China has claimed the title of the world's fastest supercomputer with its new LineShine machine, a development that underscores the limitations of U.S. export restrictions in the ongoing tech rivalry. The achievement marks a symbolic shift, as the U.S. had previously held the record with its El Capitan system at Lawrence Livermore National Laboratory.
What makes LineShine particularly notable is how it achieved this milestone. Despite years of increasingly tight U.S. controls on advanced technology exports—especially graphics processing units (GPUs) critical for high-performance computing—China's supercomputer bypassed these barriers by relying on a massive network of standard central processing units (CPUs). This workaround exposes a key vulnerability in Washington's approach to containing Chinese tech advancement.
The U.S. has long sought to hobble China's tech sector by restricting access to American-made semiconductors and manufacturing equipment. But as LineShine demonstrates, these measures often force adaptation rather than capitulation. A similar pattern emerged earlier this year when the federal government debated whether Nvidia could export its H20 chips for use in Chinese AI models. The Trump administration initially recognized that if China were to develop AI, it should at least use American chips. But Beijing responded by developing domestic alternatives, effectively telling Washington, 'thanks but no thanks.'
China is now doubling down on scaling its own tech industry, turning every restriction into an opportunity to strengthen domestic capabilities. This strategy has elevated China to a global leadership position in areas the U.S. tried to block. For Washington, the lesson is clear: restrictions alone won't secure America's sovereignty or economic future.
Congress should think carefully before expanding export controls further, especially through proposals like the Remote Access Security Act. That bill goes beyond preventing sensitive technology from reaching China—it would bar U.S. providers from serving any company that employs Chinese nationals, regardless of the technology's benign nature or the location of corporate customers. Such a broad approach risks making American platforms less attractive globally, driving customers toward competitors like Huawei. Every customer lost to U.S. cloud providers is one Huawei doesn't have to win on merit.
The real goal shouldn't just be keeping American technology out of Chinese hands—it should be ensuring the rest of the world continues to build on U.S. platforms rather than Chinese alternatives. Winning the AI race is about dominating adoption, not just denying access. As China strengthens its domestic tech manufacturing, the U.S. should do the same.
There's a solid foundation to build on. Since the CHIPS Act passed in 2022, the U.S. has treated semiconductors as industrial policy, onshoring production. A Taiwanese company is constructing a $165 billion semiconductor campus in Arizona, one of the most lucrative construction projects globally. This shows the potential for domestic growth. The U.S. also boasts advantages in attracting talent, a formidable venture capital base, and a more privatized tech sector that allows efficient pursuit of new ideas.
Owning the world's fastest supercomputer won't determine the AI race's winner. Becoming the platform the world relies on will. Washington should spend less time building barriers around American innovation and more time making sure the world chooses U.S. technology over China's.
Ram Bala is an associate professor of AI and Analytics at the Leavey School of Business, Santa Clara University.
