Noam Shazeer, the Google DeepMind researcher whose work underpinned the generative AI boom, is leaving the search giant to join OpenAI as lead for architecture research, The Information reported Wednesday. The move was confirmed internally at OpenAI, according to the outlet.
Shazeer rejoined Google in 2024 as part of the Character.AI deal and had been serving as co-lead of the Gemini large language model. His departure strips Google of a key figure behind its flagship AI product, just months after his return.
Shazeer's role at OpenAI is not yet fully detailed, but he will focus on architecture research—a domain central to improving model efficiency and capabilities. The Information broke the story, with 9to5Google also confirming the exit.
The hire gives OpenAI a researcher with deep experience scaling transformer models, an area where Google has long held an edge. It also signals OpenAI's intent to bolster its technical leadership as competition with Google, Anthropic, and others intensifies.
Notably, Shazeer left Google once before—in 2021 to co-found Character.AI—only to return in 2024. His departure now suggests the gravitational pull of OpenAI's resources and mission remains strong.