Nobel Prize in Chemistry Swaps DeepMind for Anthropic: What John Jumper's Hiring Signals for the Scientific AI Race
The creator of AlphaFold leaves Google after nine years. Anthropic has been building an AI operation for science, with its own labs and partnerships with the Allen Institute and Howard Hughes Medical Institute.
John Jumper, the 2024 Nobel Prize winner in Chemistry and architect of AlphaFold, announced on Friday, June 19, on his X account, that he is leaving Google DeepMind after nearly nine years to join Anthropic. Bloomberg confirmed the change with sources at Google, while Anthropic chose not to comment. Jumper, who won the Nobel alongside Demis Hassabis and David Baker, has yet to disclose what his role will be at his new employer.
This departure marks the most symbolic individual hiring in the AI industry since Andrej Karpathy, co-founder of OpenAI, announced in May his own move to Anthropic with the mission of "returning to R&D" in pre-training. However, Jumper operates on a different vector: he built, within DeepMind, the unit that scientifically credentialed the company the most, with AlphaFold generating predictions for over 200 million proteins and Isomorphic Labs valued, according to the latest funding round, at $7.9 billion.
The Operation That Was Waiting for a Name
Over the past eighteen months, Anthropic has quietly built a division of AI for science that seemed to be waiting for a leader like Jumper. In February of this year, the company announced partnerships with the Allen Institute for computational neuroscience and with the Howard Hughes Medical Institute for biomedical discovery. In March, it opened its first wet lab in San Francisco, hired three senior fellows with backgrounds in computational chemistry, and published a paper on Claude agents conducting bench experiments through integrations with Benchling. The company's run-rate revenue jumped from $9 billion in December 2025 to over $30 billion in June, according to figures released by CFO Krishna Rao during an investor conference.
The question is no longer whether Anthropic intends to compete in territory that was exclusive to DeepMind but whether it can deliver a second Nobel. "Achieving a fundamental scientific discovery requires a completely different structure than serving an API," said Patrick Mineault, a former DeepMind researcher now at Mila, in a comment to Nature in February. Jumper was explicitly mentioned at that time as the type of talent that defines the ceiling of the operation.
The Counter-Argument That Needs to Be Made
Not all industry veterans buy the narrative that pure science can be done at a company valued today at $965 billion with a confidential IPO at the SEC. Yann LeCun, Chief AI Scientist at Meta, published again on June 17 that "language models are not the path to real scientific discoveries; they are sophisticated statistical retrievers." For LeCun, Anthropic’s bet on AI for science recycles the publicity of AlphaFold without replicating the fundamental physics work that underpins it. The evidence that undermines Anthropic’s argument is concrete: of the $7.5 billion in annual revenue for Isomorphic, less than 8% comes from clinical stage drug discovery, according to disclosures from Google in Q4 2025. The rest consists of research partnerships and licensing, suggesting that the economic transformation of AlphaFold is still years away from quarterly balance sheets.
The Cross-Reading: What Changes for Big Pharma, Consulting Firms, and Banks
For the pharmaceutical industry in the United States, Jumper's hiring accelerates licensing negotiation strategies: Pfizer and Merck, which are already evaluating multi-year contracts with Isomorphic, now need to consider whether a direct competitor at Anthropic alters commercial terms. In Germany, BASF and Bayer, which signed a research agreement with DeepMind for the discovery of industrial catalysts in April, will closely watch the stability of their chosen partner after such a significant departure.
For the Big 4 and MBB, the interpretation is different. McKinsey QuantumBlack, BCG X, and Accenture Applied Intelligence currently sell scientific discovery packages based on partnerships with DeepMind or OpenAI. Jumper’s move diversifies the pool of strategic suppliers, which typically reopens corporate RFPs. Investment banks covering pharma, particularly Goldman Sachs and Morgan Stanley, have already begun reviewing target prices for Alphabet considering the risk of prolonged talent drain.
Anthropic, with 1,000 enterprise clients spending over $1 million per year on Claude, currently operates in a regime where each senior hire rearranges the institutional board beyond it. Jumper transitions from researcher to capital signal, and the race for scientific AI shifts from a duel between Alphabet and Anthropic to a triangulation with OpenAI in the middle, still lacking equivalent weight in biology.