Sources indicate that DeepSeek is currently creating its own semiconductor chips to facilitate the operation of its artificial intelligence systems.
Reports suggest this chip will be utilized specifically for the inference phase, rather than for constructing new models. Inference is the stage in 人工智能 where a trained model processes user queries and generates responses. DeepSeek has not provided any comment regarding these reports.
On the same day this news emerged, shares of Nvidia experienced a pre-market decline of 2.2%. This development reflects a broader industry trend where nearly all major 人工智能 developers are pursuing proprietary chip designs. The goal is to gain a competitive edge by customizing hardware that is optimally suited for their specific systems.
In a related move just last month, OpenAI unveiled its first custom processor, developed in partnership with Broadcom. Meanwhile, Amazon is reportedly in discussions to sell its custom chips to other companies for use in their data centers. Alphabet Inc.'s proprietary Tensor Processing Units (TPUs) have also seen significant market demand.
DeepSeek has positioned itself at the forefront of China's involvement in the global 人工智能 race. The company made waves in the tech industry last year with the launch of a powerful AI model, which was particularly notable for the relatively minimal amount of chips and computational resources the company claimed were required to build this cutting-edge service.
This achievement demonstrated that Chinese firms can effectively compete with leading Silicon Valley companies, even in the face of U.S. export restrictions on advanced hardware components.
Comments