Two of Google’s most prominent AI researchers announced within days of each other that they are leaving for rival labs. Noam Shazeer, a co-lead of the Gemini models, is heading to OpenAI. John Jumper, who led the AlphaFold project at Google DeepMind, is going to Anthropic.
Shazeer announced his move on X on June 18. He co-authored the paper “Attention Is All You Need,” which introduced the Transformer architecture behind most of today’s large language models. Google brought him back in 2024 through a deal with Character.AI reported at $2.7 billion, then installed him as a co-lead on Gemini. He is leaving less than two years later. OpenAI CEO Sam Altman welcomed the hire on X.
Jumper said that he would leave after nearly nine years, with plans to take time off before starting at Anthropic. He shared the 2024 Nobel Prize in Chemistry for AlphaFold, the system that predicts protein structures. Both Google DeepMind and Anthropic confirmed the move. Jumper’s work on protein structure lines up with Anthropic’s growing focus on AI for science.
Jumper wrote on X:
Demis Hassabis took a real chance letting me lead the AlphaFold team just six months after finishing my PhD.
Alphabet shares fell about 5% to 6% on June 22, with market reports tying the move to concerns about AI spending and Google’s ability to retain senior AI talent. The stock had held up in the days immediately after Shazeer’s move was reported.
Reporting on the departures points to a competitive pressure inside Google’s AI work. Bloomberg reported that staff at DeepMind have raised concerns about the company lacking a clear product for businesses building AI coding tools, an area where Anthropic and OpenAI have gained ground. It is a point Google’s own leadership has made. In May, SEJ covered CEO Sundar Pichai saying Google was “a bit behind” on agentic coding and tying the gap to a lack of developer-facing products.
Why This Matters
Shazeer and Jumper worked on technology that sits under products the search industry now depends on. Shazeer’s Transformer work is the basis for the models behind AI Overviews and AI Mode. Jumper’s research showed what AI could do in science. Where researchers at that level choose to work can shape how investors and competitors read the frontier AI race.
That doesn’t change how Gemini, AI Overviews, or AI Mode behave today. It does add a data point to how the race is being read.
Looking Ahead
Anthropic has an AI for Science event scheduled for June 30, and OpenAI has filed confidentially for an IPO. Both have been hiring from larger labs. The open question for Google is retention. The company paid a high price to bring Shazeer back once and could not keep him. Whether it changes how it holds onto senior researchers is the thing to watch.
Featured Image: Elnur/Shutterstock
