tech export controls to China have failed, and have only pushed Chinese firms to innovate more quickly. 

But some, including US Commerce Secretary Howard Lutnick, have said that Nvidia and other companies should stop helping China use “our tools to compete with us”. 

Associate Professor Marina Zhang of the University of Technology Sydney saw Mr Huang’s recent remarks as an effort to position himself – and Nvidia – as both compliant with US policy and indispensable to China’s AI ecosystem. 

Prof Zhang, who researches China and technology issues, did not see the approval of the H20 as a reversal in policy, but more of a calibration. 

“It reflects a ‘precision sanctions’ approach: allowing US firms to serve China’s mid-tier AI market – keeping American players commercially viable – while still blocking access to cutting-edge compute essential for training frontier models. 

“Washington is walking a tightrope: trying to avoid a full market exit for firms like Nvidia while continuing to constrain China’s long-term AI capabilities,” she said, adding that this policy adjustment will not reverse China’s determination to develop indigenous technologies, particularly in semiconductors. 

Dr Sun Chenghao, who researches US foreign policy and US-China relations at Tsinghua University, said the US decision to resume H20 chip sales to China did not represent a fundamental reversal of export controls. 

Rather, it is a temporary concession driven by commercial pressure and geopolitical calculations. Nvidia suffered significant losses under April’s export ban, prompting Mr Huang to lobby intensely for policy adjustments, he said. 

“While Washington green-lit the sales, it strictly capped the H20’s performance (merely 15 per cent to 30 per cent of the H100’s capabilities) to prevent its use in cutting-edge AI training – revealing its unchanged core aim: delaying China’s technological advancement.” 

Dr Sun said Washington could reimpose stricter controls if China’s domestic chip substitutes progress faster than expected, or if US political pressures intensify. 

“Notably, America has explicitly tied tech export policies to broader trade negotiations, meaning export restrictions will remain a bargaining chip to pressure China if talks stall,” he added.

Mr Huang himself saw such geopolitical issues as beyond his control. When asked about the impact of US tariffs in a press conference on July 16, he said Nvidia would simply have to adapt.

“There were trade, taxes and tariffs before I came into Nvidia. There will be trade, tariffs and taxes after I leave Nvidia.”