Google lost two more senior AI researchers to Anthropic this week, with Jonas Adler and Alexander Pritzel both confirming they are leaving the company. Bloomberg and TechCrunch reported the departures on June 24, 2026, describing both as key contributors to Google’s Gemini model.

The departures are the fourth and fifth significant exits from Google’s AI teams in under a week, following Nobel laureate John Jumper’s move to Anthropic and legendary researcher Noam Shazeer’s decision to join OpenAI.

Who Is Leaving and What They Built

Adler worked on Google’s AI coding effort, the technical work behind making Gemini useful for software development at scale. Pritzel was focused on pretraining, the foundational stage where a model learns from enormous volumes of data before it is fine-tuned for specific tasks. Both also contributed to Google DeepMind’s AlphaFold research alongside Nobel Prize winner John Jumper.

These are not peripheral roles. Pretraining quality determines the ceiling of what a model can do. Coding capability has become one of the primary battlegrounds between frontier labs. Losing the people who built those systems is not a symbolic setback, it is a capability question.

Why Researchers Are Moving

Two forces appear to be driving the departure wave.

The first is the IPO opportunity. Anthropic and OpenAI are both expected to go public, and the chance to join before that happens carries meaningful financial upside. A company already worth two trillion dollars cannot replicate that for new hires. A company approaching its public debut can.

The second is computing access and internal politics. Shortly before Noam Shazeer announced his move to OpenAI, Google reportedly reassigned computing capacity from one of his projects to another DeepMind team. Researchers who need raw compute to do their best work are sensitive to these decisions. When access gets rationed, the appetite to move increases.

Anthropic has been on an aggressive hiring run through 2026. The company has brought in researchers from DeepMind, Google Brain, and other AI organizations while simultaneously raising capital and building infrastructure for serious scientific applications.

The Broader Pattern

In the span of a few days, Google has seen:

• John Jumper (Nobel Prize in Chemistry, AlphaFold co-creator) leave for Anthropic
• Noam Shazeer (co-author of the original Transformer paper, 25-year Google veteran) leave for OpenAI
• Jonas Adler (AI coding, AlphaFold contributor) leave for Anthropic
• Alexander Pritzel (pretraining, AlphaFold contributor) leave for Anthropic

Google DeepMind remains one of the most capable AI research organizations in the world. This is not a collapse. But the pattern of movement is consistent and worth noting: the most accomplished researchers are choosing pure-play AI labs over the AI divisions of large technology companies.

What This Means for Business

If you are making decisions about which AI platforms to bet on, or trying to understand where the frontier is moving, this matters in a few ways.

Anthropic is assembling serious depth. The company has gone from a well-funded AI safety lab to a research organization that is pulling talent from the top of the AI industry. The researchers joining are not generalists. They built foundational systems. That kind of capability tends to compound.

Coding and pretraining are the strategic battlegrounds. The two specific areas Adler and Pritzel worked in, AI coding tools and model pretraining, are the same areas where enterprise AI competition is most intense right now. Claude has been gaining ground in enterprise coding deployments. Reinforcing that capability with people who built Gemini’s coding systems is a pointed move.

The talent signal matters more than the headlines. High-profile researcher departures tend to generate news cycles, then fade. The thing worth tracking is not the names but the direction. Researchers at this level pick their next employer carefully. When multiple senior people make the same choice within the same week, it reflects something about where they see the most interesting and consequential work happening.

For most businesses, the practical question is which AI tools and platforms to use, not who built the underlying models. But the underlying talent moves determine which platforms stay ahead. Watching this is a reasonable part of your AI strategy.

---

Enterprise DNA helps business owners and data professionals make sense of the AI landscape. If you want help evaluating AI platforms or building a strategy for your organization, book a call with Sam McKay.