On October 20, Arfur Grok, a prominent social media account that tracks the movement of AI talent in the United States, reported that Nathan Chen, a former researcher at Moonshot AI, would be joining Discovery Loop.
Earlier, Chen announced his departure from Moonshot AI and his move to the United States, sparking a stir in the tech community amid intensifying competition between Silicon Valley and China's technology hubs.
When contacted by the South China Morning Post (SCMP), Chen declined to comment, while a representative of Discovery Loop said the company “did not wish to comment on personnel matters.”
According to Axios, Arfur Grok has previously accurately reported several recent AI talent recruitment moves in Silicon Valley, including researcher Luke Metz leaving OpenAI to join Meta Superintelligence Labs in August, and researcher Andrew Tulloch leaving Meta to join Anthropic in September. The outlet described Chen’s move to Discovery Loop as “not surprising,” given the company’s growing appeal to AI talent worldwide.
Born in 2009, Chen interned at a Silicon Valley startup before joining Moonshot AI in 2025. He was also named a Rhodes Trust Rise Fellow, a global talent programme for teenagers aged 15 to 17, funded by former Google CEO Eric Schmidt and his wife, Wendy Schmidt.
Despite his young age, Chen was one of the researchers behind the success of the widely known Kimi K3 model, which helped the product compete on equal footing with powerful US rivals such as OpenAI’s GPT and Anthropic’s Claude. Although his tenure lasted only one year, he made important contributions to several core technical innovations at Moonshot AI, helping make the Kimi model one of the world’s leading open-weight AI systems.
Meanwhile, Discovery Loop was founded in August by a group of former Google executives, including Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Lê Viết Quốc, who is known as an “AI prodigy.” The startup aims to “solve humanity’s hardest problems.”
Despite not yet having any products, Discovery Loop has attracted considerable attention. According to Business Insider, the company was valued at US$50 billion following its first funding round, led by Radical Ventures, Khosla Ventures and Google. In August, Jeff Dean said the company’s immediate priority was to pursue RSI (recursive self-improvement), the ability to automate the development of AI through AI itself. This goal is being pursued by many leading AI companies in both the United States and China.
“By building models that are truly good at understanding science and engineering, you can help a model achieve PhD-level expertise across many fields...”
(Source: VnExpress)
