Noam Shazeer, a vice president of engineering at Google and a co-lead of the Gemini model family, announced on June 18 that he is leaving Google to join OpenAI. Sam Altman confirmed the move publicly, writing that Shazeer is one of the people he has wanted to work with since the earliest days of OpenAI. Shazeer has not disclosed a start date or specific role.
Shazeer is one of eight co-authors of the 2017 paper 'Attention Is All You Need,' the work that introduced the Transformer architecture underneath every major large language model in production today — GPT, Gemini, Claude, Llama, DeepSeek, and the rest. Google paid roughly $2.7 billion in late 2024 to bring him back from Character.AI, the startup he co-founded after first leaving Google in 2021. That reverse-acquihire was framed at the time as Google securing the talent it needed to keep Gemini competitive with GPT and Claude.
The timing is awkward for both sides. Google's Gemini 3.5 Pro slipped past the June general-availability date Sundar Pichai promised at I/O, and the company is leaning heavily on Gemini revenue to justify its 2026 capex. OpenAI is preparing to file confidentially for an IPO in the coming weeks at a private-market valuation around $730 billion, and a marquee hire from Google reinforces its narrative that the best researchers still want to ship there. Anthropic, which captured close to a quarter of the business AI subscription market in May, sat out the bidding entirely.
Takeaway for learners: the AI talent market behaves more like elite sports than like normal software hiring, and a single principal researcher can move the strategic needle for an entire company. If you are early in your career, the lesson is not that you need to be Shazeer — it is that depth in a small set of fundamental questions (architecture, training dynamics, inference efficiency) compounds far faster than breadth across tools. The Transformer paper was eight authors and eight pages, and it is still paying dividends nine years on.