Chinese startup DeepSeek is reportedly developing its own artificial intelligence chip, according to multiple sources. The initiative aims to decrease the company’s current dependence on Nvidia and Huawei chips, which it has used to train and operate its widely recognized AI models.

The new chip is intended to focus on the inference stage of AI computing, where a pre-trained model processes user inputs and generates responses. This distinguishes it from chips designed for training new AI models, which require distinct computational capabilities.

If successful, this move would represent a significant strategic pivot for DeepSeek, a company celebrated in China as a leading force in the country’s AI sector. DeepSeek gained international attention over a year ago after releasing two highly efficient AI models that achieved rapid viral popularity, drawing surprise from industry observers in Silicon Valley and the U.S. capital.

Expanding into semiconductor development could also introduce additional competitive pressures within China’s tech industry, particularly impacting Huawei, which currently holds a substantial share of AI chip production in the country. DeepSeek’s decision underscores a trend of technology companies seeking to internalize critical hardware capabilities amid ongoing geopolitical tensions and supply chain uncertainties.

Details about the chip’s design specifications, production timeline, and potential deployment have not been disclosed. However, the project signals DeepSeek’s ambitions to establish greater technological autonomy and enhance its competitiveness in the global AI market.