In artificial intelligence, the most valuable asset isn't a chip or a data centre — it's a brain. Which is why the past week has been an alarming one for Google. In roughly six days, four senior researchers walked out of Google DeepMind, Alphabet's crown-jewel AI lab, and most of them walked straight into the arms of OpenAI and Anthropic.
These aren't junior engineers. They include a co-author of the paper that started the modern AI boom and a Nobel Prize winner. The fallout was immediate: Alphabet's stock cratered, and a question that had been whispered for months got a lot louder — is Google losing the AI race?
What Happened
Over about a week in June 2026, a cluster of high-profile departures hit DeepMind in quick succession. Individually, any one of them would be notable. Back to back, they look like a talent exodus — and the market treated them as exactly that, erasing a quarter-trillion dollars from Alphabet's value almost overnight.
Who Walked Out the Door
The names are a roll call of modern AI royalty:
| Researcher | Known for | Headed to |
|---|---|---|
| Noam Shazeer | Co-author of "Attention Is All You Need"; Gemini co-lead | OpenAI |
| John Jumper | 2024 Nobel laureate; co-creator of AlphaFold | Anthropic |
| Jonas Adler | AI coding; AlphaFold contributor | Anthropic |
| Alexander Pritzel | Gemini pretraining; AlphaFold contributor | Anthropic |
To appreciate the scale: Shazeer helped invent the Transformer — the architecture underneath virtually every modern chatbot, including Google's own Gemini. Jumper won a Nobel Prize for AlphaFold, the AI system that cracked protein folding. Losing either would sting. Losing both in the same week, plus two of their key collaborators, is the kind of thing that keeps a CEO up at night.
Where They Landed
The destinations tell their own story. Noam Shazeer went to OpenAI. But the other three — Jumper, Adler and Pritzel — all went to Anthropic, the maker of Claude. Three of the four departures, including the lab's most celebrated scientist, now strengthen a single rival.
That's not a coincidence. Anthropic has been on a tear, shipping models quickly and building a reputation as a magnet for researchers who want to move fast. Picking up DeepMind's AlphaFold brain trust deepens its bench in "AI for science" — the same frontier where breakthroughs like AI cracking long-standing biology problems are starting to land.
Why They're Leaving
Neither Shazeer nor Jumper laid out their reasons publicly, but the pattern points away from money and toward culture and momentum. Current and former employees have described DeepMind as bureaucratic and risk-averse — and Shazeer himself had previously criticized the company in those terms. One analysis suggested Jumper left not for a bigger paycheck but because he believed the scientific opportunity at Anthropic was simply better.
That's the part that should worry Google most. When researchers leave for more money, you can counter with cash. When they leave because they think they can do better work somewhere else, the problem is structural — and far harder to fix.
The $250 Billion Reaction
Wall Street didn't shrug this off. As the departures piled up, Alphabet shares tumbled more than 5%, wiping out roughly $250 billion in market value in a single stretch. Let that sink in: a handful of people changing employers moved one of the most valuable companies on Earth by a quarter-trillion dollars.
It's a vivid sign of how the market now thinks about AI. Investors treat elite researchers as core infrastructure — as important as fabs or cloud regions. Lose enough of them, and the market quietly re-rates your odds of winning the most important technology race of the decade.
Is Google Falling Behind?
The departures didn't happen in a vacuum. DeepMind's recent Gemini models have frequently ranked outside the top five on AI leaderboards, and the lab has been shipping more slowly than its rivals — roughly a four-month gap separated Gemini 3.1 Pro from 3.5 Pro, while Anthropic pushed out multiple Claude updates in the same period.
None of this means Google is out of the race. It has near-limitless resources, custom TPUs, and a deep research culture that built much of modern AI in the first place. But a slower release cadence plus a visible talent drain is a worrying combination — and exactly the narrative DeepMind's leadership now has to fight.
Anthropic's Big Win
If Google is the loser here, Anthropic is the unmistakable winner. Landing three of the four departures — a Nobel laureate and two AlphaFold contributors among them — gives it a formidable "AI for science" team almost overnight. It's also a powerful recruiting signal: top people are choosing Anthropic over Google.
The timing matters too. Anthropic has been building momentum and is widely expected to pursue a public listing, and nothing strengthens that story like hiring your larger rival's brightest minds. Talent begets talent — and right now, the gravity is pulling toward Claude's maker.
Frequently Asked Questions
What is happening at Google DeepMind?
Google DeepMind — Alphabet's flagship AI lab — has lost four senior researchers in about six days, with most heading to its biggest rivals OpenAI and Anthropic. The departures include some of the most decorated names in AI, and they've rattled investors enough to knock billions off Alphabet's market value and reignite questions about whether DeepMind can stay at the frontier.
Who has left Google DeepMind?
The most prominent are Noam Shazeer, a co-author of the landmark 'Attention Is All You Need' paper and a Gemini co-lead, and John Jumper, a 2024 Nobel Prize in Chemistry laureate who co-created AlphaFold. They were followed by Jonas Adler, who worked on AI coding and AlphaFold, and Alexander Pritzel, a Gemini pretraining and AlphaFold contributor — four senior exits in roughly a week.
Where are they going — OpenAI or Anthropic?
Noam Shazeer left for OpenAI. John Jumper, Jonas Adler and Alexander Pritzel all headed to Anthropic, the maker of Claude. That makes Anthropic the clear winner of this particular wave — it picked up three of the four, including the lab's marquee scientific talent.
Why are they leaving?
Neither Shazeer nor Jumper publicly detailed their reasons, but the recurring theme is culture and momentum. Current and former staff have described DeepMind as bureaucratic and risk-averse, and one analysis suggested Jumper left because he believed the scientific opportunity at Anthropic was simply better. In short, the pull of faster-moving rivals — not just bigger paychecks — appears to be driving the exodus.
How did Alphabet's stock react?
Badly. Alphabet shares tumbled more than 5% as the news spread, wiping out roughly $250 billion in market value in a single stretch. It's a striking reminder that in the AI era, a handful of star researchers can move a trillion-dollar company's stock — investors treat top AI talent as a core asset.
What does this mean for Google's position in the AI race?
It raises real questions. DeepMind's recent Gemini models have often ranked outside the top five on leaderboards, and the lab has shipped more slowly than rivals — there was a roughly four-month gap between Gemini 3.1 Pro and 3.5 Pro, while Anthropic pushed out multiple Claude updates in the same window. Losing marquee talent on top of that fuels the worry that Google is slipping from the front of the pack.
Why is Anthropic the biggest winner?
Anthropic landed three of the four departures, including a Nobel laureate and two AlphaFold contributors — a deep bench of 'AI for science' expertise. Coming as Anthropic builds momentum and is widely expected to pursue an IPO, hiring DeepMind's scientific stars strengthens both its research and its narrative as a rising frontier lab.
Final Thoughts
Companies lose employees all the time. What makes this different is who left, how fast, and where they went. When a Transformer co-author and a Nobel laureate exit the same lab in the same week — and hand their talents to your two fiercest competitors — it stops being routine turnover and starts looking like a verdict.
Google still has the resources, the chips and the history to fight back, and it would be foolish to count DeepMind out. But the AI race is increasingly a race for people, and right now Google is losing some of its best ones at the worst possible time. How Demis Hassabis and Alphabet respond over the next few months may shape not just DeepMind's future, but the balance of power across the entire industry. We'll be watching closely.