US weighs further restrictions on chip exports to China

The Biden administration is reportedly considering placing new restrictions on the export of AI chips to China, in a move that could specifically target products previously compliant with the rules.

Chip China

US semiconductor manufacturers such as Nvidia and AMD will face new controls on exports of their chips to China, with the US Department of Commerce set to announce the restrictions as early as July, according to a report in the Wall Street Journal.

In 2022, after the US placed new restrictions on exports to China of chips used in AI systems, Nvidia said it would offer a new export-compliant advanced A800 chip to the country, in addition to making changes to its H100 so that it would also comply with the changes to regulations.

However, new restrictions could see sales of A800 chips to China banned unless Nvidia obtains a special US export license, according to the Wall Street report, citing people familiar with the situation.

In response to the report, an Nvidia spokesperson noted that company CFO Colette Kress, speaking at an investor conference Wednesday, said that the chip maker is aware of the news that the Commerce Department is considering rules that could restrict exports of the A800 and H800 products to China.

"Given the strength of demand for our products worldwide, we do not anticipate that such additional restrictions, if adopted, would have an immediate material impact on our financial results," Kress said, according to the spokesperson.

Restrictions could curb opportunities for chip makers

"Over the long-term, restrictions prohibiting the sale of our datacenter GPUs to China, if implemented, would result in a permanent loss of opportunities for U.S. industry to compete and lead in one of the world’s largest markets and impact on our future business and financial results," Kress said.

On its part, AMD earlier this month unveiled its new Instinct MI300X processor, which it claims can do the work of multiple graphics processing units (GPUs) and is, according to CEO Lisa Su, the most complex product  the company has ever built.

Alongside the new processor, AMD also launched the Instinct platform, a server reference design based on specifications from the Open Compute Project that uses eight MI300X GPUs for generative AI training and inference workloads.

AMD has also reportedly partnered with Microsoft to develop advanced processors that support AI workloads. The move, according to a Bloomberg report, potentially puts the company in direct competition with Nvidia, which currently dominates the market for GPUs used for AI applications.

Speaking on a call with analysts after the company reported its first quarter 2023 financial results in May, Su said the recent interest in generative AI requires significant increases in compute performance and that AMD is “very well positioned” to capitalize on this increased demand due to its large portfolio of high-performance compute engines and expanding software capabilities.

“We are very excited about our opportunity in AI, this is our number-one strategic priority,” she said.

AMD did not immediately respond to requests for comment on the report about the new export restrictions.

The ongoing chip trade war

President Biden’s administration started to impose trade restrictions on the export of chips to China last year in order to stop China from having access to advanced technology for military modernization and human rights abuses.

Since then, a number of companies and jurisdictions have been caught in the crossfire, causing many to enact policies designed to boost their own domestic chip manufacturing capabilities – including the US and China.

China has reportedly poured $143 billion into boosting its domestic chip manufacturing in the face of the trade restrictions, while the US government has earmarked $52 billion for manufacturing incentives to boost microchip production in the country.

(This story has been updated with comment from Nvidia.)

Copyright © 2023 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022