Anthropic and OpenAI just pulled four major AI names out of Google in a single week, including 2024 Nobel chemistry laureate John Jumper. The brain drain is accelerating right as both rivals get ready for their public listings.
Key Takeaways
- Four key Google researchers joined Anthropic and OpenAI within one week
- The Anthropic and OpenAI IPO context is fueling talent moves through pre listing equity
- Google has not publicly addressed the wave of departures
Four senior departures in seven days
Between June 18 and June 24, four prominent figures of AI research left Google. Jonas Adler and Alexander Pritzel, both key contributors to Gemini, joined Anthropic. Their arrival hands Google’s most direct competitor fresh expertise on the architecture that drives the entire Gemini family. For context, see our earlier piece on Horizon: Google Appeals Munich AI Overviews Ruling.
On June 18, Noam Shazeer announced his move to OpenAI. His Google tenure goes back to 2000, with a three year stint running Character.AI. Shazeer is one of the original co authors of the Transformer paper, the foundation of nearly every modern LLM architecture.
Two days later, John Jumper confirmed his exit from DeepMind for Anthropic. Jumper led the team behind AlphaFold, the protein structure prediction system that earned him the 2024 Nobel Prize in Chemistry. His name is one of the most prestigious currently active in AI research.
Four announcements in one week change the tone of the Google brain drain. This is no longer an isolated case. It is a clearly identifiable wave. And the destination is narrow. Three of the four moved to Anthropic, the fourth to OpenAI.
When asked for comment, Google did not respond. That silence is rare on a topic this public, and it lets the competitors own the media narrative alone.
The IPO effect on the talent race
Anthropic and OpenAI are not hiring at random. Both companies are preparing for their stock market debuts, and every executive level hire at this stage comes with an equity grid calibrated for the pre IPO window. A researcher who joins now signs with the prospect of a public valuation that can multiply the package by a large factor.
That math reframes the exits. OpenAI is heading into its IPO with 34 billion in losses, which has not stopped the secondary market from valuing the company at record levels. Anthropic is on the same path with a similar pre listing valuation engine. In both cases, the equity offered at hiring can outweigh a comparable Google offer, despite the financial strength of the Mountain View company.
Timing matters as well. Once the listing is effective, the upside on new equity grants flattens out. The current window is where the risk reward profile is most favorable for the very profiles both companies want to capture.
For the researchers themselves, the calculation is rational. The work is still applied AI research on cutting edge models. The day to day science does not change dramatically. What changes is how the time spent is valued. A few years at a pre IPO player can turn a research career into significant personal wealth.
The public debate around the role of researchers in the AI race also seeps into the decision. Sam Altman recently accused academic researchers of slowing the field down, a sign that the line between science and product has become political. Joining Anthropic or OpenAI is also a way of picking a side on that fault line.
Also on Horizon:
- Claude Tag: Anthropic Builds a Memory for Slack
- OpenAI Starts Patching Open Source Bugs
- Sam Altman Says Researchers Held the Whole Field Back
What Google loses and what comes next
In the short term, Google loses identifiable contributors on two specific fronts. Gemini on the foundation model side, and DeepMind on the applied science side. Program continuity does not collapse with these exits, but internal knowledge transfer will demand real effort. The remaining teams have to absorb expertise that was not documented at the same speed it was produced.
The reputational impact on DeepMind’s science output is immediate. A Nobel laureate leaving a few months after his recognition sends a clear signal across the ecosystem. Younger researchers hesitating between a more established Google and a riskier Anthropic will read the Jumper move as validation of the path on the other side.
In the medium term, Google will have to overhaul its retention grid. The company has the financial firepower to respond through retention bonuses, targeted grants and internal promotions. But the pre IPO mechanics of an Anthropic or an OpenAI cannot be replicated inside a company that has been public for two decades. A different value proposition will be needed, likely built on research freedom and access to massive compute.
The competition will not pause either. Anthropic and OpenAI will keep approaching Google profiles for as long as the IPO window stays open. The current movement looks like a beginning, not an end. If Google stays silent in the coming weeks, more names will surface, and the narrative of a leader slowed down by its own size will harden.
The real test will be the next Gemini release and the next major DeepMind publication. If Google proves the engine keeps producing at the same pace despite the exits, the brain drain will be quickly forgotten. If not, the perception of a leader losing ground will spread to enterprise customers, and pressure on strategic partnerships will go up another notch.
Follow the story on Horizon.
