共有:

RESEARCH TALENT SHIFT

AlphaFold's brains are moving to Anthropic

After John Jumper, the scientist who led AlphaFold, two of his co-authors are also reported to have moved to Anthropic. DeepMind's core team has effectively been dismantled, and the center of gravity in life-science AI research is quietly shifting.

AI Navigate Editorial2026.07.306 min read

Google DeepMind Anthropic AlphaFold
01
WHY NOW

Why this matters now

From an individual departure to a team's dismantling

On June 20, 2026, the news that John Jumper — the 2024 Nobel laureate in Chemistry who led AlphaFold's development for roughly nine years — had left Google DeepMind for Anthropic sent shockwaves through the industry. But that may have been only the opening act. According to a report by the Financial Times on July 29, 2026, AlphaFold co-authors Jonas Adler and Alexander Pritzel have also moved to Anthropic, reportedly following Jumper. Citing that report, an article from The Decoder states that Google DeepMind had already dismantled its dedicated AlphaFold team, having reassigned most of the original paper's authors to other roles over the past year.

AlphaFold, which has predicted and mapped more than 200 million protein structures to date, was never just a research output — it was the emblem of DeepMind's own narrative that AI could transform science. The fact that its core members have left for a competitor, and that the team itself had already been effectively dismantled, strongly suggests that the center of gravity in life-science AI research is shifting from Google to Anthropic. An analysis published by Fortune on June 23, 2026 also notes that Anthropic, OpenAI, and xAI have spent the past 18 months steadily poaching researchers from Google, Meta, Microsoft, and Amazon — meaning this episode should be read not as an isolated incident, but as part of a much larger talent flow.

02
BY THE NUMBERS

The shift, in numbers

200M+
Protein structures predicted by AlphaFold to date
9 yrs
Years Jumper led AlphaFold's development
3
Core AlphaFold members now reported at Anthropic
18 mo
Period Anthropic, OpenAI, and xAI have kept poaching researchers from Big Tech
03
TIMELINE

Timeline of the departures and dismantling

01

2026.06.20 — Jumper moves to Anthropic

The move by a Nobel laureate who had become the public face of AlphaFold was initially read as one prominent researcher's personal decision.

02

2026.07.29 — The dismantling comes to light

The Financial Times report revealed both the departures of Adler and Pritzel and the effective dismantling of DeepMind's dedicated AlphaFold team. What had looked like an individual move turned out to be a shrinking of the organization itself.

03

Throughout 2026 — Anthropic builds its own structure

In parallel, Anthropic has opened a wet lab, published research on AI agents for biology workflows, and formed research partnerships with the Allen Institute and the Howard Hughes Medical Institute — steadily assembling an "AI for Science" operation.

04
WHO IT AFFECTS

Who it affects, and how

Distinct angles for executives, PMs, and partnerships teams

Executives & business leaders

The baseline for vendor selection and partnership evaluation in life sciences is starting to move. Treating "AlphaFold = Google" as a given is no longer safe — it's worth revisiting at the next partnership decision.

Product managers

For anyone mapping AI roadmaps in drug discovery or biotech, this is one more signal to monitor: what science-focused products or APIs Anthropic ships next.

Procurement & partnerships

Tracking Anthropic's ties to the Allen Institute and Howard Hughes Medical Institute shows which research institutions and customer segments it is prioritizing in life sciences.

05
WHAT'S NEXT

What happens next, and what to do

Individual researcher moves will likely continue sporadically. But the real story is that Anthropic isn't just hiring — it's simultaneously building a wet lab, forming institutional partnerships, and publishing agent research. Whether this converts into actual products — scientist-facing APIs, productized drug-discovery partnerships — is the next thing to watch.

  1. If you're evaluating AI vendors in drug discovery or biotech, track the progress of Anthropic's science research effort on a quarterly basis.
  2. If your company has Google/DeepMind-related contracts, check with your account team whether this organizational shift affects the continuity of existing products such as the AlphaFold Server.
  3. Given how fluid talent movement is right now in life-science AI, review retention plans for people in that space that your company wants to keep.
06
DEEPMIND VS ANTHROPIC

An asymmetry in structure

Google DeepMind (now)Anthropic (now)
Most of AlphaFold's original creators have left (Jumper, Adler, Pritzel, and others)Key members, led by Jumper, have joined
Dedicated team effectively dismantled; most authors reassigned elsewhere (per FT)Building new capacity: wet lab, biology-agent research, and more
Few notable new science-institution partnerships announcedPartnerships with the Allen Institute and Howard Hughes Medical Institute
AlphaFold itself continues as an existing assetNo concrete new scientist-facing product announced yet (unverified)

It would be premature to read this with pure optimism. There's a gap between hiring prominent researchers and actually shipping products usable for drug discovery or structure prediction. The assets, data, and compute infrastructure behind AlphaFold still sit with Google — talent alone doesn't guarantee they can be replicated. What this reporting shows is a shift in where the brains are, not yet a proven shift in where the capability is. It's also worth noting that much of this departure information is report-based, not an official statement from either company.