US expands chip export curbs to China to throttle AI, supercomputer development

The US is imposing additional export curbs to restrict China from accessing advanced chips for AI and supercomputers, expanding a technology trade war that has intensified over the last year, broadly impacting the global semiconductor supply chain.

The US — which began to impose restrictions on semiconductor exports to China in 2015, extending them in 2021 and twice in 2022 — announced Tuesday that is has added new restrictions to stop the flow of advanced chips to Chinese data centers that house supercomputers and infrastructure that supports AI development.

Included in the new rules are a worldwide licensing requirement for any company that is headquartered in China, Macau, and any destination subject to the US arms embargo, or whose ultimate parent company is headquartered in those countries, according to a statement issued by the US Bureau of Industry and Security (BIS), which operates under the Commerce Department.

BIS also said that these steps are being taken to prevent firms from countries of concern from securing controlled chips through their foreign subsidiaries and branches.

Additionally, the US is also expected to impose additional license requirements on exports to more than 40 additional countries that present a heightened risk for export diversion to China.

“This will help address potential trans-shipment by non-Chinese companies purchasing chips for resale to China and provide greater visibility into procurement networks and customers of these chips,” BIS said.

Export rules include new chip specifications

The new rules also modify and expand the list of restrictions on chip-making equipment and semiconductor tech specifications. These restrictions, first published in a draft ruling earlier this month, “are critical for preventing or limiting the further development of weapons of mass destruction, advanced weapons systems, and high-tech surveillance applications that create national security concerns, including through their use in exascale supercomputing, and artificial intelligence (AI) capabilities,” BIS said.

Notably, the ruling adds a “performance density parameter,” measured in FLOPS per square millimeter, that BIS said “prevents the workaround of simply purchasing a larger number of smaller datacenter AI chips which, if combined, would be equally powerful as restricted chips.” Performance density specifications are detailed in the ruling.

Chinese officials have protested the new restrictions and said that they restrictions are not in sync with the principles of free and fair-trade practices.

Trade war for tech supremacy intensifies

However, US lawmakers have been urging the Biden administration to take further action to impede China’s progress in areas including AI  intelligence and quantum computing.

China has hit back with restrictions of its own, banning the use of semiconductors manufactured by US-based chipmaker Micron. That, in turn, sparked US lawmakers to advocate imposing trade restrictions on Chinese memory chip maker Changxin Memory Technologies.

US allies have been dragged into the tech trade war as well. In January, the US convinced the Netherlands and Japan to join it in expanding the ban on exports of chip-making technology to China.

According to analysts, Washington’s strategy to strike a deal with the two countries was a significant move, as some of the world’s largest makers of semiconductor manufacturing equipment are headquartered in these nations.

China has urged Japan to repeal the restrictions, citing international and trade regulation violations.

The trade war has also seen the EU trying to reduce its dependence on China.

In July, the EU formed a new partnership with Japan to improve cooperation on technology issues, after recently pledging billions of dollars in investments to shore up their domestic chip industries.

In April, the European Council and the European Parliament agreed to invest $3.6 billion in EU funds to build out the bloc’s semiconductor manufacturing capabilities, with the aim of attracting a further $43.7 billion in private investment. The investments could help avoid a Taiwanese bottleneck in the world’s semiconductor supply chains.

During the same time, the Japanese government said it was planning to invest $532 million (70 billion yen) in projects to develop and make next-generation chips in the country, including a deal with Rapidus to make 2nm chips in Japan by 2025.

New chip rules further impact chipmakers

The new ruling announced Tuesday is expected to affect a number of manufacturers of semiconductors and chip-making equipment. In a filing with the US Securities and Exchange Commission Tuesday, Nvidia said the rules will affect a number of products, including A800, H800, and H100 chips as well as system built on them. The H100 is part of a family of Hopper processors that Nvidia claims will underpin the world’s fastest supercomputers, and is the go-to chip for advanced AI systems.

Response from the US tech sector has been muted. “We are evaluating the impact of the updated export controls on the U.S. semiconductor industry,” according to a statement by the Semiconductor Industry Association (SIA), which claims to represent 99% of the US semiconductor industry by revenue and nearly two-thirds of non-U.S. chip firm.

 “We recognize the need to protect national security and believe maintaining a healthy U.S. semiconductor industry is an essential component to achieving that goal,” the SIA said.  “Overly broad, unilateral controls risk harming the U.S. semiconductor ecosystem without advancing national security as they encourage overseas customers to look elsewhere. Accordingly, we urge the administration to strengthen coordination with allies to ensure a level playing field for all companies.”

In the wake of the BIS announcement, share prices of chipmakers including AMD, Intel and Nvidia declined Wednesday morning. Nvidia’s shares were hit hardest, dropping by 3.54% midday to $423.87 on the Nasdaq exchange.