When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works .
According to NYT, of 3,800 procurement records relating to high-end chips and computing, some 500 instances were found where units of the Chinese military tried to get their hands on Nvidia hardware. The report claims "nearly every branch" of China's military, including units focused on nuclear explosive simulations, war games, and cyberattacks, was listed. In one specific instance, a cybersecurity unit sought Nvidia A100-powered AI servers for a password-cracking tool called hashcat.
The White House first put export controls on AI chips in 2022, citing fears that the Chinese PLA was using them to advance its own military research. But despite this ban, which made it illegal to ship advanced AI chips to China without a license (which is almost always denied), buying and using them wasn’t illegal in the East Asian country at that time. Because of this, many institutions, including those linked to the PLA, were still looking for ways to get these chips through other means.
Chinese universities performing military research acquired Super Micro servers with sanctioned Nvidia AI chips — public documents reveal purchases were completed in 2025 and 2026 despite US export controls
US closes loophole that allowed Chinese-owned subsidiaries located outside China to buy AI chips
Chinese Nvidia Cloud Partner procured 300 servers with banned AI GPUs worth $92 million
This demand for Nvidia chips in China meant that smuggling them was a lucrative effort. There have been multiple reports of enterprising individuals routing shipments through another country, like Singapore, Malaysia, Thailand, Taiwan, and even Japan, just to get these chips into China. Even Supermicro’s co-founder, Yih-Shyan “Wally” Liaw has been accused of smuggling $2.5 billion worth of Nvidia AI servers to China .
On the other hand, Nvidia CEO Jensen Huang has always been against export controls . He believes that the U.S. should actually allow its hardware to be used across the world so that technological advancements would be anchored on American AI infrastructure and that the bans would only backfire, as it would force Chinese companies to innovate. While it’s true that Washington’s moves made it harder for China-based firms to buy Nvidia’s chips, many were still able to find ways to get their hands on them. Furthermore, it fueled the rise of domestic chip manufacturers, which, although they were behind what American companies can offer, are slowly catching up with innovation and performance (but, sometimes, at the cost of efficiency ).
Huang also claimed that the Chinese military would avoid U.S. tech the same way that the U.S. doesn’t trust Chinese-made hardware. However, it seems that the leather-clad chief was mistaken, if we are to rely on this report. Still, Nvidia spokesperson John Rizzo told the NYT that advanced AI systems usually use at least a hundred thousand chips, so the minuscule amount shown on the procurement documents does not match that.
Whatever the case, this news is a nightmare scenario for many American politicians, who are wary of giving China and its military any advantage. Even though President Donald Trump made a 180-degree turn in late 2025 and allowed Nvidia to secure export licenses again for the H200 years after it was first banned, some Republican lawmakers are pushing for a law that would give Congress the power to control AI chip exports.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/tech-industry/SPONSORED_LINK_URL
- https://www.tomshardware.com/tech-industry/chinese-military-has-been-acquiring-nvidia-chips-even-post-washington-export-controls-research-claims-multiple-institutions-linked-to-the-pla-asked-for-nvidia-ai-chips-according-to-publicly-available-documents#main
- https://www.tomshardware.com
- Why Financial Institutions Are Converging on Transaction Foundation Models to Build Their Own Intelligence
- Intel warns it has 'a healthy dose of paranoia' over Nvidia entrance into PC market — company says RTX Spark is 'great for the market' while touting the virtues
- Lenovo Yoga Slim 7x review: Snapdragon X2 Elite makes its case
- Bambu Lab Launches Big Bed Slinger: A2L — company's 'H2S Lite' is half the cost of H2S at just $469
- SoftBank to spend up to $87 billion on French AI data centers — country offers ample nuclear grid that US sites lack
Informational only. No financial advice. Do your own research.