Noam Shazeer, the Google engineer who co-authored "Attention Is All You Need" — the 2017 paper that introduced the Transformer architecture underpinning nearly every modern AI system — is leaving the company for OpenAI, multiple outlets including Reuters, the Decoder, and Computerworld reported this week. The move is being called the biggest AI talent story of the year.
Shazeer most recently served as a vice president of engineering at Google and co-led the company's Gemini model efforts alongside Jeff Dean and Oriol Vinyals. "It was a difficult decision to move on," Shazeer wrote, according to the Decoder, in a message confirming his departure. For more context on this story, see our ongoing AI industry coverage.
A Decades-Long Career at Google
Shazeer's relationship with Google stretches back more than two decades. He joined the company in 2000 and contributed to core products, including improvements to the search engine's spell checker — work that, at the time, was considered among the most sophisticated language-processing systems in widespread commercial use. Over the years, his name appeared on a string of influential papers spanning everything from mixture-of-experts architectures to large-scale language modeling.
He later became a central figure in Google's deep-learning research. In 2017, he was one of the eight researchers who authored "Attention Is All You Need," the Google paper that proposed the Transformer model. That architecture replaced earlier recurrent and convolutional approaches and became the foundation for large language models like GPT, Claude, and Gemini. The paper has since become one of the most cited works in the history of computer science.
The Character.AI Detour
In 2021, Shazeer left Google to co-found Character.AI, a chatbot startup, with fellow researcher Daniel De Freitas. The company attracted a large consumer following for its conversational personas and reached a multibillion-dollar valuation by letting users chat with AI versions of historical figures, celebrities, and fictional characters.
In 2024, Shazeer and De Freitas returned to Google as part of a roughly $2.7 billion deal that brought the founders back along with portions of their research team and a license to Character.AI's technology. The arrangement was widely seen as a creative workaround to acquire talent without triggering the antitrust scrutiny that a traditional acquisition might invite. At the time, Google framed the move as a way to strengthen its reasoning models, which had not yet caught up with those from OpenAI and Anthropic.
Why OpenAI, and Why Now
Shazeer's jump to OpenAI arrives as the company prepares for a public offering. Reuters reported that the researcher would join "IPO-bound OpenAI," linking the hire to the company's broader effort to consolidate talent ahead of a landmark stock-market debut. Analysts and commentators have read the recruitment as an explicit signal that OpenAI wants to lock in the people most associated with the technology it sells, particularly as it courts investors.
Jim Cramer described the hire as a "coup," according to 24/7 Wall St., framing it within an intensifying contest for the small pool of researchers who understand frontier model training at the deepest level. Fast Company and Quartz both reported the departure as a major loss for Google's Gemini team.
The Talent War Intensifies
The move fits a broader pattern of high-profile defections reshaping the AI industry. The Decoder noted that Shazeer's departure ranks alongside Andrej Karpathy's decision to join Anthropic as one of the most consequential talent shifts of the year. OpenAI has separately hired Dean Ball, the lead author of the Trump administration's AI Action Plan, to lead a new Strategic Futures team — a signal that the company is investing in policy expertise as much as raw research firepower.
For Google, losing a Gemini co-lead is a significant blow at a moment when the company is racing to prove its models can match or exceed those of OpenAI and Anthropic. Gemini has been positioned as Google's flagship response in the frontier-model competition, and Shazeer was central to that effort. His exit leaves Jeff Dean and a reshuffled leadership group to carry the program forward.
For OpenAI, adding one of the original architects of the Transformer is as much a branding victory as a technical one. Few researchers carry the credibility of a co-author of the paper that effectively created the modern AI era — and that pedigree carries weight with both customers and regulators.
What Comes Next
Shazeer's specific role at OpenAI has not been fully detailed in public reporting, but his background in both foundational research and large-scale model deployment suggests he will work on next-generation model development. His arrival also deepens OpenAI's bench as it navigates an increasingly competitive and politically charged landscape, one marked by export controls, global standards debates, and a looming IPO.
The hiring underscores a simple reality of the current AI industry: the technology may be advancing rapidly, but the people who truly understand how to build it remain extraordinarily rare — and extraordinarily valuable. In a market where a single researcher can reshape a product roadmap, Shazeer's move to OpenAI is a reminder that the talent war is now inseparable from the race for AI supremacy itself.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
