China Is Racing to Exploit an Nvidia Loophole Before the Trump Admin Slams It Shut

For years, Nvidia has been the bedrock of the global AI industry, commanding an estimated 90% of the market for advanced AI chips. That foundation has been badly shaken by Thursday’s resignation of Jensen Huang, the man who cofounded Nvidia and served as its CEO since 1993, following revelations of his extramarital affairs with a Chinese spy and, according to the New York Times, a fellow Nvidia executive. Now Chinese AI companies are reportedly rushing to secure supply of much-needed Nvidia chips before the house comes crashing down around them, cutting off their supply to the computing resources on which much of their AI infrastructure depends. Fearing a tightening on U.S. export controls in the aftermath of Huang’s resignation, Chinese tech firms are turning to third-party cloud providers in other countries, Reuters reported on Thursday. Citing three anonymous sources, the outlet reports that one major Chinese cloud provider (also not identified by name) has held internal discussions around whether it should reserve Nvidia-powered compute from data center providers in other countries in an effort to maintain access to advanced chips despite potential new U.S. restrictions. Other Chinese companies have reportedly already been in contact with overseas cloud services to discuss the same subject. Alice Huang, the suspected Chinese spy whom Nvidia’s former CEO reportedly had an affair with, is a former executive at Megaspeed, a Singapore-based cloud provider suspected by some U.S. government officials of illegally funneling Nvidia chips to China. (The company has denied the allegations, while the Chinese government has denied that Alice Huang was an intelligence agent working on its behalf.) On Wednesday, both Axios and The New York Times reported on growing tensions within the Trump administration over why Megaspeed had not been officially banned from receiving shipments of Nvidia chips. According to Axios, Treasury Secretary Scott Bessent confronted Commerce Secretary Howard Lutnick about the issue, with Lutnick promising to look into it. During a subsequent Oval Office meeting, Lutnick reportedly warned that tightening restrictions would “start a trade war with China.” The Commerce Department has denied that the confrontation between Bessent and Lutnick occurred. There are many unanswered questions about Lutnick’s apparent reluctance to blacklist Megaspeed from receiving Nvidia chips, just as there are around why the Commerce Department reportedly halted its months-long investigation into Megaspeed after being contacted by Nvidia’s general counsel, and why Trump reportedly took no action after being notified last year of Huang’s involvement with a woman suspected of being an intelligence asset working on behalf of Beijing. Lutnick’s Commerce Department has also been criticized by some within the Trump administration for not doing enough to prevent Chinese companies from getting ahold of Nvidia chips through illegal side doors. But it now appears to be changing its tune: The Information reported Wednesday that the Department is finalizing a new rule which would make it more difficult for Chinese AI developers to exploit the remote access loophole. If implemented, the rule would require local cloud providers in Singapore, Malaysia, Thailand, Brazil, Indonesia, Norway, and South Africa to obtain a license from the U.S. government before receiving Nvidia’s Blackwell chips or more advanced models. The new rule would also require companies to submit reports to the Department outlining exactly who will receive the chips and what they’ll be used for. U.S.-operated data centers would be exempt from the licensing requirements, provided they block remote access for customers in restricted countries like China. Nvidia has previously opposed the rule in its lobbying efforts, arguing that blocking remote access would backfire on the U.S. by forcing China to spearhead its own domestic chipmaking efforts, and potentially get a leg up in the AI race.