Chinese AI startup DeepSeek is developing its own AI chip, Reuters reports, citing three people familiar with the matter. Bloomberg and other outlets repeat the same claim, saying the effort would reduce DeepSeek’s reliance on U.S.-based Nvidia and China’s Huawei chips, which the company has used to train and run its AI models. The chip is reportedly intended for inference—the computing step where a trained model generates responses for users—rather than for training new models. If it succeeds, the move would represent a strategic shift for DeepSeek, which is widely known for delivering highly efficient AI models rather than for building semiconductor technology. Multiple sources frame the potential impact in industry and geopolitics, including possible added pressure on Huawei, which currently supplies parts of China’s AI hardware market. Separate reporting also notes that DeepSeek’s next-generation model work can be affected by U.S. export restrictions on Nvidia chips such as the H20, which have limited availability. Overall, the reporting aligns on the chip-development plan, its inference focus, and its goal of easing dependency on existing AI hardware suppliers.