Tuesday, September 27

US AI chips banned by analysis are in high demand in Chinese state institutes

By Eduardo Baptista

BEIJING (Reuters) – Top universities and state-run research institutes in China are relying on a U.S. computer chip to power their artificial intelligence (AI) technology, but whose export to the country Washington is now limited, according to a study by Reuters.

US chip designer Nvidia Corp said last week that US government officials had ordered it to stop exporting its A100 and H100 chips to China. Local peer Advanced Micro Devices Inc (AMD) also said new licensing requirements now prevent the export to China of its advanced AI chip MI250.

The development signaled a major escalation in a US campaign to thwart China’s technological capability as tension mounts over the fate of Taiwan, where chips for Nvidia and nearly every other major chip company are made.

China views Taiwan as a rogue province and has not ruled out force to bring the democratically-ruled island under its control. In response to the restrictions, China called them a futile attempt to impose a technological blockade on a rival.

A Reuters review of more than a dozen publicly available government tenders over the past two years indicated that among some of China’s most strategically important research institutes, there is strong demand – and a need – for A100 chips from Nvidia.

Tsinghua University, the world’s top-ranked Chinese institution of higher learning, spent more than $400,000 last October on two Nvidia AI supercomputers, each powered by four A100 chips, according to one of the appeals. offers.

In the same month, the Institute of Computing Technology, part of a leading research group, the Chinese Academy of Sciences (CAS), spent around $250,000 on A100 chips.

A CAS university’s school of artificial intelligence in July this year also spent about $200,000 on high-tech equipment, including a server partly powered by A100 chips.

In November, Guangdong-based Jinan University’s Cybersecurity College spent more than $93,000 on an Nvidia AI supercomputer, while its School of Intelligent Systems Science and Engineering spent nearly $100,000. dollars for eight A100 chips last month.

Lesser-known institutes and universities supported by municipal and provincial governments, such as in Shandong, Henan and Chongqing, have also purchased A100 chips, according to the tenders.

None of the design firms responded to requests for comment on the effect on their designs of the A100 export curb.

Nvidia did not respond to a request for comment. Last Wednesday, it said it recorded $400 million in Chinese sales of the affected chips this quarter, which could be lost if its customers decide not to purchase alternative Nvidia products. He also said he plans to seek exemptions from the new rules.


The lack of chips like Nvidia and AMD is likely to hamper Chinese organizations’ efforts to cost-effectively perform the kind of advanced computing used for tasks such as image and speech recognition.

Image recognition and natural language processing are common in consumer apps such as smartphones that can respond to queries and tag photos. They also have military uses such as searching satellite images for weapons or bases and filtering digital communications for intelligence gathering purposes.

Experts said few Chinese chipmakers could easily replace these advanced Nvidia and AMD chips, and buyers could instead use multiple low-end chips to replicate the processing power.

Reuters could not locate any Chinese government tenders mentioning the other two restricted chips – Nvidia’s H100 and AMD’s MI250.

But some of the tenders showed, for example, chip purchases from US tech firm Intel Corp and bids to buy less sophisticated Nvidia products, underscoring China’s dependence on a range of US chip technologies. .

A tender in May showed that the Chinese Academy of Surveying and Mapping, a research institute under the Ministry of Natural Resources, was considering an Nvidia AI 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 with 40GB of memory, which will greatly improve the data carrying capacity and computing speed, shorten the scientific research process, and achieve scientific research results faster. and better,” reads the RFP. .

The National University of Defense and Technology (NUDT), which describes itself as a “military university” and “under the direct guidance of the Central Military Commission”, China’s highest military body, is also among the buyers of A100 chips.

The NUDT, home to Tianhe-2, one of the world’s most powerful supercomputers, has been on a US blacklist since 2015 over national security concerns, eliminating the university’s access to the Intel processors it uses in its supercomputers.

A May tender showed the institute planned to purchase 24 Nvidia graphics processing units with AI applications. The RFP was re-issued last month, indicating that NUDT had yet to find the right deal or supplier.

NUDT did not respond to a request for comment.

(Reporting by Eduardo Baptista; Additional reporting by Josh Horwitz; Editing by Miyoung Kim and Christopher Cushing)