Noam Shazeer, vice president of engineering at Google and co-lead of its Gemini AI models, announced on Wednesday that he is leaving the company to join OpenAI. He shared the news in a post on X. “I’m excited to share that I’ll be joining OpenAI and look forward to working with the exceptional team there,” he wrote, adding that the decision had been a difficult one. “I’m incredibly proud of the amazing team at Google and everything we’ve built together.”
His exact role at OpenAI has not been disclosed. Google responded to the news in a statement to Reuters: “We are grateful for Noam’s meaningful contributions to Google over the years.”
The departure is notable given how recently Shazeer returned to Google. He had left the company in 2021 to co-found Character.AI, the conversational AI startup, where he served as CEO. In August 2024, Google struck a licensing deal with Character.AI and brought Shazeer back, reportedly paying around $2.7 billion as part of an arrangement that also saw a number of his researchers rejoin Google. He was appointed co-lead of Gemini’s development shortly after, a role he has held since.
Shazeer is widely regarded as one of the central figures behind Gemini’s progress in narrowing the gap with OpenAI’s own ChatGPT, which makes his move to the rival company particularly striking. His history with Google goes back much further than his recent return: he first joined the company in 2000 and was a co-author of the influential 2017 research paper that introduced the transformer architecture, a foundational piece of work credited with catalysing the broader generative AI boom that followed.
The move adds to a string of high-profile talent shifts across the AI industry, where leading companies are competing as intensely for researchers as they are for compute and model performance.