DeepSeek is secretly developing its own AI inference chips

Odaily reported that sources revealed Chinese AI large model company DeepSeek is secretly developing AI inference chips. The project was initiated approximately one year ago and is primarily focused on model inference rather than training. According to reports, DeepSeek has been continuously recruiting chip design engineers in recent months to advance this project's development.

Currently, DeepSeek's model training and inference mainly rely on chips from Nvidia and Huawei. If its self-developed inference chips are successfully deployed, it could reduce dependence on external suppliers and enhance hardware cost control capabilities. However, the project remains in its early stages and also faces chip manufacturing and memory acquisition constraints imposed by US export controls. (Reuters)