State-run research institutes and high-profile universities in China have been depending on a U.S. computing chip to power their artificial intelligence (AI) technology, but whose export to the country Washington has now banned, according to a Reuters review.
U.S. chip designer Nvidia Corp (NVDA.O) last week said U.S. government officials have instructed it to halt exporting its H100 and A100 chips to China. Local rival Advanced Micro Devices Inc (AMD) (AMD.O) also said new license requirements now restrict export to China of its advanced AI chip MI250.
The development indicated a major escalation of a U.S. drive to hamper China’s technological capability as tension brims over the fate of Taiwan, where chips for Nvidia and almost every other major chip firm are designed.
China considers Taiwan a rogue province and has not excluded force to deliver the democratically governed island under its control. Retaliating against the restrictions, China labeled them a futile attempt to place a technology blockade on a rival.
A Reuters review of over a dozen publicly available government tenders over the last two years showed that among some of China’s most strategically major research institutes, there is surging demand – and need – for Nvidia’s signature A100 chips.
Tsinghua University, China’s top-ranking higher education institution worldwide, spent over $400,000 last October on two Nvidia AI supercomputers, each running on four A100 chips, one of the tenders showed.
In the same month, the Institute of Computing Technology, part of a leading research group, the Chinese Academy of Sciences (CAS), spent about $250,000 on A100 chips. The school of artificial intelligence at a CAS university in July 2022 also spent around $200,000 on high-tech equipment including a server partly running on A100 chips.
In November, the cybersecurity college of Guangdong-based Jinan University spent over $93,000 on an Nvidia AI supercomputer, while its school of intelligent systems science and engineering spent about $100,000 on eight A100 chips in August.
Less renowned universities and institutes supported by provincial and municipal governments, such as Henan, Shandong, and Chongqing, also purchased A100 chips, the tenders showed. None of the research departments reacted to requests for comment on the impact on their projects of the A100 export blockade.
Nvidia failed to respond to a request for comment. Last Wednesday, it said it had recorded booked $400 million in Chinese sales of the affected chips this quarter which could go down the drain if its customers decided not to purchase alternative Nvidia products. It also said it wanted to seek exemptions to the new rules.
REPLACEMENTS
The lack of chips from the like of AMD and Nvidia is expected to hinder efforts at Chinese organizations to cost-effectively execute the advanced computing used for tasks such as speech and image recognition.
Image recognition and natural language processing are popular in consumer applications such as smartphones that can tag photos and answer questions. They also have military uses such as filtering digital communications for intelligence-gathering purposes and scouring satellite imagery for bases and weapons.
Experts said there are few Chinese chipmakers that could easily replace such advanced AMD and Nvidia chips, and purchasers could instead use numerous lower-end chips to replicate the processing power. Reporters could not find any Chinese government tenders citing the other two restricted chips – AMD’s MI250 and Nvidia’s H100.
But several tenders showed, for example, chip buyers from U.S. technology firm Intel Corp (INTC.O) and proposals for purchasing less-advanced Nvidia products, highlighting China’s dependence on a range of U.S. chip technology.
One tender in May showed the Chinese Academy of Surveying and Mapping, a research institute of the Ministry of Natural Resources, looking at an Nvidia AI supercomputer to boost its ability to create three-dimensional images from geographic data.
The tender read:
“The proposed NVIDIA DGX A100 server will be equipped with 8 A100 chips with 40GB memory, which will greatly improve the data-carrying capacity and computing speed, shorten the scientific research process, and get scientific research results faster and better.”
The National University of Defense and Technology (NUDT), which calls itself a “military university” and “under the direct leadership of the Central Military Commission”, China’s main military body, is also one of the purchasers of A100 chips.
Buy Bitcoin NowThe NUDT, home of Tianhe-2, one of the world’s most powerful supercomputers, has been put on a U.S. blacklist since 2015 because of national security concerns, ending the university’s access to the Intel processors it uses in its supercomputers.
One May tender showed the institute wanted to purchase 24 Nvidia graphics processing units with AI applications. The tender was issued again in August, suggesting NUDT had not yet found the right deal or supplier.