The United States has introduced restrictions limiting technology company Nvidia Corp. and semiconductor company Advanced Micro Devices (AMD), Inc. sales of advanced AI chips to China. The restrictions are a tactical foreign policy and national security move to safeguard the US advantage in the race to develop artificial intelligence technology. How will this regulatory shift affect the AI chip industry in the short term and the long term?
These new restrictions do not completely cut off the flow of AI chips from US companies to China. Rather, they focus on advanced products with powerful computing capabilities, specifically the Nvidia’s A100 and H100 GPUs. Nvidia addressed the new licensing requirements in its 8-K, released on August 26. The company reported that $400 million in potential sales to China could be affected by the new license requirement.
The new restrictions affect AMD’s M1250 chip, according to the Reuters report. Both Nvidia and AMD took hits in the stock market following the news of the restrictions.
Nvidia and AMD are not the only US-based companies developing powerful AI chips. “The new limits may have a substantial effect on other tech firms in addition to Nvidia Corp. and Advanced Micro Devices (AMD) Inc., including Intel Corp., Dell Technologies Inc., Super Micro Computer, Inc., Hewlett Packard Enterprise, KLA Corp., Lam Research Corp. and Applied Materials Inc.,” says Emmanuel Ekwedike, PhD, an AI research scientist and engineer.
The Impact to China
China accounts for one-quarter of the global AI chip market, according to the Center for Security and Emerging Technology (CSET). Restricting US companies’ sales to that market will impact their revenue and market share. “In the short term, they’ll feel some pain because they’ll sell less chips. But in the long term, it will inhibit China from advancing as quickly to be a prime competitor to those chip companies,” Jack Gold, president and principal analyst with industry analyst firm J. Gold Associates, contends.
Nvidia will be allowed to continue the development of the H100 chip in China, and it will be able to continue shipping AI chips out of its Hong Kong facility through September 2023, according to CNBC.
“We are working with our customers in China to satisfy their planned or future purchases with alternative products and may seek licenses where replacements aren’t sufficient. The only current products that the new licensing requirement applies to are A100, H100 and systems such as DGX that include them,” Nvidia said in an emailed statement.
The ultimate goal of these restrictions is to preserve the US’s global competitive advantage in the AI market. Military applications for powerful AI chips are also a consideration. “There is a growing concern that the new, highly capable AI chips could be utilized (misused) for military reasons, as evidenced by the fact that the new limits only apply to specific chips with specific performance capabilities,” Ekwedike says.
Restricting AI chip exports may slow China down in the AI race, but they will not put an end to the competition. “We’re reducing the speed and ease at which new big models can be developed, but customized applications actually don’t necessarily depend on these really high performance [chips], like Nvidia’s A100 and H100 type systems,” explains Bob Rogers, PhD, CEO of Oii, an AI supply chain management software company. Rogers also cautions that these restrictions could cause China to ramp up efforts to manufacture its own AI chips.
These restrictions seem to herald the beginning of further government scrutiny of the AI chip industry. “A wider export ban is probably on the horizon, a wide campaign … limiting the availability of specific advanced chipmaking machinery and specific items that make use of the targeted advanced AI chips,” Ekwedike suggests.
Already, the US government is planning further AI chip export restrictions, which will affect KLA Corp., Lam Research Corp. and Applied Materials, Inc., according to Reuters. The companies will need to obtain licenses from the Commerce Department in order to export chipmaking equipment to China.
What to Read Next:
If you’d like to read the original source of this article please click here Visit Source