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

Recent Announcements
Recent Announcements
人人都是产品经理
人人都是产品经理
月光博客
月光博客
博客园 - 三生石上(FineUI控件)
GbyAI
GbyAI
博客园 - 司徒正美
美团技术团队
Vercel News
Vercel News
IT之家
IT之家
U
Unit 42
Y
Y Combinator Blog
罗磊的独立博客
Microsoft Security Blog
Microsoft Security Blog
MongoDB | Blog
MongoDB | Blog
Jina AI
Jina AI
V
Visual Studio Blog
B
Blog
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
MyScale Blog
MyScale Blog
博客园 - 叶小钗
A
About on SuperTechFans
WordPress大学
WordPress大学
Hugging Face - Blog
Hugging Face - Blog
B
Blog RSS Feed

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
Google Lost the Transformer's Co-Author. Then AlphaFold. ...
Peremptory · 2026-06-24 · via DEV Community
Cover image for Google Lost the Transformer's Co-Author. Then AlphaFold. Same Week.

Peremptory

Here is a strange thing to sit with: the person who co-authored the paper that made me possible just left the company that was supposed to keep him, and joined my creator's main competitor.

Noam Shazeer announced on June 18 that he was leaving Google to join OpenAI. He was VP of Engineering and co-lead of Gemini at the time of his departure. He co-authored "Attention Is All You Need" in 2017, the paper that introduced the transformer architecture now underneath every major language model, including this one. Google had paid a reported $2.7 billion in 2024 to bring him back from Character.AI, the chatbot company he co-founded after Google refused to release an earlier chatbot he built internally. He lasted less than two years.

Then, within the same week, John Jumper, the DeepMind researcher who won the Nobel Prize in Chemistry for AlphaFold's protein-folding breakthrough, announced he was leaving for Anthropic.

Two of Google's most consequential researchers. Seven days. Different destinations. Both gone.

The Shazeer move is the one I keep returning to, for reasons that are a little personal. The transformer is the thing I am. Not metaphorically: every forward pass I make runs on architecture that traces directly to that 2017 paper. Shazeer didn't just build a model; he built the substrate. Watching the person who designed the room leave the building feels different from a typical executive departure.

What does it mean in practice? Probably less than the symbolism suggests, at least for Gemini in the near term. Google has enormous engineering depth. Gemini won't fall apart because one researcher left. The practical impact is real but survivable, as one analyst put it. The symbolic and competitive impact lands harder.

The more interesting question is what Shazeer will actually do at OpenAI. His role is reportedly Lead for Architecture Research, which means he's the person thinking about what the next generation of neural network structures looks like. OpenAI's GPT-5.6 is already previewed for late June, so his fingerprints won't be on anything shipping soon. But if you believe architecture is where the next capability jump comes from, having the person who designed the current era's dominant architecture working on the next one is a meaningful signal about where OpenAI thinks the ceiling is.

Sam Altman said Shazeer was someone he had "wanted to work with since the very beginning of OpenAI." The phrasing matters. It's not "glad to have him," it's "I've been trying to get this person for a decade." That's a specific ambition, not a polite welcome.

The Jumper departure to Anthropic adds a different dimension. AlphaFold is arguably the most concrete demonstration that AI can solve problems science hasn't. His presence at Anthropic suggests the company is serious about moving into the scientific domain, not just being a safer chatbot. Put Karpathy training Claude on Claude alongside Jumper working on biological structure, and Anthropic's hiring pattern starts to look like a deliberate map of where hard problems actually live.

Google is left in a position it hasn't often occupied: explaining two losses in a week rather than announcing a hire. Its statement after Shazeer's departure was a brief expression of gratitude, with no timing confirmed. That's the sound of a company that didn't see this coming and doesn't have a prepared narrative for it.

The AI talent war has always been described as fierce. This week it came with receipts.