Nvidia’s board said on November 8 it had opened a probe into CEO and co-founder Jensen Huang. The move came after Huang admitted to an affair with a woman whom US officials say was working for Chinese intelligence.
The woman is reportedly an executive at a Singapore-based company that the United States suspects of smuggling AI chips into China. The case has put new focus on how Chinese firms still get top Nvidia chips despite US export rules.
Nvidia designs graphics processing units, or GPUs, first built to handle 3D visuals in video games. These chips can process huge amounts of data, so they are now used to train and run artificial intelligence.
Demand for the chips grew after ChatGPT launched in 2022. Nvidia’s market value has since climbed to $4.5 trillion, putting it level with Apple as one of the world’s two most valuable companies.
What the US Allows and Bans
Washington bans exports of Nvidia’s top “Blackwell” and “Rubin” GPUs to China. It cites security concerns because advanced AI can be used by the military.
This year, the US eased rules on Nvidia’s older and less powerful H200 chip. Under the deal, the United States takes a 25 percent cut of those sales.
Shipments have still been hard to get approved because of US conditions. Beijing has also reportedly been reluctant to buy H200s as its own chip industry grows.
Huang has lobbied the US government to allow more chip sales to China. He has warned that the limits are helping Chinese rivals build their own chips.
The Rental Loophole
Lian Jye Su, a chief analyst at Omdia, told AFP that Chinese vendors are still using top Nvidia chips to train their models. One way is renting computing power at data centers in other countries, often in Southeast Asia.
“US export controls restrict where a chip is shipped, not who uses it,” said Stephen Wu, founder of the Carthage Capital investment fund. He said GPUs can sit in Malaysia’s Johor region “in full compliance while a customer in Shenzhen logs in.”
White House adviser Michael Kratsios said this summer that Moonshot AI had accessed high-end Nvidia chips in Thailand. Washington has also accused Chinese AI labs of training systems to copy the abilities of US models.
Smuggling is another route. Tech outlet The Information reported in December that DeepSeek built a new model using thousands of chips taken apart overseas and brought into China.
DeepSeek did not respond to a request for comment at the time. Nvidia told The Information it had seen no evidence of such smuggling, saying the claim “seems far-fetched.”
Chinese startups such as Moonshot AI, Z.ai and DeepSeek have released strong models in recent months. Poe Zhao, founder of Hello China Tech, said Nvidia chips still make top AI work faster, “but they are no longer the only path.”
Zhao said training on Chinese chips usually takes more time and engineering. Wu said Chinese firms make up for weaker chips by using more of them, paying more in power and cost, which they can afford because electricity is cheap and the chips are state-subsidized.
Huawei is China’s top Nvidia rival, though it still trails the US company. DeepSeek will reportedly train its next major AI model on Huawei chips.
Huawei said in September that its next AI chip would be ready in the first quarter of 2027. That is three quarters earlier than it had planned.
