Noam Shazeer, co-head of Google Gemini, quits OpenAI: The battle for AI talent cannot be retained even with $2.7 billion

On June 18, the AI ​​industry ushered in a heavyweight talent earthquake:Google Gemini co-head Noam Shazeer announced his resignation and officially joined OpenAI. One of the "Eight Transformers" and co-inventor of the modern large language model infrastructure, less than two years after Google spent $2.7 billion to "recruit" him from Character.AI, he left again.

Who is Shazeer? A "living fossil" figure in the AI ​​world

For those who are not familiar with the AI ​​research circle, the name Noam Shazeer may be unfamiliar - but he is one of the most important founders of contemporary AI. Joined Google in 2000 and co-published a landmark paper in 2017《Attention Is All You Need》, the Transformer architecture proposed in this paper directly gave birth toChatGPT、Gemini、ClaudeWait for all mainstream large language models today. It can be said that without the contributions of Shazeer and others, there would be no generative AI today.

In 2021, Shazeer and colleague Daniel De Freitas left Google to foundCharacter.AI, creating the most popular AI character chat platform at the time. In August 2024, Google will$2.7 billionHe "bought back" the two people together with the team at a sky-high price, and made him the co-head of Gemini model development. The deal was once seen as a key victory for Google in the AI ​​talent war.

Even 2.7 billion cannot be kept: Why leave again?

Shazeer said on the

Although he did not publicly explain the reason for his resignation, there are several backgrounds worthy of attention:

  • OpenAI is about to IPO: OpenAI filed confidentially for an IPO earlier this month, with a possible valuation of as much as $852 billion. For core technical staff who join early, the ceiling for equity incentives is much higher than Google’s salary package.
  • Delay of Gemini 3.5 Pro: The Gemini 3.5 Pro promised by Google I/O has not yet been released, and Pichai also publicly admitted that Google lags behind its competitors in "agent programming."
  • Character.AI Frustration: Shazeer originally left Google because the company was unwilling to radically advance the chatbot project. Now similar differences in product rhythm may happen again.

The battle for AI talent intensifies

Shazeer's job-hopping is the latest epitome of the war for AI talent. From Anthropic’s Dario Amodei (former OpenAI employee) to Inflection’s Mustafa Suleyman (now Microsoft AI CEO), the flow of top AI talents is reshaping the industry landscape at an unprecedented speed.

For OpenAI, winning the co-author of the Transformer paper and the core person in charge of Gemini on the eve of the IPO not only brings technical strength, but also aA strong “vote of confidence”——After Google spent 2.7 billion to poach people, they were still "reverse poached", which shows that OpenAI's attraction to top AI talents is still unmatched.

Summary: A person’s job-hopping is a signal for an industry

Noam Shazeer's switch from Google to OpenAI is ostensibly a personal career choice, but in essence reflects the deep changes the AI ​​industry is undergoing:OpenAI, on the eve of its IPO, is making every effort to recruit the world's top talents, while giants such as Google face not only technological gaps in the large-model arms race, but also challenges to organizational speed and innovation culture.. Who will be the next AI star to leave a big company and join a startup? We'll see.

Continue to track the latest trends in the AI ​​industry, all at AI Dash——Discover the best AI tools.

🔗 Share: Twitter Weibo Copy link

📬 Like this article?

Weekly selected AI tool reviews + practical tutorials, delivered directly to you.

Subscribe to the weekly AI picks →

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top
Tool Picks
1
AI Coding
DeepSeek V4 Pro 0813 in-depth review: 1M context + MoE architecture, flagship performance at an affordable price
8.9
2
AI Coding
Qwen3.8 2.4T A95B in-depth review: 2.4 trillion parameter open source weight, a milestone for Alibaba Tongyi Qianwen
8.5
3
AI Coding
GLM 5.3 in-depth review: Zhipu’s millions of contextual reasoning models, fully upgraded programming agent capabilities
8.4
💻AI Coding 📊AI Productivity 📝AI Writing 🎨AI Image Gen
📬 Weekly AI Picks