US tech leaders aim for fewer export curbs on AI chips for China

The symbol for Intel appears on a screen at the Nasdaq MarketSite, in New York. [AP Photo]

Intel Corp has introduced a processor in China that is designed for AI deep-learning applications despite reports of the Biden administration considering additional restrictions on Chinese companies to address loopholes in chip export controls.

The chip giant’s product launch on July 11 is part of an effort by U.S. technology companies to bypass or curb government export controls to the Chinese market as the U.S. government, citing national security concerns, continues to tighten restrictions on China's artificial intelligence industry.

CEOs of U.S. chipmakers including Intel, Qualcomm and Nvidia met with U.S. Secretary of State Antony Blinken on Monday to urge a halt to more controls on chip exports to China, Reuters reported. Commerce Secretary Gina Raimondo, National Economic Council director Lael Brainard and White House national security adviser Jake Sullivan were among other government officials meeting with the CEOs, Reuters said.

The meeting came after China announced restrictions on the export of materials that are used to construct chips, a response to escalating efforts by Washington to curb China's technological advances.

VOA Mandarin contacted the U.S. chipmakers for comment but has yet to receive responses.

Reuters reported Nvidia Chief Financial Officer Colette Kress said in June that "over the long term, restrictions prohibiting the sale of our data center graphic processing units to China, if implemented, would result in a permanent loss of opportunities for the U.S. industry to compete and lead in one of the world’s largest markets and impact on our future business and financial results."

Before the meeting with Blinken, John Neuffer, president of the Semiconductor Industry Association, which represents the chip industry, said in a statement to The New York Times that the escalation of controls posed a significant risk to the global competitiveness of the U.S. industry.

"China is the world’s largest market for semiconductors, and our companies simply need to do business there to continue to grow, innovate and stay ahead of global competitors,” he said. “We urge solutions that protect national security, avoid inadvertent and lasting damage to the chip industry, and avert future escalations.”

According to the Times, citing five sources, the Biden administration is considering additional restrictions on the sale of high-end chips used to power artificial intelligence to China. The goal is to limit technological capacity that could aid the Chinese military while minimizing the impact such rules would have on private companies. Such a move could speed up the tit-for-tat salvos in the U.S.-China chip war, the Times reported.

And The Wall Street Journal reported last month that the White House was exploring how to restrict the leasing of cloud services to AI firms in China.

But the U.S. controls appear to be merely slowing, rather than stopping, China’s AI development.ILE - A Nvidia Corp. sign is shown in Santa Clara, Calif., May 31, 2023. AI chips and their leading designer, Nvidia, are now at the center of what some experts consider an AI revolution that could reshape the technology sector.

Last October, the U.S. Commerce Department banned Nvidia from selling two of its most advanced AI-critical chips, the A100 and the newer H100, to Chinese customers, citing national security concerns. In November, Nvidia designed the A800 and H800 chips that are not subject to export controls for the Chinese market.

Taiwan chip industry emerges as battlefront in US-China showdown

According to the Journal, the U.S. government is considering new bans on the A800 exports to China.

According to a report published in May by TrendForce, a market intelligence and professional consulting firm, the A800, like Nvidia's H100 and A100, is already the most widely used mainstream product for AI-related computing.

Combining chips

Robert Atkinson, founder and president of the Information Technology and Innovation Foundation, told VOA in a phone interview that although these chips are not the most advanced, they can still be used by China.

“What you can do, though, is you can combine lesser, less powerful chips and just put more of them together. And you can still do a lot of AI processing with them. It just makes it more expensive. And it uses more energy. But the Chinese are happy to do that,” Atkinson said.

As for the Chinese use of cloud computing, Hanna Dohmen, a research analyst at Georgetown’s Center for Security and Emerging Technology, told VOA Mandarin in a phone interview that companies can rent chips through cloud service providers.

In practice, it is similar to a pedestrian hopping on an e-share scooter or bike — she pays a fee to unlock the scooter’s key function, its wheels.

For example, Dohman said that Nvidia's A100, which is “controlled and cannot be exported to China, per the October 7 export control regulations,” can be legally accessed by Chinese companies that “purchase services from these cloud service providers to gain virtual access to these controlled chips.”

Dohman acknowledged it is not clear how many Chinese AI research institutions and companies are using American cloud services.

“There are also Chinese regulations … on cross-border data that might prohibit or limit to what extent Chinese companies might be willing to use foreign cloud service providers outside of China to develop their AI models,” she said.

Black market chips

In another workaround, Atkinson said Chinese companies can buy black market chips. “It's not clear to me that these export controls are going to be able to completely cut off Chinese computing capabilities. They might slow them down a bit, but I don't think they're going to cut them off."

According to an as yet unpublished report by the Information Technology and Innovation Foundation, China is already ahead of Europe in terms of the number of AI startups and is catching up with the U.S.

Although Chinese websites account for less than 2% of global network traffic, Atkinson said, Chinese government data management can make up for the lack of dialogue texts, images and videos that are essential for AI large-scale model training.

“I do think that the Chinese will catch up and surpass the U.S. unless we take fairly serious steps,” Atkinson said.