China’s Tech Giants Shift to Homegrown AI Chips

Chinese technology leaders Alibaba and Baidu are making decisive moves to reduce reliance on foreign semiconductors by deploying their own in-house AI chips. Alibaba’s processors are approaching the performance of NVIDIA’s H100 for smaller-scale AI models, while Baidu’s Kunlun P800 is already powering training for its flagship Ernie AI model.

This trend aligns with Beijing’s broader strategy to achieve technological self-sufficiency. With the government encouraging domestic companies to innovate and adopt homegrown processors, the push reflects both national security concerns and the need to secure critical supply chains.

Meanwhile, the U.S. has intensified export controls on advanced GPUs, curbing China’s access to NVIDIA’s most powerful chips. Although a recent agreement under the Trump administration allowed limited sales tied to revenue guarantees, it has failed to slow China’s pivot toward indigenous technology.

For NVIDIA, the stakes are high. China represents its largest overseas market, and losing ground to local competitors could weaken its dominance in global AI hardware.

The shift also highlights a broader AI decoupling between the U.S. and China. As Chinese firms scale domestic innovation and the U.S. tightens restrictions, the race for AI supremacy is set to intensify, reshaping the global technology landscape.