Noam Shazeer is best known as one of the co-authors of the 2017 “Attention Is All You Need” paper, which introduced the Transformer architecture that now powers most large language models. He also worked on several major Google AI projects, including LaMDA, before leaving the company in 2021 to co-found Character.AI. He also authored the Sparsely-gated Mixture of Experts (2016) paper, which is popular among the AI community.
After falling behind OpenAI and Anthropic a couple of years ago, Google brought Shazeer back in 2024 as part of a major deal with Character.AI. Through this deal, along with Noam, several other researchers returned to Google DeepMind. More recently, he was a vice president of engineering at Google and a technical co-lead for Gemini.
Today, Noam Shazeer announced on X that he is leaving Google and joining OpenAI. In his post, Shazeer said it was a difficult decision to move on, adding that he was proud of the Google team and what it had built together.
I’m excited to share that I’ll be joining OpenAI and look forward to working with the exceptional team there.
— Noam Shazeer (@NoamShazeer) June 18, 2026
It was a difficult decision to move on. I’m incredibly proud of the amazing team at Google and everything we’ve built together. It has been an honor and a pleasure to…
OpenAI CEO Sam Altman welcomed the move with a post of his own, saying Shazeer was one of the people he had most wanted to work with since OpenAI’s early days.
noam is one of the people I have most wanted to work with since the very beginning of openai.
— Sam Altman (@sama) June 18, 2026
only took 10 years.
i think it will be worth the wait! https://t.co/Hj2fjCywjv
Google has made strong progress with Gemini over the past year, closing the gap with OpenAI in several areas. But losing Noam Shazeer is a major talent setback for them, especially after bringing him back less than two years ago by spending a fortune. For OpenAI, the hire adds one of the industry’s most experienced language model researchers to a team that is already pushing ahead with ChatGPT, Codex, and its next generation of frontier models.