U.S. export controls designed to limit China’s access to advanced AI chips have proven largely ineffective due to widespread evasion, with the help of U.S. firms and East Asian allies.

By Samuel Hammond | City Journal

Summary

The U.S. was believed to have a significant lead over China in artificial intelligence (AI), but recent developments suggest that China is now only six to nine months behind, if at all. Chinese AI startup DeepSeek’s rapid replication of OpenAI’s breakthrough model R1 highlights the minimal technical barriers to competing at the AI frontier.

U.S. export controls, intended to restrict China’s access to advanced AI chips, have been largely ineffective due to widespread evasion. Chinese buyers have circumvented these controls by routing chips through third parties in nearby regions, with American firms and East Asian allies facilitating the flow of chips to China. This has been evidenced by Nvidia’s release of China-specific chips like the H800 and H20, and the significant increase in Nvidia’s sales to Singapore, which is a major hub for chip diversion into China.

The stakes are high, as China’s military relies on advanced AI hardware for cyber-attacks, autonomous weapons, and intelligence processing. With the potential advent of Artificial General Intelligence (AGI), access to large amounts of processing power could lead to transformative economic growth and technological innovation, providing a decisive geopolitical advantage.

Despite the challenges, export controls remain America’s best tool for maintaining its strategic edge in AI, as modern AI systems rely on algorithms, data, chips, and energy—areas where China is at parity or ahead of the U.S., except for chips. Without swift action, the U.S. risks losing its AI leadership to China, along with economic and military superiority.

Read the report in full at the publisher’s website here.