Uncovering the Secret: How Chinese Entities Defy US Chip Bans for AI Advancement

Uncovering the Secret: How Chinese Entities Defy US Chip Bans for AI Advancement

January 16, 2024

Chinese military bodies, state-run AI research institutes, and universities have managed to bypass US chip bans to acquire Nvidia semiconductors, according to a Reuters review of tender documents. These purchases by largely unknown Chinese suppliers highlight the challenges the US faces in completely cutting off China’s access to advanced US chips. Despite the bans, buying or selling high-end US chips is not illegal in China, and the tender documents reveal that Chinese entities have successfully acquired and utilized Nvidia semiconductors. These include the A100 and H100 chips, which were banned from export to China and Hong Kong in September 2022, as well as the A800 and H800 chips, which were developed specifically for the Chinese market but were also banned in October.

Nvidia’s graphic processing units (GPUs) are considered superior to rival products for AI work due to their ability to efficiently process large amounts of data required for machine learning tasks. The fact that there is still a demand for and access to banned Nvidia chips underscores the lack of viable alternatives for Chinese firms, despite the development of rival products by Huawei and others. Prior to the bans, Nvidia held a 90% share of China’s AI chip market.

Interestingly, the buyers of these banned chips include elite universities and entities subject to US export restrictions, such as the Harbin Institute of Technology and the University of Electronic Science and Technology of China. These entities have been accused of having military affiliations or involvement in military matters against US national interests. For example, the Harbin Institute of Technology purchased six Nvidia A100 chips in May for deep-learning model training, while the University of Electronic Science and Technology of China purchased one A100 chip in December 2022.

It is important to note that neither Nvidia nor approved retailers were among the suppliers identified in the review. The sources of these chips procured by Chinese entities are not entirely clear. However, an underground market for such chips has emerged in China since the US imposed restrictions. Chinese vendors have claimed to acquire excess stock from the market after Nvidia ships large quantities to major US firms. They may also import through locally incorporated companies in places like India, Taiwan, and Singapore.

Both Nvidia and the US Department of Commerce declined to comment on the matter. US authorities have expressed their intention to close loopholes in export restrictions and limit access to these chips by Chinese companies located outside of China. However, experts argue that it is unrealistic to expect export restrictions to be completely watertight, as chips are small and easily smuggled. The primary aim of these restrictions, according to Chris Miller, a professor at Tufts University and author of “Chip War: The Fight for the World’s Most Critical Technology,” is to impede China’s AI development by making it difficult for them to build large clusters of advanced chips for training AI systems.

The review reveals that military and AI buyers in China have been procuring Nvidia chips through tenders. For instance, Tsinghua University, often referred to as China’s Massachusetts Institute of Technology, has purchased approximately 80 A100 chips since the ban in 2022. Several other tenders also indicate that these chips are being used for AI purposes. While the quantities of most purchases are small and insufficient for building sophisticated AI models from scratch, they can still be used to run complex machine-learning tasks and enhance existing models.

It is worth noting that many of the tender requirements specify that suppliers must deliver and install the products before receiving payment. Most universities have published notices indicating the completion of these transactions. For example, Shandong Artificial Intelligence Institute awarded a contract to Shandong Chengxiang Electronic Technology for five A100 chips, worth 290,000 yuan ($40,500), last month.

The ability of Chinese entities to bypass US chip bans and acquire banned Nvidia chips raises concerns about the effectiveness of these export restrictions. However, the US remains committed to closing these loopholes and limiting China’s access to advanced chips. As the race for AI supremacy continues, the battle for control of critical technologies intensifies on a global scale.