Company Baidu, known as the largest search engine in China, in recent years has refocused its business on the development of autonomous vehicles and artificial intelligence. A key element of this strategy is Kunlunxin, a subsidiary engaged in designing AI chips. Baidu presented a five-year plan for the development of its Kunlun chips, starting with the M100 models in 2026 and M300 in 2027. The company already uses its own chips in its data centers for working with ERNIE models, as well as Nvidia products. Baidu earns by selling chips to third-party companies building data centers and providing computing power for rent through cloud services. Baidu aims to become a supplier of comprehensive AI solutions, offering infrastructure that includes chips, servers, data centers, as well as models and applications.
In 2025, Kunlunxin received orders from suppliers like China Mobile, one of the country’s largest mobile operators. Deutsche Bank analysts note that Kunlunxin has become a leading domestic chip developer, focusing on high-performance solutions for training and logical inference of large language models, cloud computing, telecommunications, and corporate tasks. While Nvidia GPUs are considered the most advanced for AI training and deployment, the US government has prohibited the company from selling its top products in China. Additionally, it is reported that Beijing is encouraging local tech companies not to buy the H20, a less powerful Nvidia chip designed for the Chinese market and permitted for export.
As Huawei, previously holding leading positions thanks to its chip clusters, faces challenges, analysts believe Baidu will fill the resulting vacuum, and a period of rapid growth is anticipated. JPMorgan analysts forecast Baidu chip sales will increase sixfold, reaching 8 billion yuan (about $1.1 billion) by 2026. Macquarie analysts estimate the Kunlun chip division’s value at about $28 billion. The shortage of AI chips in China creates additional opportunities for Baidu. Other Chinese tech giants, such as Alibaba and Tencent, report a lack of components for building data centers. Tencent President Martin Lau noted that planned capital expenditures would be lower than expected due to limited chip availability. This is both due to global demand and supply chain issues in semiconductors, as well as the effective blockade of Nvidia chips in China.
