Four Gemini Researchers Leave Google in Six Days for Anthropic and OpenAI

Introduction

Within a span of six days in June 2026, Google lost four of its most prominent AI researchers to Anthropic and OpenAI. Jonas Adler, a key contributor to Google's AI coding efforts, and Alexander Pritzel, who led core pretraining work on the Gemini model, are departing for Anthropic — joining John Jumper and Noam Shazeer, who announced their own exits the previous week. The departures, reported by Bloomberg on June 24, represent the most concentrated talent exodus from Google's AI division in the company's history.

The story is no longer about individual career moves. Four senior departures in six days constitutes a visible pattern — and raises serious questions about whether Google can retain the researchers it needs to remain competitive at the frontier of AI model development.

Who Are Adler and Pritzel?

Jonas Adler specialized in AI coding at Google, working on the coding capabilities central to Gemini's software engineering features. Coding is one of the most commercially competitive dimensions of modern LLM deployment, with Anthropic's Claude Code, OpenAI's Codex, and competing products fighting for developer market share. Adler's expertise covers the exact area where Google most directly competes with Anthropic's strongest product line.

Alexander Pritzel focused on pretraining — the foundational first stage of LLM development where models ingest vast quantities of data to build broad world knowledge and language understanding. Pretraining researchers are among the most scarce specialists in the field; their methodological experience directly shapes the quality ceiling of frontier models. Both Adler and Pritzel also contributed to Google DeepMind's AlphaFold research alongside John Jumper — the 2024 Nobel laureate who departed for Anthropic the week prior.

Neither departure had been formally announced at the time Bloomberg reported it, and both Google and Anthropic declined to comment. This mirrors the pattern of earlier departures: no formal announcements, details confirmed only through independent reporting.

The Six-Day Timeline

Shazeer's departure is particularly notable in historical context. He left Google in 2021 to co-found Character.AI and returned when Google effectively acquired the company for $2.7 billion, partly to bring him back to work on Gemini. That he has now left again — this time for OpenAI — underscores how difficult it is for an established corporation to retain top talent through compensation alone when competitors offer pre-IPO equity.

Why Researchers Are Leaving

Two structural forces are driving these departures. The first is pre-IPO equity. Both Anthropic and OpenAI are expected to go public, potentially at valuations well into the hundreds of billions of dollars. Researchers who join now receive equity stakes that could produce transformative personal financial returns — compensation that Google, as a mature public company, cannot easily match through stock grants structured around its own share price.

The second force is access to compute. Bloomberg's reporting noted that Google researchers faced internal queues for TPU time, suggesting that even at one of the world's largest AI hardware operators, internal resource allocation had become a practical bottleneck for research velocity. Anthropic, backed by Amazon's deep investment and AWS infrastructure, and OpenAI, with its Microsoft Azure partnership, both offer researchers direct access to frontier-scale compute — a consideration that can be more decisive than salary in determining where cutting-edge work gets done.

Broader Industry Implications

For Google: The stakes are substantial. Gemini is Google's primary commercial response to Claude and ChatGPT, and losing researchers who specialize in coding capability and pretraining touches exactly the areas where Gemini competes most directly with Claude. Google has announced deep Gemini integration across Search, Workspace, Android, and hardware — but executing that roadmap at a frontier level requires the researchers who built the model in the first place. Replacements, however qualified, will need time to ramp up on Gemini's architecture and internal tooling.

For Anthropic: The additions of Adler and Pritzel — following Jumper the previous week — give Anthropic direct access to knowledge about Gemini's pretraining methodology and coding architecture. Researchers carry methodological intuition, empirical knowledge of what works at scale, and practical experience with edge cases in training runs. The accumulation of three Gemini researchers at Anthropic in a single week is a meaningful technical reinforcement of its research bench.

For the AI field: Talent concentration at Anthropic and OpenAI may accelerate their respective model development timelines. Conversely, if Google's research velocity slows, it could reduce competitive pressure on Claude and ChatGPT in the second half of 2026 — shifting the dynamic in what has been a tightly contested frontier model race.

Usability Analysis: What This Means for Enterprise Customers

For enterprises evaluating which AI models to bet on for long-term deployment, talent mobility at this level is a relevant signal. A sustained pattern of senior researcher departures from Google's Gemini team could slow the cadence of Gemini capability improvements — particularly in coding and reasoning, the two areas most cited by enterprise buyers as decision criteria.

In contrast, Anthropic's Claude roadmap benefits from the influx of researchers who have hands-on experience training and scaling frontier models. For CIOs and AI procurement teams watching the 2026 model landscape, the talent picture now favors Anthropic in terms of pretraining and coding expertise at the senior researcher level.

Pros and Cons

For Anthropic:

• Access to Gemini pretraining and coding expertise at the senior researcher level
• Three departures in one week signals growing cultural and financial appeal for frontier researchers
• AlphaFold-adjacent expertise (Jumper, Adler, Pritzel) creates a rare concentration of cross-disciplinary AI knowledge

For Google:

• Loss of coding and pretraining specialists in the exact areas where Gemini competes most directly with Claude
• Four departures in six days creates reputational pressure for talent recruitment
• Replacements will need significant ramp-up time on Gemini's internal tooling and architecture

For the industry:

• Pre-IPO equity as a recruitment mechanism is now demonstrably more powerful than established tech company compensation
• Compute access is emerging as a non-salary factor in researcher retention decisions

Outlook

The central question is whether Google can respond with structural compensation changes before the talent pressure deepens further. Options include special pre-IPO-equivalent equity programs, accelerated vesting, or internal spinout vehicles that offer Google researchers a similar upside profile to joining a pre-IPO startup. None of these are quick to implement within a large public company.

Google retains significant structural advantages: the world's largest internal compute infrastructure, unmatched training data access through its consumer products, a massive internal research organization, and Gemini's distribution footprint across Android devices, Search, and Workspace. These moats matter. But frontier AI development is increasingly a talent-constrained problem at the senior researcher level, and Google's position in that competition has visibly weakened in June 2026.

If Anthropic successfully goes public in 2026 and the equity window closes, the structural pull toward Anthropic may moderate. Until then, the incentive gradient continues to run against Google in the frontier talent market.

Conclusion

The departure of Jonas Adler and Alexander Pritzel, coming immediately after Noam Shazeer's move to OpenAI and John Jumper's move to Anthropic, marks the most significant concentrated AI talent event in the industry's history. Four senior departures in six days is not a coincidence — it reflects structural forces (pre-IPO equity and compute access) that Google has not yet resolved. For AI industry observers, this six-day window in June 2026 may prove to be a meaningful inflection point in the competition between Google Gemini and Anthropic Claude for frontier model leadership.

Editor's Verdict

Four Gemini Researchers Leave Google in Six Days for Anthropic and OpenAI earns a solid recommendation within the it news space.

The strongest case for paying attention is establishes pre-IPO equity as the dominant structural force reshaping frontier AI talent markets in 2026, which raises the bar for what readers should now expect from peers in this space. Reinforcing that, adler and Pritzel bring direct hands-on experience with Gemini pretraining and coding architecture — not publicly documented knowledge adds practical value rather than just headline appeal. The broader signal worth registering is straightforward: four senior departures in six days constitutes a pattern rather than coincidence, reflecting structural incentive misalignment between established Big Tech and pre-IPO AI labs. On the other side of the ledger, talent exodus of this scale could slow Google Gemini's capability improvement cadence in the second half of 2026 is a real constraint, not a marketing footnote, and it should factor into any serious decision. Layered on top of that, neither Google nor Anthropic confirmed the departures officially, leaving some details unverified at time of reporting narrows the set of teams for whom this is an obvious yes.

For AI industry watchers, strategy teams, and decision-makers tracking platform shifts, this is a serious evaluation candidate, not just a curiosity to bookmark. For everyone else, the safer posture is to monitor coverage and revisit once the use cases that matter to your team are demonstrated in the wild.