Google Loses Four AI Stars to Rivals in Six Days

Between June 18 and June 24, 2026, Google lost four of its most consequential AI researchers to rivals OpenAI and Anthropic - a departure wave that sent Alphabet shares tumbling and raised pointed questions about whether the world's largest AI research organization can hold onto the talent that built it.

TL;DR

Noam Shazeer, co-author of the transformer paper that underpins modern AI, left Gemini for OpenAI on June 18
John Jumper, 2024 Nobel laureate and AlphaFold lead, announced his departure for Anthropic on June 19
Bloomberg reported June 24 that two more Gemini contributors - Jonas Adler and Alexander Pritzel - are also joining Anthropic
Alphabet stock closed at $345.29 on June 24, down nearly 10% from its May 13 all-time high of $402.38
Google DeepMind CEO Demis Hassabis called the departures "expected" - investors weren't reassured

Six Days, Four Departures

Noam Shazeer - The Transformer Pioneer Goes to OpenAI

On Wednesday, June 18, Noam Shazeer announced on X that he was leaving Google to join OpenAI as Lead for Architecture Research. The announcement was quiet by tech standards - a short post expressing gratitude and excitement - but the industry implications were anything but.

Shazeer co-authored "Attention Is All You Need," the 2017 paper that introduced the transformer architecture and set off the modern AI revolution. He originally left Google in 2021, frustrated by the company's reluctance to publicly release its Meena chatbot, and co-founded Character.AI. Google bought him back in 2024 as part of a $2.7 billion licensing deal for Character's technology, installing him as co-lead of Gemini.

That arrangement lasted less than two years.

"I'm excited to share that I'll be joining OpenAI and look forward to working with the exceptional team there. It was a difficult decision to move on. I'm incredibly proud of the amazing team at Google and everything we've built together."
Noam Shazeer, June 18, 2026

OpenAI CEO Sam Altman didn't hide his enthusiasm about landing the hire: "Noam is one of the people I have most wanted to work with since the very beginning of OpenAI. Only took ten years."

John Jumper - The Nobel Laureate Takes His Leave

One day later, on June 19, John Jumper announced he was leaving Google DeepMind for Anthropic. The timing was remarkable. Within 48 hours, Google had lost the person who co-designed the architecture that powers its frontier models and the person who used AI to solve one of biology's hardest problems.

Jumper, who turned 40 this year, is the co-creator of AlphaFold - the AI system that has predicted protein structures for more than 200 million sequences and is now used by over two million scientists in 190 countries. He shared the 2024 Nobel Prize in Chemistry with Demis Hassabis for that work, becoming the youngest chemistry laureate in over 70 years at the time of the award.

Reflecting on his tenure, Jumper acknowledged what it meant to have Hassabis back him early in his career: "Demis Hassabis took a real chance letting me lead the AlphaFold team just six months after finishing my PhD."

Hassabis, for his part, offered a gracious public response: "What we achieved with AlphaFold changed the world, and showed the field what was possible with AI for science and medicine, lighting the way for how AI can benefit humanity."

John Jumper and Demis Hassabis celebrating the 2024 Nobel Prize in Chemistry for AlphaFold Demis Hassabis and John Jumper, whose AlphaFold work earned them the 2024 Nobel Prize in Chemistry, shortly before Jumper's departure for Anthropic. Source: deepmind.google

Four Days Later, Two More Gemini Researchers Exit

Jonas Adler

Bloomberg reported on June 24 that Jonas Adler is also planning to leave Google for Anthropic. Adler worked on Google's AI coding efforts and contributed to the AlphaFold project alongside Jumper. Internally, he was considered one of the key technical contributors to Gemini's development.

Alexander Pritzel

Also named in Bloomberg's reporting was Alexander Pritzel, who was involved in pretraining - the foundational stage in which a model learns from massive datasets. Pritzel, like Adler, worked on AlphaFold research in addition to his Gemini work. Both are expected to join Anthropic.

The four departures across six days, combined with David Silver's departure earlier this year to found Ineffable Intelligence - a London lab that raised $1.1 billion at a $5.1 billion valuation before shipping a single product - paint a picture of systematic talent drain from one of the field's most storied research institutions.

What Google Paid to Lose

The Shazeer departure is particularly costly to understand in dollar terms. Google's deal with Character.AI in 2024 cost roughly $2.7 billion. Much of that was understood as a talent acquisition - specifically to bring Shazeer, who had publicly criticized the company's pace and risk tolerance, back inside the tent.

Less than twenty-four months later, he walked out again. This time, to Google's direct competitor.

For Jumper, the cost is harder to quantify in cash but no less significant strategically. AlphaFold is widely regarded as the most consequential scientific AI system ever built. The research that earned him a Nobel Prize was done at DeepMind. That lineage, and whatever he builds next at Anthropic, now belongs to the competition.

AlphaFold protein structure prediction visualization AlphaFold's protein structure predictions have been used by over two million researchers globally - a scientific achievement now heading to Anthropic's research portfolio. Source: deepmind.google

Hassabis Responds

Pressed on the departures, Hassabis told reporters that "movement between leading laboratories was expected in the current market" and maintained that Google retained "the largest and broadest research team in the industry."

Both statements are defensible in isolation. Talent mobility between AI labs has accelerated across the board. And DeepMind's headcount, measured in raw numbers, remains enormous. But Hassabis was careful not to directly address what any outside observer could plainly see: the departures aren't random. They're concentrated at the very top of the talent distribution, and they are flowing in one direction.

The Pattern Beneath the Exits

Four departures in six days don't happen by accident. The picture that emerges from multiple sources is one of an organization whose culture - bureaucratic, risk-averse, large - is increasingly mismatched with the pace at which the field is moving.

Internally, staff have raised concerns about Google's lack of a competitive AI coding product, exactly the category where Anthropic's Claude Code and OpenAI's Codex have gained decisive market traction. The gap isn't just technical - it's visible to users and enterprise customers in ways that benchmark rankings don't always capture.

The competitive dynamics are deepening. Anthropic recently closed a $65 billion funding round at a $965 billion valuation, making it the most valuable private AI company in the world. That kind of capital, combined with a clear product roadmap, creates a powerful pull for researchers who want to see their work shipped and used.

Shazeer left Google once before because the company wouldn't move fast enough. He came back for $2.7 billion. He left again two years later for a competitor. Whatever is happening inside Google's AI organization, money alone wasn't enough to fix it.

Alphabet stock closed at $345.29 on June 24, down from its all-time high of $402.38 on May 13 - a decline of nearly 10% over six weeks that analysts linked in part to the cascading news of the departures.

The exits are individually explicable - researchers move, opportunities arise, organizations evolve. But four in six days, each landing at a direct rival, each carrying deep institutional knowledge of Google's most competitive work, represents something beyond ordinary churn. It's a signal about who believes the next five years of AI are going to be won, and by whom.

Sources: