Baidu unveils AI computing platform powered by Chinese chips to push a domestic tech stack

Baidu, one of China's artificial intelligence champions, on Thursday unveiled its Baige 5.0 AI infrastructure platform - powered by a mix of semiconductors, including those designed by its Kunlunxin unit - to raise the efficiency of DeepSeek's open-source models.

The Beijing-based AI and internet search giant's upgraded platform delivers a speedier network connection, increased computing power, and enhanced model training and inferencing capabilities, according to executive vice-president Shen Dou, who also serves as president of Baidu AI Cloud Group, speaking at a corporate event held in China's capital on Thursday.

Baige's inferencing system - backed by adaptive and smart resource allocation technologies, which speed up data throughput and lessen communications latency - improved the inferencing efficiency of DeepSeek's R1 reasoning model by around 50 per cent, according to Shen. Inferencing is the process that a trained AI model uses to draw conclusions in response to human queries.

Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.

"That means, with the same time and cost ... we could have the model 'think' 50 per cent more [or] work 50 per cent more," Shen said.

The launch of Baige 5.0 reflects the growing efforts across the mainland's AI and semiconductor sectors to push forward the development of a domestic AI technology stack, reducing the impact of US trade restrictions on China.

Baidu executive vice-president Shen Dou serves as the president of the company's AI Cloud Group. Photo: Handout alt=Baidu executive vice-president Shen Dou serves as the president of the company's AI Cloud Group. Photo: Handout>

Shen said the Kunlunxin Super Node, which supports hundreds of interconnected AI chips, had gone live on the Baige 5.0 platform, making it capable of deploying and running a trillion-parameter AI system within minutes. The number of parameters in an AI model dictates the size and complexity of its use.

Baidu unit Kunlunxin designs integrated circuits to compete with those from US semiconductor powerhouse Nvidia and mainland rivals that include Huawei Technologies' HiSilicon and Cambricon Technologies.

Kunluxin last week said it won chip orders worth over 1 billion yuan (US$139 million) for the AI projects of state-owned telecommunications network operator China Mobile, according to a Reuters report.

Story Continues

Cambricon, meanwhile, on Tuesday reported a revenue surge in the first half of 2025 on the back of "continued market expansion and active support for the implementation of AI applications", according to the firm's earnings report.

The launch of the Kunlunxin Super Node on the Baige platform comes as more Chinese enterprises adopt domestic chips to power AI development projects. This trend received a big boost last week, when Hangzhou-based DeepSeek suggested in a cryptic one-line post on WeChat that China's next-generation AI chips would soon be released.

Baidu's Hong Kong-listed shares closed 1.25 per cent lower to HK$86.60 on Thursday.

This article originally appeared in the South China Morning Post (SCMP), the most authoritative voice reporting on China and Asia for more than a century. For more SCMP stories, please explore the SCMP app or visit the SCMP's Facebook and Twitter pages. Copyright © 2025 South China Morning Post Publishers Ltd. All rights reserved.

Copyright (c) 2025. South China Morning Post Publishers Ltd. All rights reserved.