AI researchers keep leaving Google: Adler and Pritzel head to Anthropic

By TechCrunch · June 24, 2026.

Google is experiencing an unprecedented drain of scientific talent in its recent history. As Bloomberg reports and TechCrunch picks up, researchers Jonas Adler and Alexander Pritzel, both key players in the development of the Gemini artificial intelligence model, have announced their departure from the company to join Anthropic. The news comes at a particularly delicate moment for Google, which in the span of barely a week has seen several of its most illustrious names announce their exit to direct competitors.

This double departure is not an isolated event, but the continuation of a trend that has accelerated visibly. The previous week, the legendary researcher Noam Shazeer announced that he was leaving Google to join OpenAI. Shazeer is a figure of historic weight at the company: he had been working at Google since the year 2000, with a single three-year hiatus during which he founded the controversial chatbot startup Character.AI. The covert acquisition of that company by Google —valued at approximately $2.7 billion— was partly motivated, according to the article itself, by bringing Shazeer back to work on the development of Gemini. That this monumental financial effort was not enough to retain him speaks for itself about the magnitude of the crisis.

A few days after Shazeer made his decision public, it was the turn of John Jumper, director of Google DeepMind. Jumper is not just any researcher: alongside Demis Hassabis, CEO of DeepMind, he received the Nobel Prize in Chemistry in 2024 for his work on AlphaFold, the system capable of predicting three-dimensional protein structures from amino acid sequences. That an active Nobel laureate decides to leave an organization is an extraordinary event in any scientific field; that he does so to go precisely to Anthropic, Google's most direct competitor in the segment of high-performance language models, adds a strategic and symbolic dimension that goes beyond the personal.

The pattern drawn by this wave of departures is clear: Google is losing talent to the two major rivals that dominate the narrative of generative AI in 2026, OpenAI and Anthropic. This is especially significant considering that Google has for years been regarded as the world's leading incubator of artificial intelligence research, with milestones ranging from the publication of the original paper on the Transformer architecture —the basis of practically all of today's major models— to AlphaFold itself.

The TechCrunch article points to a structural factor that could be accelerating these departures: both OpenAI and Anthropic are preparing to go public. In that context, the ability of these companies to offer equity-based compensation packages —shares that are expected to appreciate significantly at the time of the IPO— represents an extraordinarily powerful recruiting magnet. For an elite researcher, the promise of participating in the value creation of a company in a hypergrowth phase, with a public offering in sight, can outweigh any salary offer that an already publicly traded company like Google is able to make.

As sector context, the war for AI talent is not new, but it has reached an unprecedented intensity in recent years. The most cutting-edge researchers in the field —especially those with experience training and scaling foundational models— are an extremely scarce resource. Universities do not train enough profiles of this level, and technology companies compete fiercely not only on salaries, but on research autonomy, access to state-of-the-art computational infrastructure and the possibility of working on projects that define the global state of the art.

For Google, the situation raises a fundamental strategic question: can a company of its size, with the corporate culture typical of a large consolidated tech firm, compete in terms of appeal with startups that offer both the promise of transformative impact and that of sudden enrichment? The acquisition of Character.AI, designed in part as a mechanism to retain Shazeer, suggests that Google is aware of the problem and is willing to pay astronomical prices to solve it. However, recent events indicate that money, on its own, is not enough.

From the perspective of agentic AI —the thematic core of this newsletter—, these departures have direct implications. Both Anthropic and OpenAI are intensifying their bets on agentic systems: models capable of planning, executing multi-step tasks, using external tools and operating autonomously in complex environments. Hiring researchers with experience in the development of Gemini —a model designed precisely to compete in that space— could significantly accelerate Anthropic's agentic capabilities. John Jumper, for his part, brings a scientific depth in structural and biological reasoning that could be transferred to agentic applications in specialized domains such as materials science or drug discovery.

As for Google, the question is whether the loss of these profiles will affect Gemini's roadmap and the company's competitive position in the foundational model market. Google has historically demonstrated a remarkable ability to regenerate its scientific talent, but the current pace of departures —especially of such visible figures— may also affect the external perception of the company as a destination for ambitious researchers, creating a snowball effect that is difficult to stop.

In the absence of official statements from Google —TechCrunch indicates that it contacted the company to obtain its version—, the picture that emerges is that of a company which, despite its dominant position in infrastructure, data and capital, is losing the battle to retain the scientists who shape the future of artificial intelligence. If the trend continues, it could have lasting consequences for its ability to maintain technical leadership in a field where people —more than resources— remain the fundamental differentiator.