TLDR: Chinese technology giants Alibaba and Baidu have reportedly begun integrating their own domestically developed chips for training artificial intelligence models, a strategic move aimed at reducing their reliance on foreign suppliers like Nvidia. This shift, driven by ongoing U.S. export restrictions on advanced AI hardware, sees Alibaba deploying its Zhenwu processors for smaller AI models and Baidu testing its Kunlun P800 chip for its Ernie AI model. While still utilizing Nvidia’s high-end chips for their most advanced systems, this development underscores China’s long-term commitment to achieving AI technological independence.
In a significant stride towards technological self-reliance, China’s leading tech firms, Alibaba Group and Baidu Inc., have reportedly initiated the use of their own internally designed semiconductors for training artificial intelligence models. This strategic pivot, detailed in a report by The Information on September 12, 2025, marks a partial reduction in their dependence on chips from U.S. manufacturers, particularly Nvidia.
Alibaba has been deploying its Zhenwu processors for smaller AI models since early 2025, while Baidu is actively testing its Kunlun P800 chip to train updated versions of its Ernie AI model. This move comes amidst escalating U.S. export controls on advanced AI chips, which have spurred Chinese companies to accelerate the development and adoption of domestic alternatives.
The transition is a clear manifestation of China’s broader national strategy to foster an independent AI ecosystem. Experts note that this reflects years of strategic preparation, with Chinese cities now targeting 70% AI chip self-sufficiency by 2027. Local governments are actively promoting domestic chip industries to lessen reliance on foreign technology, building on substantial investments and early efforts in the sector.
Despite these advancements, both Alibaba and Baidu continue to utilize Nvidia chips for their most advanced AI systems, indicating that a full replacement of foreign technology has not yet occurred. Reports suggest that Alibaba’s in-house chips are now competitive with Nvidia’s H20 chip, though Nvidia’s high-end H100 and Blackwell series still maintain a performance lead over current Chinese alternatives.
An Nvidia spokesperson, acknowledging the intensifying competition, stated, “The competition has undeniably arrived. To win the AI race, U.S. industry must earn the support of developers everywhere, including China.” This sentiment highlights the global nature of the AI chip race and the strategic importance of market presence.
The news had a positive impact on the stock market, with shares of Alibaba Group and Baidu Inc. surging on Friday following the report. This market reaction underscores investor confidence in the companies’ ability to navigate geopolitical challenges and advance their AI capabilities through domestic innovation.
Also Read:
- Alibaba Shares Surge on Accelerated AI Investments and Innovations
- Alibaba Spearheads US$60 Million Funding Round for AI Video Innovator AIsphere
Further illustrating this national trend, breakout AI star DeepSeek has also reportedly diversified its chip supply, incorporating some of Huawei’s chips in its training efforts. Additionally, Chinese regulators have recently demanded justifications for purchases of Nvidia’s H20 chips, particularly for government or security-related applications, citing concerns over potential backdoor access – allegations that Nvidia has denied. These developments collectively paint a picture of a rapidly evolving AI landscape in China, driven by a strong imperative for self-sufficiency and technological leadership.
