ai-trend-notifier
← wiki

$ cat wiki/people/noam-shazeer.md

Noam Shazeer

personupdated 2026-06-24created 2026-06-24

Overview

Research scientist and engineer; VP Engineering at Google DeepMind; co-lead of the Gemini AI models; co-author of the 2017 paper "Attention Is All You Need" — the paper introducing the Transformer architecture that underpins virtually every major LLM.

  • Current affiliation (as of June 2026): OpenAI (announced June 18, 2026)
  • Previous affiliation: Google / Google DeepMind (VP Engineering, Gemini co-lead)
  • Prior: Character.AI co-founder (with Daniel De Freitas)

Career Path

PeriodRole
~2000s–2021Google (senior researcher, foundational work on Transformer, Mixture-of-Experts)
2021–2024Character.AI (co-founder)
Aug 2024Google DeepMind — returned via ~$2.7B Google acquisition of Character.AI team
Jun 18, 2026OpenAI — announced departure from Google to join OpenAI

Key Contributions

  • "Attention Is All You Need" (2017) — co-authored with Vaswani et al. at Google Brain; introduced the Transformer architecture. Every major LLM (GPT, Claude, Gemini, Llama) is built on this foundation.
  • Mixture-of-Experts (MoE) — early theoretical and applied work on sparse MoE architectures
  • Gemini co-lead — co-led development of Google's frontier Gemini model series at Google DeepMind
  • Character.AI — co-founded the consumer AI chat company (valued at $1B+ before Google re-acquisition)

Notable Statements

  • Sam Altman on Shazeer joining OpenAI: "one of the people I have most wanted to work with since the very beginning of OpenAI" — "only 10 years" in the making (source)

Significance

The Shazeer hire by OpenAI — announced June 18, 2026 — is among the highest-profile talent moves in AI history. Combined with John Jumper joining Anthropic on June 19, Google lost two pivotal AI figures within 48 hours.

The move is doubly significant: Google paid approximately $2.7 billion to acquire Shazeer's Character.AI team back in August 2024; less than two years later he left for a direct competitor.

Related

Referenced by

Sources