The Nobel scientist behind AlphaFold leaves DeepMind for Anthropic
John Jumper shared a 2024 Chemistry Nobel for cracking protein folding with AlphaFold. Nine years at DeepMind, and now he is walking across the street to a rival lab that just bought a biotech startup.
Two senior names left Google DeepMind in two days: Noam Shazeer, then Jumper. The thread treated it less as one departure than as a pattern, and started guessing what is going wrong inside Google.
One read tied the move to Anthropic's recent purchase of a biotech outfit and a bet on using AI to discover drugs. A skeptic countered that the whole industry is now a revolving door priced around the next IPO.
A protein-folding laureate crosses to a competitor
The news arrived as a one-line tweet, and the thread filled in everything around it. John Jumper is not a typical lab hire: with Demis Hassabis he shared the 2024 Nobel Prize in Chemistry for AlphaFold2, the model that predicted the structures of nearly all 200 million proteins known to science and has since been used by more than two million researchers in 190 countries. After nine years at Google DeepMind, he is joining Anthropic.
Hassabis posted a farewell on X, quoted in the thread:
"Thanks John for an extraordinary partnership and wonderful collaboration over the past 9 years! What we achieved with AlphaFold changed the world…" — as relayed in the thread
Two exits in two days, and a guess at why
What caught the thread was not the hire alone but the timing. Noam Shazeer had departed the day before, and the pattern set off speculation about what is pushing senior people out of Google.
"Something spicy must have happened internally at Google. This rapid fire high level attrition isn't just down to the bureaucratic quagmire." — the chief scientist of Sibylline Software
Others watched for the real signal. The tell, one regular argued, is whether the very top moves:
"The real tell will be if Demis makes a move. Jeff Dean seems more like a lifer to me." — a long-tenured commenter
The drug-discovery thesis
The sharpest explanation reframed the hire as strategy, not poaching. Anthropic recently acquired the biotech startup Coefficient Bio, and a protein-folding laureate fits a specific ambition:
"I am guessing this is related to Anthropic's recent acquisition of Coefficient Bio, and their interest on essentially using AGI to discover novel drugs" — in a reply tracing the acquisition
The talent-stacking drew its own awe — and a hedge:
"Anthropic legit builds one the strongest if not the strongest IC team in the history of computational technology… either we will witness a legendary run, or a new LTCM" — a data-focused commenter invoking the famous hedge-fund blowup
The cynical read: priced to the IPO
Not everyone bought the legend. The bleakest take treated the move as a bet on a payout, not a mission:
"He will be back at Google after the Anthropic IPO… everyone here is easily fooled by the Anthropic hype." — a commenter skeptical of the valuation
Another put it more plainly: "I would just join a lab for 4 years and retire at this point." Whether Jumper's move is about building drug-discovery models or about timing an exit, it lands the same way for Google: the person whose name is on a Nobel is now wearing the competitor's badge.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
Norway approves the world's first full-size ship tunnel through a mountain
For decades, ships have waited days for the Stadhavet sea to calm, the stretch that has killed 33 sailors since WWII. Norway's answer: bore a 1,800-metre tunnel, 49 metres tall, straight through the Stad peninsula so ships sail under the mountain.