# Tom Brown

> Tom Brown is an Anthropic co-founder and its Chief Compute Officer, responsible for the technical organization that secures, scales, and uses the computing resources behind Claude. His career moved from MIT and early startups through MoPub, the group-dating company Grouper, a deliberate self-taught turn into machine learning, Google Brain, and OpenAI, where he led the GPT-3 paper. The distinctive thread is not a tidy academic ascent but a willingness to become a beginner, paired with an engineer's instinct for making large systems work.

- **Role:** Co-Founder and Chief Compute Officer at Anthropic
- **Organizations:** Anthropic, OpenAI, Google Brain, Grouper, MoPub, Linked Language, South Park Commons
- **Education:** B.S. and M.Eng.; computer science and cognitive science, Massachusetts Institute of Technology
- **Known for:** Lead author of the 2020 GPT-3 paper 'Language Models are Few-Shot Learners.', Co-author of the 2020 paper identifying empirical scaling laws for neural language models., Co-author of OpenAI's 2017 work on learning from human preferences and creator of its open-source RL-Teacher implementation.

## Career timeline

- **2009** — Joined friends' startup Linked Language as its first employee while at MIT.
- **2010-2011** — Worked as an early engineer at mobile advertising startup MoPub, building early versions of its core server.
- **2012-2014** — Co-founded the Y Combinator company that became Grouper, serving as CTO of the group-dating service.
- **2015** — Joined South Park Commons, set aside a focused self-study period and retrained in machine learning.
- **2016** — Joined OpenAI after roughly six months of focused study; later spent a period at Google Brain before returning.
- **2017** — Co-authored OpenAI's work on deep reinforcement learning from human preferences and released the RL-Teacher implementation.
- **2020** — Co-authored the neural-language-model scaling laws paper and was lead author of the 175-billion-parameter GPT-3 paper, 'Language Models are Few-Shot Learners.'
- **2021** — Co-founded Anthropic with six other former OpenAI colleagues.
- **2025** — Publicly described Anthropic's multi-chip compute strategy and the expansion of AI infrastructure on Y Combinator's Lightcone podcast.
- **2026** — Serves as Anthropic's Chief Compute Officer, leading the technical organization responsible for securing, scaling and effectively using compute.

## Achievements

- Lead author of the 2020 GPT-3 paper 'Language Models are Few-Shot Learners.'
- Co-author of the 2020 paper identifying empirical scaling laws for neural language models.
- Co-author of OpenAI's 2017 work on learning from human preferences and creator of its open-source RL-Teacher implementation.
- Co-founded Anthropic in 2021.
- Leads Anthropic's compute organization across a strategy that uses GPUs, Google TPUs and AWS Trainium accelerators.

## Latest updates

- **2026-09** — Appeared in a prominent fireside-chat role at the G20 Innovation Ministerial in Chapel Hill as Anthropic took a more visible role in US AI policy discussions.
- **2026-09** — Anthropic lists Brown as Co-Founder and Chief Compute Officer, leading the technical organization responsible for securing, scaling and effectively using compute resources.
- **2025-08** — Joined Y Combinator's Lightcone podcast to discuss his route from startups to GPT-3, Claude Code, scaling laws and Anthropic's compute strategy.
- **2025-04** — Discussed Anthropic's role in Amazon's Project Rainier and compared a frontier lab's partnership with a cloud provider to Netflix helping AWS mature at scale.

## Links

- Website: https://www.anthropic.com/company/leadership
- LinkedIn: https://www.linkedin.com/in/nottombrown
- Twitter/X: https://x.com/nottombrown
- GitHub: https://github.com/nottombrown
- YouTube: https://www.youtube.com/watch?v=JdT78t1Offo

---

Profile page: https://yespress.io/tom-brown
Published by YesPress — https://yespress.io
Last updated: 2026-09-26