The architect behind the technology underlying nearly every modern large language model is switching sides, and the industry is taking notice. Shazeer confirmed the move himself on X: "I'm excited to share that I'll be joining OpenAI and look forward to working with the exceptional team there."
A $2.7 Billion Retention Deal That Didn't Last
Reported cost of Google's 2024 Character.AI deal to bring Shazeer back
Year the "Attention Is All You Need" Transformer paper was published
Time between Shazeer's Google return and his departure for OpenAI
Shazeer spent 2000 to 2021 at Google Brain, where he co-developed the T5 model and led the Meena and LaMDA projects that later became the foundation for Bard. He left Google in 2021 to co-found Character.AI, only to return to Google in late 2024 through an unusual deal: rather than a straightforward acquisition, Google licensed Character.AI's technology and brought Shazeer and a slice of his research team back in-house, a structure that drew regulatory scrutiny at the time.
Shazeer co-invented the very architecture that the "T" in GPT stands for. His move to the company that commercialized the Transformer most successfully carries outsized symbolic weight in the AI research community.
Why OpenAI Wants Him Now
OpenAI is widely reported to be heading toward an IPO, and hiring a researcher of Shazeer's stature serves a dual purpose: advancing work on new model architectures while signaling to the market that top talent wants to be there. OpenAI has spent the past year aggressively recruiting from rivals while expanding distribution deals that put its models in front of far more enterprise customers.
The departure comes as Google intensifies its enterprise AI competition with OpenAI and Anthropic, shortly after rebuilding its AI platform around agents at Cloud Next. Losing a Gemini co-lead now stings more than usual.
A New Data Point in the Talent War
The move has become a talking point about whether enormous compensation packages can actually buy loyalty. Google's reported $2.7 billion effort to reacquire Shazeer and his research team was held up as proof that the giants would pay almost anything to keep frontier-model talent. Less than two years later, that bet didn't hold — suggesting mission and research direction may matter more than capital in retaining top AI researchers.
| Timeline | Event |
|---|---|
| 2000–2021 | Shazeer at Google Brain; co-develops T5, leads Meena/LaMDA |
| 2021 | Leaves Google, co-founds Character.AI |
| Late 2024 | Returns to Google via ~$2.7B Character.AI deal, named Gemini co-lead |
| June 18, 2026 | Announces move to OpenAI |
- Transformer co-inventor Noam Shazeer is leaving Google's Gemini project for OpenAI
- The move comes less than two years after a ~$2.7 billion deal brought him back to Google
- OpenAI gains both research capability and a market signal ahead of its anticipated IPO
- Google faces a leadership gap in Gemini just as enterprise AI competition intensifies
Neither company has disclosed the exact terms of Shazeer's new role, his start date, or what specifically he'll work on at OpenAI beyond new architectures. What's confirmed is the announcement itself — and what it represents: less than two years after a multibillion-dollar effort to bring him home, Google's Gemini co-lead is walking across the street to its biggest rival.