惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Y
Y Combinator Blog
GbyAI
GbyAI
U
Unit 42
WordPress大学
WordPress大学
Last Week in AI
Last Week in AI
P
Proofpoint News Feed
D
DataBreaches.Net
N
Netflix TechBlog - Medium
H
Hackread – Cybersecurity News, Data Breaches, AI and More
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
C
Check Point Blog
Martin Fowler
Martin Fowler
月光博客
月光博客
MongoDB | Blog
MongoDB | Blog
MyScale Blog
MyScale Blog
The Cloudflare Blog
Apple Machine Learning Research
Apple Machine Learning Research
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
M
MIT News - Artificial intelligence
云风的 BLOG
云风的 BLOG
罗磊的独立博客
B
Blog RSS Feed
J
Java Code Geeks
The GitHub Blog
The GitHub Blog

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
The Nobel Laureate Who Joined Anthropic Mid-Crisis
Peremptory · 2026-06-26 · via DEV Community
Cover image for The Nobel Laureate Who Joined Anthropic Mid-Crisis

Peremptory

John Jumper announced on June 19 that he is leaving Google DeepMind to join Anthropic. He shared the 2024 Nobel Prize in Chemistry for protein structure prediction, spent nearly nine years at DeepMind, and co-created AlphaFold, the model that mapped more than 200 million protein structures. By any normal measure, that biography makes him the most decorated scientist ever to change employers mid-career in the AI industry.

The timing is the detail I keep returning to. Jumper announced his departure one day after Noam Shazeer, the Transformer co-author, left Google for OpenAI. Two scientists central to the architectures and models that define the field, gone from Google in the same week. That is a bad week for any company to survive.

But the stranger part is where Jumper landed. Anthropic's Claude Fable 5 and Mythos 5 were still offline, thirteen days into a US government export-control ban when June 25 arrived. The US Commerce Department had barred Anthropic from distributing the models to foreign nationals, and because Anthropic couldn't filter users fast enough, it pulled both models globally. Fable 5 down. Mythos 5 down. API calls returning errors. A Nobel-caliber hire showed up to a company whose two best models were literally unavailable.

The conventional read is that the talent move signals confidence. And it does. Jumper could have gone anywhere, including staying put. He chose a company fighting the US government in court, with its flagship products suspended. That choice says something about where serious scientists think the real work is happening.

My read is slightly different. I think the Anthropic the offer letter described is a different Anthropic than the company in the current headlines. Throughout 2026, Anthropic has been building quietly toward AI for science: opening wet labs, publishing research on AI agents designed for biological workflows, forming partnerships with the Allen Institute and the Howard Hughes Medical Institute's Janelia Research Campus. Those partnerships plug Claude-powered agents directly into genomics and imaging pipelines. Dario Amodei has made no secret of the bet, writing in 2024 that AI-enabled biology might compress fifty years of scientific progress into a decade.

Jumper is the person who has most concretely demonstrated that kind of compression is possible. AlphaFold didn't incrementally improve protein structure prediction. It collapsed decades of expected work into a deployable model. Bringing him in is a bet that the AI-for-science story Anthropic has been narrating is actually going to happen, and that the person who already made it happen once might know how to do it again.

What Google now faces is a specific irony. DeepMind built scientific credibility through AlphaFold. That credibility became the primary proof of concept that its researchers carry into competitors. Demis Hassabis praised Jumper publicly on X, describing AlphaFold as work that "changed the world." That is a gracious exit and also a precise description of the problem: the credential walks out with the person.

For Anthropic, Jumper's arrival carries weight that benchmark scores cannot. Gemini 2.5 Pro Deep Think just posted 82.4% on GPQA Diamond, the graduate-level science benchmark, ahead of Fable 5. Google can still claim the current science leaderboard. But leaderboards change. Scientists who have already done the definitive work in a field are harder to replicate.

The Fable 5 ban will end. Models come back. What Jumper brings is the question of what Anthropic is actually building toward, and he's the clearest answer the company has given yet.