Google's Top AI Researcher Shazeer Joins OpenAI
Alina Collins
Google DeepMind researcher Noam Shazeer is leaving for OpenAI — his second departure from Google in four years, and a signal that even a $2.7 billion talent deal couldn't keep a co-architect of the Transformer.
Who is Shazeer, and why does the industry care?
Noam Shazeer co-authored the 2017 Transformer paper — the foundational architecture behind every major large language model today, from GPT to Gemini to Claude.
In plain terms = every AI chatbot you use runs on a framework he helped invent.
OpenAI informed select employees of the hire on Wednesday, according to The Information.
Why is his career path called "winding"?
In 2021, Shazeer left Google to co-found Character.AI, a chatbot startup.
In 2024, Google brought him back into DeepMind through a $2.7 billion deal — This means → Google was willing to pay an extraordinary sum to reclaim him, a clear signal of how irreplaceable it considered him.
Now he is leaving again for OpenAI — two departures from Google in under two years, each time to a direct competitor.
What does this mean for Google and OpenAI?
For Google: a $2.7 billion talent buy-back has failed. This reflects a structural difficulty in retaining top AI researchers, even when money is no object.
For OpenAI: it gains a researcher with foundational contributions to the architecture that underpins large models — a direct boost to its technical depth.
This means → the AI talent war has moved past the stage where the highest bid wins. Researchers are voting with their feet for the platform where they believe the next breakthrough will happen.
Content is for reference only, not financial advice.