Google DeepMind lost two of its most significant researchers within three days of each other in June 2026, in a sequence that analysts have widely described as a serious blow to the lab’s ability to retain top AI talent amid intensifying competition from rivals OpenAI and Anthropic. On June 18, Noam Shazeer — Google’s vice president of engineering and co-lead of its Gemini AI model family, and co-author of the seminal 2017 “Attention Is All You Need” paper that introduced the Transformer architecture underlying nearly all modern large language models — announced he was leaving Google to join OpenAI as its Lead for Architecture Research. The following day, June 19, John Jumper — winner of the 2024 Nobel Prize in Chemistry alongside Google DeepMind CEO Demis Hassabis, and co-creator of AlphaFold, the AI system that has predicted the three-dimensional structures of more than 200 million proteins — announced he was leaving DeepMind after nearly nine years to join Anthropic.
Shazeer’s departure carries a particular irony given his own history with Google: he originally left the company in 2021 to co-found Character.AI alongside fellow researcher Daniel De Freitas, before Google paid a reported $2.7 billion in 2024 specifically to bring Shazeer and De Freitas back into its DeepMind unit as part of a licensing and talent deal with Character.AI. His second departure from Google, less than two years after that costly reacquisition, underscores how difficult it has become even for extraordinarily well-compensated researchers to remain with a single lab amid the current competition for frontier AI talent.
A Nobel Laureate’s Departure Signals a Different Kind of Race
Jumper’s move to Anthropic is notable for a different reason: unlike Shazeer’s departure, which reflects competition over large language model architecture research, Jumper’s expertise lies specifically in applying AI to biological science — AlphaFold’s protein structure predictions have already had a substantial, well-documented impact on biomedical research and drug discovery worldwide, earning Jumper and Hassabis their joint Nobel Prize. Two additional AlphaFold contributors, Jonas Adler and Alexander Pritzel, followed Jumper to Anthropic, suggesting the move involved relocating a meaningful portion of DeepMind’s core protein-folding research capability rather than a single individual departure. Industry observers have interpreted Anthropic’s specific interest in Jumper as a signal that the company wants to establish genuine scientific credibility in AI-for-science applications, rather than competing solely on consumer and enterprise chatbot capability, as the next major front in AI lab competition.
What Analysts Say Is Driving the Exodus
Reporting from Fortune and other outlets covering the departures has pointed to internal bureaucracy, friction across Google’s multiple overlapping AI product lines, and a broader lack of startup-like focus and speed as root causes behind the wider pattern of senior AI researcher departures from Google DeepMind, of which Shazeer and Jumper’s back-to-back exits are simply the two most prominent recent examples. The timing also coincided with prediction market Polymarket pricing in roughly a 74% chance that Anthropic would complete an initial public offering by the end of 2026, adding financial upside as an additional draw for researchers weighing a move to a company positioning itself for a near-term public listing.
A Pattern That Extends Beyond These Two Departures
Shazeer and Jumper’s back-to-back exits were the most prominent examples of a wider pattern of senior AI researcher departures from Google DeepMind that has continued through 2026, with several other researchers across both the lab’s language-model and applied-science teams also leaving for rival labs during the same period. The scale and seniority of the departures — a Transformer co-inventor and a sitting Nobel laureate leaving within a single week — has prompted renewed public debate over whether Google’s famously deep bench of AI research talent, built over more than a decade, is beginning to erode meaningfully relative to smaller, more focused rivals like OpenAI and Anthropic that can offer researchers greater autonomy and a more direct path to shipping their work.
Google has continued to publicly defend its overall AI research output and Gemini model roadmap despite the departures, and DeepMind CEO Demis Hassabis — Jumper’s own Nobel co-laureate — has remained at the company, a detail some analysts have cited as evidence the exodus reflects specific role and mission dissatisfaction among individual researchers rather than a wholesale loss of confidence in DeepMind’s leadership or research direction.
What This Means for Philippine Founders
The concentration of world-class AI research talent moving specifically toward AI-for-science applications — as illustrated by Jumper’s move to Anthropic — is a meaningful signal for Philippine biotech, pharmaceutical, and life sciences researchers and founders about where genuinely frontier AI capability in drug discovery and protein science is likely to be built and commercialized over the next several years. Filipino AI founders more broadly should also note the underlying lesson from Shazeer’s costly, short-lived return to Google: even massive compensation packages have proven insufficient to retain top AI talent against genuine mission and autonomy concerns, a dynamic worth weighing carefully for any local company trying to build or retain a strong technical team in a competitive global market for AI talent.
Share this article