Baidu accelerates its AI-chip push as China scrambles for computing power

China’s race to build its own AI-chip supply is now being pulled forward by Baidu, according to CNBC, as U.S. export rules under President Donald Trump keep Nvidia’s strongest processors out of the country and leave a wide opening in a market worth billions.

What used to be a simple search company is now trying to replace foreign hardware in a sector where every major tech giant is fighting for compute.

Baidu has spent the past few years turning its attention toward self-driving cars and AI, and that shift includes Kunlunxin, a chip business Baidu controls outright.

Analysts have raised their expectations for Baidu’s stock over recent weeks because they believe this semiconductor arm will pull in more orders from inside China.

The upgrades come as companies that used to rely on Nvidia’s GPUs face new restrictions and start hunting for domestic supply.

Baidu lays out chip roadmap as orders arrive

This month, Baidu revealed a five-year plan for its Kunlun chips. The M100 is scheduled for 2026 and the M300 for 2027. Baidu already runs its ERNIE models with a mix of its own processors and Nvidia units still available in China.

The company sells its chips to clients building data centers and also rents out computing power through its cloud platform. Baidu calls this a full-stack setup that includes hardware, data centers, AI models, and finished applications.

Kunlunxin has begun to pick up speed. Early this year, the unit secured orders from suppliers connected to China Mobile, one of the country’s biggest telecom operators. In a note, Deutsche Bank analysts said Kunlunxin has become a leading local AI-chip builder focused on high-performance processors for training and running large language models, cloud workloads, and telecom and enterprise tasks.

Nvidia still makes the world’s most capable GPUs for AI, but the U.S. government blocks its top chips from entering China. Beijing has also urged local firms not to buy the H20, a lower-power Nvidia model approved for the market.

With Huawei limited by supply difficulties, analysts now expect Baidu to absorb a big share of the remaining demand.

JPMorgan analysts said domestic AI-compute demand in China is still intense, and hyperscalers are moving toward local equipment. They wrote that the Kunlun line is among the best positioned in the market.

JPMorgan expects Baidu’s chip revenue to grow sixfold to 8 billion yuan, or about $1.1 billion, by 2026. Macquarie analysts estimate Kunlunxin could carry a valuation near $28 billion.

Baidu isn’t the only company designing its own chips. Cryptopolitan reported in August that Alibaba is working on its next AI processor. Even so, every major tech firm in China is dealing with supply pressure.

AI-chip shortage widens as demand outpaces supply

The push from Baidu arrives while Chinese tech firms warn they are short on chips. Eddie Wu, the chief executive of Alibaba, said “the supply side is going to be a relatively large bottleneck” for the next two to three years.

He said the shortage affects the parts needed to build data centers. Martin Lau, the president of Tencent, said its 2025 spending will fall from earlier plans, not because demand is weak but because not enough chips are available. “It is indeed a change in terms of the AI chip availability,” Martin said.

Some of the shortage comes from global demand and slowdowns in the semiconductor supply chain. Another part comes from the limits placed on Nvidia’s hardware in China. Companies have tried to manage the squeeze by using chips they had stored and by making their models more efficient, but the gap keeps growing.

China also faces limits in manufacturing. Its biggest foundry, SMIC, cannot match the production scale or technology offered by TSMC. That makes it difficult for China to produce enough high-end chips for domestic needs.

Even with these limits, demand for AI systems keeps rising. Eddie said customers want more processing power than Alibaba can deploy. He said they “are not even able to keep pace with the growth in customer demand.” That demand gives Baidu room to expand further into the hardware market.

Nick Patience, who leads AI research at The Futurum Group, said Baidu’s chip push is both urgent and timed well for China’s needs. Nick said Chinese firms can no longer rely on U.S. GPUs and now have a large local market shaped by export rules and Beijing’s goals.

He said if Baidu ships its Kunlun generations on schedule, it will not only fix its own shortages but will become a major supplier for the rest of China’s AI industry.

Join a premium crypto trading community free for 30 days - normally $100/mo.