Sourcery with Molly O'Shea · Nov 10, 2026 · 54m · 44m saved
With Brendan Foody, Co-Founder and CEO of Mercor
Mercor has just raised $750 million in a Series D at a $20 billion valuation, with Nvidia, Benchmark, General Catalyst, Apollo-affiliated funds, XTX Ventures and Salesforce taking part. Brendan Foody, co-founder and CEO of Mercor, told Molly O'Shea on Sourcery with Molly O'Shea that the company has grown more than 5x in about a year, on revenue he puts in the mid-nine figures.
Much of Silicon Valley treats paying human experts to train AI models as a temporary business that better models will make unnecessary. Brendan's case is the reverse: cheaper, more productive experts have so far meant more demand for their work, not less.
"Data is still contrarian... a lot of the top firms in Silicon Valley just thought that the market would dissipate within a couple of years." — Brendan Foody
Brendan built Mercor's first AI interviewer with his co-founder while both were still in college, and the company now pays out more than $5 million a day to the experts on its platform.
The full interview is covered here so you can skip it. 54 minutes of audio, 10 minutes of reading.
Here are the 11 arguments that matter.
Key Takeaways
Making experts more productive has raised demand for human training data, not cut it
Brendan cites Jevons paradox: AI co-pilots make experts roughly 3-5x more productive per hour
Mercor's product is the reinforcement-learning environment, not the marketplace of experts
Mercor went from paying out about $1M a year to more than $5M a day in 24 months
Every Fortune 500 company will come to resemble a "Neolab" that builds its own training data
Letting a frontier lab train on that data would be like handing it to a competitor
Enterprises spending $10-100M on inference start putting about 10% of it toward evals
Brendan has turned down offers to sell Mercor, and an IPO is not coming in the next 12 months
1. $750M at a $20B Valuation
The Series D closes a year in which Mercor grew more than 5x since Brendan and his co-founder Adarsh Hiremath last spoke with Molly. Brendan's case for Mercor becoming a trillion-dollar company starts from a bet about how work changes.
Brendan Foody expects people to be paid to train agents rather than do repetitive work
Most human labor today is repetitive: answering support tickets, redlining contracts, editing podcasts. Brendan expects people instead to train an agent once to automate a task, and to be paid more and more for writing the evaluation and skills files that teach it.
"We've now created the fastest-growing job category in history." — Brendan Foody
He calls that change in how people contribute to the economy the reason the category can be "legendary" if Mercor executes well.
