Gemini Co-Lead Noam Shazeer Leaves Google for OpenAI
Noam Shazeer, the Google engineering vice president who has co-led the company’s Gemini AI models, is leaving to join OpenAI, according to a statement he made this week confirming earlier reports. The move ends Shazeer’s second stint at Google and sends one of the most recognizable names in AI research to the company’s chief rival. It also deepens a talent war that has reshaped the industry’s research community, with OpenAI and Google trading senior researchers as they race to build the next generation of AI systems.
Shazeer is not an ordinary engineer. He was one of the authors of the 2017 paper that introduced the transformer, the architecture underlying nearly every modern AI system, and he spent years at Google building the technologies that became the foundation of its search and language products. He left Google in 2021 to co-found Character.AI, a startup that let users chat with AI versions of celebrities and fictional characters, which became one of the fastest-growing consumer AI products of its era. Google rehired him in 2024, bringing him back with his team in a deal that valued Character.AI’s technology while the startup’s founders returned to the search giant.
His return was meant to anchor Google’s AI ambitions. Shazeer took a leading role in Gemini, the family of models Google has positioned against OpenAI’s GPT series, and his name became associated with the company’s efforts to stay competitive in the race it had effectively started. His departure now raises questions about that effort’s continuity. Google has said Gemini will continue under its existing leadership structure, and the company’s research bench remains deep, but losing the person most associated with its flagship models is a blow to morale and to the story Google tells about its AI future.
For OpenAI, the hire is the latest in a pattern of high-profile acquisitions from rivals. The company has built its research team by combining its own early talent with senior researchers recruited from across the industry, and Shazeer’s arrival adds one of the few people in the field whose name is known outside it. His expertise in the architecture of large models, and his experience building consumer AI products at Character.AI, gives OpenAI a resource it can apply to both its frontier models and its products. The company has not said what role he will take, and Shazeer did not specify his mandate.
The move also illustrates how the economics of AI research have shifted. A decade ago, the best researchers were employees of large technology companies, doing foundational work as part of teams that published papers and moved between companies rarely. Today, the value of top researchers is measured in the tens of millions of dollars, and the competition for them is conducted with compensation packages, equity and the promise of working on the most ambitious projects in the field. OpenAI’s position as the company with the most visible frontier models makes it a magnet for that talent, and Google’s attempts to hold its own researchers have required matching offers it would not have considered a few years ago.
The stakes of the talent war go beyond individual hires. The research community is small, and the movement of senior people carries knowledge about techniques, priorities and strategy that is hard to value and impossible to protect. When a researcher of Shazeer’s standing changes companies, the teams around them often follow, and the direction of an entire program can shift. Google has lost several senior AI researchers over the past two years, and while the company has continued to produce strong models, the perception that it is losing the talent battle has become part of the industry’s conversation.
There is also a question of what the move says about Google’s position. The company has the resources, the data and the computing power to compete at the frontier of AI, and its Gemini models have closed much of the gap that once separated it from OpenAI. But the flow of people has been more one-directional, and the departure of a figure like Shazeer risks reinforcing the narrative that the most ambitious researchers see OpenAI as the place where the next breakthroughs will happen. Google’s counter is its scale and its ability to deploy AI across products used by billions of people, an advantage no rival matches.
For the industry, Shazeer’s move is another data point in the consolidation of AI talent around a small number of companies. The field’s history has been shaped by the movement of its most creative people, and each high-profile departure reshapes the map. Shazeer has now been on both sides of that process, leaving Google for a startup and returning, and leaving again for its rival. Where he lands inside OpenAI, and what he builds there, will be watched by everyone in the field, because his career has a way of predicting where the technology goes next. This article was prepared by Rhino Finance’s editorial team based on public reporting.
