John Jumper Leaves DeepMind for Anthropic

He and DeepMind CEO Demis Hassabis won the Nobel Prize in chemistry in 2024 for AlphaFold, an AI model that can predict the 3D structure of proteins based on their genetic sequences. That work reshaped computational biology and turned DeepMind from a research lab with a reputation into a lab with a Nobel on the shelf.

So when the person most associated with DeepMind defining scientific achievement decides to leave for a competitor, it reads as more than a routine hire. It is a signal about where top researchers think the frontier is moving.

In his post, Jumper was gracious about the place he is leaving. He wrote that Hassabis took a real chance letting him lead the AlphaFold team just six months after finishing his PhD, and that the entire GDM team taught him a great deal about how to do great science. He added that GDM is a special place and that he will still be excited to hear about what they discover next.

How AlphaFold actually worked

To understand why Jumper is so sought after, it helps to understand what AlphaFold solved.

Proteins are chains of amino acids that fold into specific three-dimensional shapes, and that shape determines what the protein does in the body. Predicting that shape from the underlying genetic sequence had been an open problem in biology for decades.

AlphaFold attacked it as a prediction task. Given the sequence, the model outputs the likely folded structure, collapsing work that used to take labs months or years into a computational step. The achievement was not a single clever trick but the demonstration that a deep learning system could reach scientifically useful accuracy on a hard, real-world problem.

That is the kind of capability every major AI lab now wants in-house: not just chatbots, but models that produce verifiable scientific results.

Jumper is not the only one leaving

The bigger story is that Jumper is part of a cluster of exits.

Character AI co-founder Noam Shazeer also announced this week that he is leaving DeepMind. In Shazeer case, he is joining OpenAI rather than Anthropic, which means two well-known figures are leaving the same lab in the same week for two different rivals.

According to Bloomberg, Jumper was a key member of Google team developing coding tools, products the company has reportedly struggled to sell to businesses. That detail matters. It suggests the moves are happening not only at the pure-research layer but also around the commercial AI products Google has been trying to push into the enterprise.

When a Nobel laureate and a co-founder of a major AI startup leave the same lab in the same week for two different competitors, treat it as a leading indicator of where frontier talent expects the next breakthroughs to happen, not as isolated resignations.

What to watch over the next 24 to 72 hours

Expect the immediate conversation to center on what Jumper will actually do at Anthropic.

Anthropic has been known primarily for Claude and for a heavy emphasis on AI safety. Bringing in the architect of AlphaFold raises an obvious question about whether the company is widening its ambitions toward AI for science, not just conversational and coding models. Watch for any statement from Anthropic clarifying his role.

Watch DeepMind response too. Losing two high-profile names in a single week invites questions about retention, and Google may move to reassure staff and the public that its research pipeline is intact.

• Look for an official Anthropic announcement describing Jumper focus and team.
• Track whether more DeepMind researchers follow, which would turn two exits into a trend.
• Note any signal that Google reshuffles or doubles down on the coding tools Jumper worked on.

For now, the facts are narrow but striking. Jumper says he is leaving after nearly nine years, he is heading to Anthropic, and Shazeer is heading to OpenAI. The talent at the very top of AI is in motion, and the labs that capture it are betting that people, more than any single model, decide who leads next.

Source: TechCrunch