Noam Shazeer, a top AI researcher at Google DeepMind, has joined OpenAI. Shazeer was one of the architects behind the transformer technology that powers modern language models. His departure signals another win for OpenAI in the AI talent wars.

Shazeer is best known for co-authoring the seminal 2017 paper Attention Is All You Need. That paper introduced the transformer architecture. Nearly every large language model today relies on this technology. The paper revolutionized artificial intelligence.

Google acquired Character.AI in 2024 for $2.7 billion largely to bring Shazeer back. The company had previously left Google for a startup venture. Bringing him back cost the tech giant billions. His latest departure represents a significant loss for Google.

At OpenAI, Shazeer will serve as Lead for Architecture Research. He will explore next-generation AI model architectures. His role involves driving evolution of the transformer technology he helped create. The position suits his expertise precisely.

Shazeer studied mathematics and computer science at Duke University. He won a gold medal at the International Mathematical Olympiad. His academic pedigree is exceptional. Early achievement predicted his later contributions to AI.

At Google, Shazeer worked on Gemini, Google’s flagship AI model. He served as vice president of engineering. He co-led the Gemini development effort. This executive experience complements his technical background.

His departure from Google signals the company may be losing ground in the AI race. Top talent moving elsewhere indicates internal concerns. OpenAI attracts researchers with its mission and resources. Google’s bureaucracy may feel constraining to some.

OpenAI has become a magnet for AI talent. The company’s focus on frontier AI research appeals to ambitious scientists. Significant compensation packages help. But the mission matters to many researchers. Shazeer’s move comes as AI companies battle for talent. Money alone doesn’t always win. Culture, mission, and direction matter deeply.