DeepSeek secretly develops its own inference chip

According to monitoring by Beating, Reuters cited sources saying that Chinese large model company DeepSeek is developing its own AI inference chips. This development work started about a year ago, primarily for model inference rather than model training. In recent months, the company has been privately recruiting chip design engineers.

Currently, DeepSeek's model operation and training heavily rely on chips from Nvidia and Huawei. If the self-developed inference chip succeeds, it will reduce dependence on external suppliers and gain more cost-effective control over hardware. However, this project is still in its early stages and faces manufacturing and memory access restrictions due to U.S. export controls.