Tom Brown

Overview

Tom Brown is a co-founder of Anthropic and the company’s Chief Compute Officer — the co-founder whose day job is the physical foundation of the AI boom: compute (Reuters). Forbes lists his citizenship as United States and residence as San Francisco, California, while Bloomberg lists his country/region as United States; neither source establishes a birthplace (Forbes; Bloomberg). While most coverage of him leads with the biographical hook — a B-minus in linear algebra not long before co-founding one of the world’s leading AI labs — the distinctive verified fact is what followed: Brown was the first author of the 2020 paper “Language Models are Few-Shot Learners” that introduced GPT-3, a 175-billion-parameter model ten times larger than any previous non-sparse language model (arXiv; The Times of India). The man who wrote the paper is now the executive who buys the hardware: in September 2026 he told U.S. Commerce Secretary Howard Lutnick that AI demand was driving “this incredible industrial build-out, the likes of which we’ve never seen in the history of humanity, even more than the railroads of the 1800s” (The Singju Post).

The 2014 conversation that started it

In Brown’s own telling, the pivot came in 2014, when, as he put it at the G20 Innovation Summit, “a few of my friends came to me and they were like, ‘Tom, it seems like we might end up getting human-level intelligence from AI within our lifetimes, so in the next 50 years’” (The Singju Post). The proposition sounded far-fetched, but its stakes were impossible to ignore: “if it does happen, that would be the biggest, most important thing that ever happens to humanity.” He framed it with a comparison he still uses — the Industrial Revolution transformed society, but “the second most important thing about humans is our bodies. The main thing is our minds” (The Singju Post).

So he left conventional engineering and began teaching himself AI. About a year later OpenAI was founded, and he went “to see if I could help out in some way” — joining, in his words, “as one of the first 20 employees basically to do engineering work” (The Singju Post). Before AI, he had experimented with startups, including Grouper, a group-dating app, where he became close with Greg Brockman, later OpenAI’s co-founder and president: “I ended up being close with Greg (Brockman, co-founder and president of OpenAI), which ended up being my connection at OpenAI,” he said on Y Combinator’s Lightcone Podcast (The Times of India). That connection became his route into OpenAI through his friend Greg Brockman.

His self-study was deliberate and concentrated: roughly six months working through Sheldon Axler’s Linear Algebra Done Right, Google DeepMind’s How to Scale Your Model e-book, Coursera courses, Kaggle projects, and a GPU bought with Y Combinator alum credit, according to reporting drawing on Business Insider (The Times of India).

First author of the GPT-3 paper

Brown is listed as the first author of “Language Models are Few-Shot Learners,” submitted in May 2020, the paper that introduced GPT-3 (arXiv). The author list runs to 31 names — including Jack Clark, who would go on to co-found Anthropic alongside Brown (arXiv; Chris Olah). The abstract describes GPT-3 as “an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model,” trained without task-specific fine-tuning and evaluated in the few-shot setting (arXiv).

Brown identifies the 2019 scaling-laws studies as the turning point of his career. Researchers including future Anthropic co-founders had done the kind of science you would do, he said, “if you were in the early days of a combustion engine” — finding the exact recipe of compute, data, and model size that produced a given level of intelligence (The Singju Post). Those studies gave the field confidence that the trend — models getting bigger, more compute, better results — held across orders of magnitude. For Brown, that collapsed the timeline from fifty years to under ten: “that ended up leading to the founding of Anthropic” (The Singju Post). There he joined fellow co-founders including Chris Olah, who leads the company’s interpretability research (Reuters; Chris Olah).

The compute mandate

As Chief Compute Officer, Brown is Anthropic’s public voice on the infrastructure race. In June 2026, when Anthropic signed a supply-and-investment agreement with Micron Technology, he said: “Our compute strategy depends on getting every layer of the stack right, and memory and storage are central to how efficiently we can train and serve Claude” (Reuters). The company had in recent months signed several major agreements to secure computing capacity, including deals with CoreWeave, Broadcom and SpaceX (Reuters).

In September 2026, in a fireside conversation with U.S. Commerce Secretary Howard Lutnick at the G20 Innovation Ministers’ meeting in Chapel Hill on September 2, 2026, he laid out three takeaways for the assembled ministers: AI capabilities have been increasing very rapidly and will continue to; demand is growing exponentially, “increasing in multiples every year”; and as a result, “we have this incredible industrial build-out, the likes of which we’ve never seen in the history of humanity, even more than the railroads of the 1800s” — an expansion he argued any country could still participate in (The Singju Post).

What this profile does not claim

This profile does not assert Brown’s birth date, birthplace, or degree — no source read for this profile confirmed them; the only education detail on record is the B-minus in linear algebra. It does not repeat the widely reported claim that Anthropic’s March 2025 funding round made its seven co-founders billionaires, because that claim was not re-verified from a readable source in this pass. It states that Brown was the first author of the GPT-3 paper, which is verified from the paper’s own author list; it does not assert the stronger claim that he “led the engineering of GPT-3,” which appears only in non-citable third-party commentary. The 2021 founding year of Anthropic is carried in the timeline for consistency with the site’s published Anthropic profile, but was not independently stated in a source read for this pass — see openItems. Finally, this is the Anthropic co-founder and AI researcher, not the British chemist Tom Brown who shares the name.