Google DeepMind is experiencing a significant exodus of its most accomplished researchers, including AlphaFold co-creator John Jumper and nearly 25% of the original protein-folding team, as the company reorganizes to prioritize commercial AI products over fundamental research. The departures reflect a broader strategic shift where Google increasingly treats AI research as infrastructure for its business rather than as an end in itself.

What's Driving the Departure of AlphaFold Researchers?

The reshuffling began in earnest around mid-2025, when DeepMind started reassigning AlphaFold's original researchers to other projects including Gemini model development, enzyme design, nuclear fusion, genomics, and Isomorphic Labs. By late July 2026, the Financial Times and DeepMind confirmed that almost a quarter of the paper's original full-time authors had left Google entirely.

John Jumper, who won the 2024 Nobel Prize in Chemistry alongside Demis Hassabis for AlphaFold's breakthrough in predicting protein structures, departed for Anthropic on June 19, 2026, after nearly nine years at DeepMind. Within weeks, two other AlphaFold alumni, Jonas Adler and Alexander Pritzel, both Gemini contributors, joined him at Anthropic.

"The lab's strategy had moved away from dedicated teams built around single grand challenges," stated Pushmeet Kohli, a DeepMind researcher.

Pushmeet Kohli, DeepMind

This admission came less than two years after the Nobel Prize award, highlighting how quickly the organization's priorities shifted. The pattern extends beyond AlphaFold; David Silver, one of DeepMind's earliest employees and central figure in AlphaGo, AlphaZero, and reinforcement learning research, left in January 2026 to found Ineffable Intelligence, which has since raised $1.1 billion at a $5.1 billion valuation.

Why Are Top Researchers Choosing to Leave?

Analysts describe a consistent pattern in the departures: researchers leaving want to conduct research for its own sake rather than build products, and increasingly see Anthropic, OpenAI, or new startups as better places to pursue that work. This tension reflects a fundamental disagreement about what DeepMind should prioritize.

The organizational changes accelerated after Koray Kavukcuoglu, who was chief technology officer of Google DeepMind, was promoted to chief AI architect in June 2025, a role reporting directly to Google CEO Sundar Pichai. His mandate was clear: bring Google's AI models into its products faster, with quicker iteration and less duplicated work. By August 5, 2026, Kavukcuoglu stepped into the role of senior vice president overseeing Gemini model development, frontier research, the Gemini app, and Google DeepMind's developer teams.

How Google's Business Success Is Reshaping Its Research Culture

• Cloud Revenue Surge: Google Cloud revenue rose 82% to $24.8 billion in the second quarter of 2026, with operating income more than tripling to $8.8 billion and operating margin expanding from 20.7% to 35.6%
• Enterprise Adoption: Nearly 90% of the Fortune 100 now uses Gemini Enterprise, and the Gemini app has reached 950 million monthly users
• Contracted Backlog Growth: Cloud's contracted backlog reached $514 billion, up more than $50 billion in just three months

These financial results explain the logic behind the reorganization. When Demis Hassabis stepped down as chief executive of Google DeepMind in early August 2026 to become chairman and chief scientist of Alphabet (a newly created role focused on long-range artificial general intelligence strategy), the company signaled that commercial success now drives organizational decisions.

Hassabis noted in his announcement that the change would give him more time for Isomorphic Labs, the drug-discovery unit he continues to run as chief executive, and for AGI research itself. However, the timing of his departure, combined with Jeff Dean's exit just hours later, sent a clear message about the company's shifting priorities.

Jeff Dean, Google's chief scientist and a 27-year employee, announced his departure on X without even mentioning Google. He co-founded Discovery Loop, a public benefit corporation designed to automate scientific research, starting with machine-learning experiments that improve AI itself. His co-founders include Sanjay Ghemawat, a Google senior fellow and his engineering partner of two decades; Oriol Vinyals, a Google DeepMind vice-president and Gemini co-lead; and Quoc Le, a co-founder of Google Brain. Alphabet is a founding investor in Discovery Loop and will supply its cloud compute for at least the first year.

What This Means for the Future of AI Research at Google

The departures represent a historical pattern at Google. Eight researchers published "Attention Is All You Need," the 2017 paper that introduced the Transformer architecture underlying nearly every modern AI model, and all of them had left Google by August 2023. Google built BERT, Meena, and LaMDA on that architecture years before ChatGPT existed, yet failed to commercialize these breakthroughs as effectively as competitors.

The current reorganization suggests Google is determined not to repeat that mistake. By placing product-focused leadership like Kavukcuoglu in charge of model development and keeping research-oriented scientists like Dean outside the company while maintaining financial stakes in their work, Alphabet appears to be hedging its bets. The company keeps a stake in research without running it directly, funding Discovery Loop while keeping Isomorphic Labs separate from Kavukcuoglu's product-focused organization.

The stock market reacted sharply to the news, with Alphabet shares falling about 5% and erasing an estimated $160 billion to $200 billion in market value. The decline came after the stock had rallied 13% from a post-earnings low in the three weeks before the announcements.