Brendan Foody

When the principal moved to shut down his donut stand, Foody moved the stand twenty feet past the property line, outside school jurisdiction, and kept selling. He was in eighth grade, buying wholesale from Safeway, reselling at school by bike, and undercutting a rival stand on price until it went out of business.¹ He is the co-founder and CEO of Mercor, a company that uses artificial intelligence to evaluate job candidates and, in its largest line of business, finds and vets human experts for the AI labs that need them to train frontier models. Founded in 2023, Mercor grew from roughly $1 million to more than $1 billion in annualized revenue in under two years, a pace described as the fastest revenue growth of any startup on record. By his early twenties, Foody and his two co-founders were among the youngest self-made billionaires in the world.

Background

Foody grew up in the Bay Area. His mother, watching the donut hustle escalate, worried it would tip into something worse and enrolled him in a Catholic high school, Bellarmine College Preparatory in San Jose. There, on the debate team around age fourteen, he met Adarsh Hiremath and Surya Midha, his future co-founders. Foody is dyslexic, which he has said made him weaker at debate than the other two. In high school he also ran a consulting business filing for unclaimed AWS startup credits on behalf of sneaker resellers, earning hundreds of thousands of dollars before finishing school.¹ He did not want to go to college but enrolled at Georgetown under parental pressure. Hiremath went to Harvard, and Midha also went to Georgetown, rooming with Foody.

Getting started

Mercor began at a hackathon in São Paulo in early 2023, when Foody and his co-founders built a tool matching freelance developers, largely in India, with US companies, assessing them with language models. All three dropped out of college during sophomore year to pursue it full time, raising a seed round of just over three million dollars led by General Catalyst. The founding insight sharpened when Scale AI, then the dominant AI training-data company, came to Mercor's platform to hire thousands of workers itself. Mercor was flooded with support tickets from unpaid workers accusing it of running a scam, since Mercor had referred them there; investigating, Foody concluded that Scale's infrastructure, built for labeling autonomous-vehicle footage cheaply in the Philippines, was not built for vetting elite experts. That pushed Mercor to go directly to the AI labs, building its own vetting infrastructure for a market moving from low-skill crowdsourcing to high-skill vetting: finding Goldman Sachs analysts, McKinsey consultants, FANG engineers, doctors, and lawyers to work alongside AI researchers.¹

What he built

Mercor's core product predicts how well a candidate will perform in a job, automating resume review, interviewing, and the hire decision. Its largest business supplies the human experts AI labs need to build "RL environments," the task simulations that have become a primary data source for training frontier models. Mercor says it holds roughly half to three-fifths of that market, paying workers an average of about $95 an hour, well above the roughly $30 an hour associated with earlier crowdsourced platforms like Scale and Surge.¹ It raised a Series A led by Benchmark at a $250 million valuation on about $1.5 to $2 million in run-rate revenue, then a Series B four months later at a $2 billion valuation on roughly $20 million in run rate. It crossed roughly $500 million in annualized revenue about seventeen months after founding, then $1 billion within twenty months, before doubling to around $2 billion by mid-2026.¹ By October 2025 investors valued Mercor at $10 billion, and by mid-2026 it was reportedly in talks to raise at $20 billion. Mercor also built APEX, a benchmark top AI labs use to measure model performance on economically valuable professional work, and has begun hiring scientists to run physical experiments that models design, using the results as training data.

How he operates

Foody describes his own drive as obsession rather than discipline: "I've never been super disciplined," he has said, offering instead an involuntary pull toward problems he cannot stop thinking about, even away from work.² His filter for what to build is whether a problem keeps returning to mind unbidden, where importance is high and attention from others is thin. The same instinct, he has noted, ran through the donut stand, the AWS credit business, and Mercor: a structural inefficiency spotted and the arbitrage automated at increasing scale. The same shape shows up identically across ventures built years before he had language for it. The obsession half of that description is his own, in his own words; the systematizing half is the archive's observation rather than anyone's account of him, which is why no plate is entered against his name. Foody holds an uncompromising bar on Mercor's first ten hires, believing they shape the next hundred, and treats fundraising as an input rather than a milestone; both Series A and Series B were raised without a slide deck.² He calls the business capacity-constrained, turning down projects it could otherwise staff, and prioritizes long-term measures like marketplace depth over short-term revenue. He credits treating workers well, reflected in Mercor's above-market wages, as the most important advantage a marketplace business can have.¹

Where things stand

As of mid-2026, Foody remains CEO of Mercor, running it privately alongside Hiremath and Midha, intending to stay private as long as possible on the advice of investor Jack Dorsey.¹ The company was reportedly in talks to raise fresh capital at roughly double its prior valuation, serving most leading AI labs alongside a growing base of enterprise customers. Foody says publicly that he sees the bottleneck on applying AI across the economy as the human labor needed to build evaluation frameworks for every task, and that he expects this need to grow rather than shrink as AI capability advances.²

Key facts

• Grew Mercor from about $1 million to more than $1 billion in annualized revenue in under two years, the fastest such ramp of any startup on record.
• Co-founded Mercor in 2023 at 19 with two Bellarmine College Preparatory debate-team friends, Adarsh Hiremath and Surya Midha, all of whom dropped out of college to build it.
• Became one of the youngest self-made billionaires in the world in October 2025, when Mercor was valued at $10 billion; by mid-2026 it was reportedly in talks to raise at roughly $20 billion.
• Ran a donut resale business in eighth grade and an AWS-credit consulting agency in high school, each earning hundreds of thousands of dollars before college.
• Is dyslexic, a detail he has disclosed publicly in discussing his non-traditional path.
• Built APEX, a benchmark for economically valuable professional AI capability now used by multiple leading AI labs.