Daya Guo Leaves DeepSeek: Implications for AI Competition

Rumors are circulating about the resignation of Daya Guo, a core researcher at DeepSeek and one of the primary authors of the renowned DeepSeek-R1 paper. Guo, who holds a PhD from Sun Yat-sen University, joined DeepSeek in July 2024, focusing on code intelligence and the reasoning capabilities of Large Language Models (LLMs).

During his time at DeepSeek, Guo contributed to key projects such as DeepSeekMath, DeepSeek-V3, and, most notably, DeepSeek-R1, whose research was published on the cover of Nature.

Talent War and Computing Resources

The reasons behind Guo's departure are said to be related to the intense competition among companies to acquire the best talent in the field of AI. There are rumors of more lucrative financial offers from giants like Baidu and ByteDance, as well as the availability of greater computing resources, a crucial factor for those working on the development of increasingly powerful LLMs. It seems that a salary disparity had arisen within DeepSeek between new hires and long-standing team members.

The loss of Guo may not be an isolated incident: there is talk of other "important figures" at DeepSeek in negotiations with other companies, attracted by roles with greater visibility and better resources. The ability of AI startups to retain their talent in the face of the firepower of web giants is being put to the test.