In 2011, Baidu started work on the Kunlun AI chip project. Initially, Baidu researched and emulated its many-small-core XPU microarchitecture using FPGAs, but in 2018 finally built a dedicated silicon that was built using one of Samsung Foundry's 14 nm process.[1] In July 2018, Baidu unveiled the Kunlun AI chip to the public marking its debut. The new design was 30 times faster than the original FPGA-based processor.[4] The chip was deployed in Baidu's search engine.[3]
In August 2021, Kunlunxin unveiled the Kunlun II AI Chip which was comparable to the Nvidia A100.[1] It was introduced to the Ernie Bot deep learning model as well in Baidu robotaxi and Apolong.[3]
In April 2025, at its developer conference, Baidu unveiled a 30,000-chip cluster powered by its third-generation P800 Kunlun chips. Baidu pitched the system as capable of training "DeepSeek-like" models with hundreds of billions of parameters.[5] They would be used to handle the vast majority of the Baidu's inference tasks.[3]
In August 2025, Baidu's AI servers powered by Kunlun chips won the bidding for a centralized procurement project held by China Mobile.[3]
In November 2025, Baidu stated it would release the Kunlun M100 chip in 2026 and the Kunlun M300 chip in 2027.[3][6]
In December 2025, Kunlunxin reached a valuation of almost US$3 billion. It was reported that there were plans for an initial public offering on the Hong Kong Stock Exchange.[7]
