
China’s Alibaba and Baidu have begun using their own internally designed chips to train artificial intelligence models, partially replacing Nvidia hardware. Alibaba has deployed its chips for smaller AI models since early 2025, while Baidu is testing its Kunlun P800 chip to train new versions of its Ernie AI model. Although the shift marks an important development in China’s tech sector, both firms continue to rely on Nvidia for their most advanced projects.
This move reflects a significant change in the AI landscape, where reliance on Nvidia processors has long dominated. With U.S. export restrictions tightening, Chinese companies face increasing pressure to strengthen their domestic technology capabilities. As a result, developing in-house chips allows Alibaba and Baidu to reduce dependency on U.S. suppliers and secure greater control over their AI strategies.
Impact on Nvidia and Market Competition
Despite these changes, Nvidia remains a key player. Its H20 chip, while less powerful than the H100 and Blackwell series, still outpaces most Chinese alternatives. Nevertheless, reports indicate Alibaba’s new AI chip now performs competitively against the H20, according to employees familiar with the technology. This development signals that the gap between domestic and foreign hardware is narrowing faster than expected.
An Nvidia spokesperson responded, stating: “The competition has undeniably arrived … We’ll continue to work to earn the trust and support of mainstream developers everywhere.” Although Nvidia faces challenges, the company still supplies hardware for China’s most advanced AI models, highlighting a dual approach where local and foreign chips coexist.
Rising Pressure and Strategic Shifts
Export controls from Washington have forced China’s tech giants to adapt quickly. Companies are under growing pressure from Beijing to adopt homegrown technology, and in-house chips represent a critical step in meeting those demands. At the same time, global competition in AI hardware is intensifying, which could reshape supply chains and influence long-term partnerships.
Meanwhile, Nvidia CEO Jensen Huang recently said talks with the White House over selling a scaled-down version of the next-generation chip to China will take time. Although challenges remain, the shift by Alibaba and Baidu toward self-reliance demonstrates China’s determination to carve out a stronger position in the global AI race.