On September 6, according to a report obtained by Reuters, China's leading universities and research institutions have been relying on American computing chips to power their artificial intelligence (AI) technology, but the United States has now restricted the export of these chips to China .

U.S. chip designer Nvidia said last week that U.S. government officials had ordered it to stop exporting its A100 and H100 chips to China. Rival AMD has also said new licensing requirements now prevent the export of its advanced artificial intelligence chip, the MI250, to China.

A Reuters review of more than a dozen public government tenders over the past two years shows that some of China's most strategic research institutions are heavily reliant on Nvidia's signature A100 chip.

Tsinghua University, China's highest-ranked higher education institution in the world, last October spent more than $400,000 on two Nvidia AI supercomputers, each powered by four A100 chips, according to one of the tenders.

In the same month, the Institute of Computing Technology under the Chinese Academy of Sciences spent about $250,000 on the A100 chip.

In July, the Institute of Artificial Intelligence at the University of Chinese Academy of Sciences also spent about $200,000 on high-tech equipment, including servers partly powered by the A100 chip.

In November, the School of Cyber ​​Security at Jinan University spent more than $93,000 on an NVIDIA AI supercomputer, while its School of Intelligent Science and Engineering spent nearly $100,000 on eight A100 chips last month.

Research institutes and universities backed by municipal governments in Shandong, Henan and Chongqing also bought the A100 chips, the tender showed.

None of the research departments responded to a request for comment on the impact of the A100 export ban on its project, the report said.

replacement product

The lack of chips from companies like Nvidia and AMD could hinder Chinese companies from cost-effectively doing the advanced computing needed for tasks like image and speech recognition.

Image recognition and speech processing are common in consumer applications, such as smartphones that can query and tag photos. Additionally, they have military uses, such as searching satellite imagery of weapons or bases and filtering digital communications for intelligence-gathering purposes.

Few Chinese chipmakers could easily replace such advanced chips from Nvidia and AMD, but buyers could instead use multiple low-end chips to replicate the processing power, experts said.

Some Chinese government tenders also show reliance on a range of U.S. chip technologies, the report said.

Among them, a tender in May showed that the China Academy of Surveying and Mapping, a research arm of the Ministry of Natural Resources, was considering using Nvidia's artificial intelligence supercomputer to improve its ability to create three-dimensional images from geographic data.

"The proposed NVIDIA DGX A100 server will be equipped with 8 A100 chips and 40GB of memory, which will greatly improve the data carrying capacity and computing speed, as well as shorten the scientific research process, so as to obtain scientific research results faster and better," reads the tender.

The National University of Defense Technology is also one of the buyers of the A100 chip, the report said.

A tender in May revealed that the institute plans to buy 24 Nvidia graphics processing units with AI applications. But when the tender was published again last month, it said it had yet to find a suitable deal or supplier.