July 7 (Reuters) - Chinese startup DeepSeek is developing its own AI chip, according to three people familiar with the matter, a push that could reduce its reliance on Nvidia and Huawei chips, which it has depended on to train and run its globally popular models.
The chip is designed for inference — the stage of AI computing in which a trained model generates responses for users — rather than for training new models, the sources said.
If successful, DeepSeek's expansion into semiconductor development would mark a major strategic shift for a company widely hailed in China as the country's AI champion, potentially adding to challenges faced by Chinese tech giant Huawei. DeepSeek rose to global fame more than a year ago after releasing two highly efficient AI models that went viral worldwide, surprising many in Silicon Valley and Washington.
The company has long been known for emphasizing AI model breakthroughs rather than commercializing its technology.
Although Huawei's offerings still lag Nvidia's most advanced chips by a wide margin, a U.S. ban on their exports to China has helped Huawei gain around half of the $50 billion domestic AI chip market, supplying DeepSeek and several other leading industry players.
However, Huawei's hold on the market is already weakening as tech rivals Alibaba and Baidu develop their own AI chips and gain market share.
DeepSeek's effort to join that race remains at an early stage, with the company reaching out to external partners and holding discussions with chip-design, foundry and memory companies, the three sources said. The effort began about a year ago, one of the people said.
The Hangzhou-based company has also increased hiring of chip-design engineers in recent months, but recruitment has been done privately without job postings on public hiring platforms, two of the sources said.
All three people declined to be identified because the information is not public. Despite becoming a standard-bearer of China's AI ambitions, DeepSeek has kept a low profile. The company did not respond to a request for comment.
FOLLOWING GLOBAL TRENDS
With an in-house chip, DeepSeek would be joining other global AI developers in seeking greater control over the hardware behind their models and reduced dependence on Nvidia's.
OpenAI last month unveiled Jalapeno, its first custom inference chip, developed with Broadcom, while Anthropic has been weighing building its own AI chips, Reuters reported in April.
For DeepSeek, the effort carries an added strategic dimension. U.S. export controls bar Chinese companies from buying Nvidia's most advanced chips, and Beijing has been pressing its technology champions to build domestic alternatives.