China's AI Chip Deficit: Why Huawei Can't Catch Nvidia (www.cfr.org)

On December 8, the U.S. administration announced a significant loosening of export controls on AI chips by approving the sale of Nvidia's most powerful AI chips, the H200, to China. This move comes amid concern that Huawei might emerge as a competitor to Nvidia in the AI chip market. However, analysis indicates that Huawei is not catching up; instead, it is falling behind due to stringent export restrictions hampering its production capabilities. Currently, Nvidia's AI chips outperform Huawei's best offerings by approximately five times, a gap projected to expand to seventeen times by 2027. Furthermore, Huawei's roadmap suggests that its upcoming chips will actually be less powerful than its current models, indicating major hurdles in domestic chip production. This development is pivotal for the AI and machine learning community as it highlights the critical edge the U.S. has in AI chip manufacturing amid increasing global demand for AI computing power. With China's fast-growing AI sector requiring more advanced chip technologies, the inability of Huawei to produce competitive chips may lead to a more acute chip deficit in China. As U.S. export controls loosen, the prospect of China gaining access to substantial AI computing power raises concerns about potential shifts in global AI dynamics and infrastructure competition, emphasizing the need for monitoring the balance of technological capabilities between the U.S. and China.