
Baidu Inc is considering listing its Kunlunxin AI chip business in Hong Kong, people familiar with the matter said, as it looks to tap investor interest in China’s attempts to design products to compete with Nvidia Corp.
Kunlunxin, which makes chips that power servers in data centers, is valued at at least $3 billion, the people said, asking not to be identified because the information is private. Deliberations are at an early stage and details such as the IPO timing and size are uncertain, the people said.
Baidu shares jumped as much as 7.8 percent in Hong Kong on Friday, the most since Sept 17, after Reuters first reported that the deliberations were underway.
ALSO READ: Baidu unveils new Kunlun chips
Baidu didn’t immediately respond to an emailed request for comment.
Kunlunxin is one of a few Chinese companies capable of designing the powerful accelerators essential for developing AI — central to Beijing’s effort to wean the country off US technology such as Nvidia chips. Led by Huawei Technologies Co and Cambricon Technologies Corp, these firms hope to fill a void created by Nvidia’s forced departure from the world’s largest semiconductor arena.
Smaller player Moore Threads Technology Co rose more than 500 percent at one stage in its Shanghai debut Friday, underlining bets that homegrown players will climb the tech ladder despite persistent US sanctions.
READ MORE: Baidu adds $6b in day in Hong Kong as buy calls get louder in AI hype
Huawei and Cambricon are sharply ramping up output in 2026, anticipating accelerating demand for chips from Alibaba Group Holding Ltd, DeepSeek and other companies that run AI services.
Kunlunxin was created in part to sate Baidu’s enormous appetite for computing power to run its online businesses. JPMorgan last month singled out the division — along with the giant cloud computing business — as among Baidu’s biggest growth engines.
“The market is underestimating this transformation,” analysts, including Alex Yao wrote in a note, upgrading Baidu shares to overweight from neutral.
