Chinese AI lab DeepSeek has started developing its own artificial intelligence chips in an effort to reduce dependence on Nvidia, according to sources familiar with the matter.
The Hangzhou-based company is building an internal chip design team focused on developing application-specific integrated circuits optimized for AI inference workloads. The move comes as export controls continue to restrict Chinese companies’ access to advanced Nvidia processors.
DeepSeek’s chip effort mirrors similar initiatives by other Chinese tech giants including Baidu, Alibaba, and Huawei, all of whom have invested heavily in custom silicon. The company has been recruiting chip design engineers from leading semiconductor firms across China.
The shift toward in-house chip development reflects a broader trend among AI model developers who are finding that general-purpose GPUs are not always the most efficient solution for inference workloads. By designing custom chips, these companies can optimize for their specific model architectures and reduce costs at scale.
DeepSeek made headlines earlier this year with its cost-efficient AI models that achieved performance comparable to leading Western models at a fraction of the training cost. The company’s push into chip design suggests it is betting that similar efficiency gains can be achieved on the hardware side.
