DeepSeek reportedly builds its own AI chip

DeepSeek has spent roughly a year developing its own AI inference chip, according to three sources cited in a Reuters exclusive published July 7, a move to reduce reliance on Nvidia and Huawei hardware amid US export controls. DeepSeek founder Liang Wenfeng has previously called those controls a major obstacle; unresolved challenges reportedly include foundry access and HBM memory supply. DeepSeek has not confirmed the project itself.