Will blocking China from accessing relatively high-performing AI chips help the U.S. maintain global AI leadership? The answer is no. Restricting China’s access to U.S.-designed AI chips will not strengthen America’s leadership in AI and related technologies. Instead, it will prompt China to accelerate the development and deployment of its own AI chip capacity and ecosystem, ultimately weakening U.S. leadership.
The seesawing policy decisions on AI chip exports to China reflect the competing views on this issue. A key example is the Nvidia graphics processing unit (GPU) known as the H20. GPUs are essential for the computations required to train and run large AI models. Nvidia, a leading supplier of high-performance GPUs, designed the H20 to comply with export regulations. Initially, the Commerce Department imposed new restrictions on the sale of the H20 to China, citing concerns about China’s advancing AI capabilities. Nvidia reported a significant financial impact due to these restrictions. However, subsequent policy shifts allowed Nvidia to resume applications for selling the H20 to China, though the actual level of permitted sales remains uncertain.
Limiting AI Chip Sales to China: A Shortsighted Perspective
One argument for restricting AI chip sales to China is that it will limit China’s access to the computing power needed to compete effectively, thereby cementing America’s status as the global AI leader. However, this perspective is shortsighted for several reasons.
Firstly, U.S.-imposed scarcity will drive greater AI investment within China. Companies like Huawei have significant human and financial capital that can be directed towards AI development. The restrictions will incentivize increased investment in AI within China, leveraging the expanded market opportunities created by the U.S.-imposed shortage.
Secondly, the future of AI extends beyond high-performance chips in data centers. AI is an ecosystem that includes billions of edge devices, such as mobile phones, laptops, connected vehicles, and factory automation solutions. The AI solutions of the future will increasingly rely on a combination of computing at both the edge and the data center. Huawei has a substantial presence in many markets outside the U.S., with a large and rapidly growing global installed base of edge devices. This provides Huawei with a competitive advantage when companies in non-U.S. jurisdictions decide how to build their AI data centers. If Huawei’s chips become nearly as capable as U.S. chips, the global market may shift towards all-Huawei AI ecosystems, potentially leading the U.S. to lose the global AI race despite a strong domestic AI chip market.
Ensuring U.S. Leadership in AI
The best approach to ensure U.S. leadership in AI is to limit the economic incentives for China to accelerate its AI chip advancements. This involves allowing U.S. companies to sell highly capable (though not the most capable) chips into the Chinese market. More broadly, it requires a policy focus on positioning the U.S. as the world’s supplier of choice for various components of the AI ecosystem, including both AI data center chips and edge devices. By doing so, the U.S. can maintain its technological edge and ensure that the global AI landscape remains favorable to American interests.
