Skip to content
News Leadership

Six Google DeepMind researchers left for Anthropic and OpenAI in five months, Alphabet stock fell more than $245 billion

· by Pondero Newsdesk

The short version

From Noam Shazeer in February to Jonas Adler and Alexander Pritzel on June 24, six senior Google DeepMind researchers departed for rival AI labs as Alphabet absorbed its steepest market-cap loss in over a year.

Six Google DeepMind researchers left for Anthropic and OpenAI in five months, Alphabet stock fell more than $245 billion

Four senior Google DeepMind researchers left for rival AI labs in ten days during June 2026, capping a five-month run of departures that cost Alphabet more than $245 billion in market capitalization in a single trading session and put Sergey Brin on the record about an "urgent" gap in agentic execution.

What happened

The sequence began quietly in February when Denny Zhou, who led reasoning research at Google DeepMind, departed for Meta's Superintelligence Lab. The pace accelerated sharply in June.

On June 18, Noam Shazeer posted that he was joining OpenAI. Shazeer had been a Google engineer since 2000, left briefly to found Character.AI (which Google effectively acquired for $2.7 billion to bring him back), and returned to work on Gemini. He is a co-author of the 2017 "Attention Is All You Need" paper that introduced the Transformer architecture the entire field now builds on per TechCrunch.

Two days later, John Jumper announced he was leaving Google DeepMind for Anthropic. Jumper shared the 2024 Nobel Prize in Chemistry alongside DeepMind CEO Demis Hassabis for AlphaFold, the protein-structure prediction system. His departure alone erased roughly 6% of Alphabet's share price, more than $245 billion in market value, in a single session per The Next Web.

On June 24, Bloomberg reported that two more researchers were set to leave. Jonas Adler, who worked on Google's AI coding products, and Alexander Pritzel, a pretraining specialist, both confirmed departures for Anthropic per TechCrunch. Adler and Pritzel played key roles in Gemini development, according to the Bloomberg report.

Six departures total across five months, spread across reasoning, architecture, structural biology, AI coding, and pretraining: the range covers nearly every capability domain Google has been building.

Why it matters

For AI operators watching the competitive landscape, the talent signal is specific: Anthropic picked up a Nobel laureate in structural biology, a pretraining lead, and an AI coding product engineer inside a week. Those three roles map directly onto Anthropic's publicly known priorities. Claude Code is Anthropic's flagship coding product. The company has separately signaled interest in scientific and healthcare applications. Pritzel's pretraining expertise sits underneath all of it.

The departures accelerated just as Anthropic and OpenAI began moving toward IPO. Pre-IPO equity packages at fast-growing private labs can offer upside that a stock already priced at a multi-trillion-dollar valuation cannot easily match. That dynamic is structural, not the result of any single offer per The Next Web.

There is also a compute dimension. The Next Web reported that internal Google researchers have had to queue for the company's own tensor processing units as external demand soaks up capacity. Anthropic, notably, is one of the external customers buying that TPU capacity from Google.

For anyone evaluating Claude Code or Anthropic's research roadmap, the Adler and Pritzel additions represent a direct transfer of Gemini-level coding and pretraining expertise into the team building the competitor.

Context

Alphabet CEO Sundar Pichai and DeepMind CEO Demis Hassabis both addressed the departures publicly, though neither made specific retention commitments public. Sergey Brin, by this point directly overseeing Google's AI Coding Strike Team, put the stakes plainly in an internal memo: "To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers of final code," per the Bloomberg report cited by The Next Web coverage.

Google reorganized the Coding Strike Team around the same time, expanding its scope beyond AI coding tools to include a dedicated midtraining phase. Bloomberg described the expanded scope as a structural response to capability gaps left by the departing engineers.

Gemini remains a competitive frontier model, and Google has the resources to hire and has weathered talent cycles before, but the pattern that matters here is the direction: four senior researchers left in ten days, all to the two rivals best positioned to use them.

What to watch next

The practical question for operators is whether Anthropic announces new research initiatives tied specifically to Adler, Pritzel, and Jumper. The combination of their expertise covers the core stack for a research-grade coding agent: pretraining, scientific reasoning, and product-side coding tooling. Watch Alphabet's Q2 2026 earnings call for any executive commentary on retention strategy and whether the Coding Strike Team produces visible product results in the second half of the year.

Sources