The archive · AI & Models · Strategic decision · 2023-2026
Mercor's expert-feedback bet: teen recruiting startup to $10B, $2B ARR
Three 21-year-olds bet AI labs would pay a premium for expert human feedback - pivoting an AI recruiting marketplace into a $2B-run-rate data network
Mercor
What the business is
Marketplace that recruits and vets domain experts (scientists, doctors, lawyers, engineers) and routes them to AI labs for model training and evaluation, charging an hourly finder's fee and matching rate.
Starting capital：$3.6M seed (2023, General Catalyst); $32M Series A at $250M (2024, Benchmark); $100M Series B at $2B (Feb 2025, Felicis); $350M Series C at $10B (Oct 2025, Felicis)
How it started
Founded in 2023 by Brendan Foody (CEO), Adarsh Hiremath (CTO) and Surya Midha (COO), three 21-year-old Thiel Fellows, as an AI-driven hiring platform: automated resume screening, 20-minute AI interviews, candidate matching and payroll, initially focused on connecting software engineers with US companies.
What happened
Pivoted after discovering AI labs would pay hourly finder's fees for domain experts to train models via RLHF and evals. Raised a $3.6M seed (2023), $32M Series A at $250M (2024), then $100M Series B at $2B in Feb 2025 when ARR reached $75M, most of it from AI labs. By Oct 2025 it paid over $1.5M/day to 30,000+ experts earning $85+/hour on average, with OpenAI and Google DeepMind among its biggest customers after they reportedly cut ties with Scale AI. Gross annualized revenue crossed $500M in September 2025, $1B around February 2026, and $2B by June 2026.
How it ended up
Still scaling fast: TechCrunch reported on July 8, 2026 that Mercor crossed $2B in gross annualized revenue as of June - four months after the $1B milestone - making it one of the fastest-growing AI companies by revenue velocity.
Background
Mercor was founded in 2023 by Brendan Foody, Adarsh Hiremath and Surya Midha - three 21-year-old Thiel Fellows - as an AI-driven hiring platform. Candidates took 20-minute AI interviews that built skill profiles; employers uploaded job descriptions and Mercor matched engineers to full-time, part-time or hourly roles. The bet behind the company, however, was bigger than recruiting: that human expertise would become the scarce input for frontier AI.
The founders discovered that AI labs would pay a premium for the same network of domain experts - scientists, doctors, lawyers, bankers - to train models through reinforcement learning from human feedback and evaluations. Mercor pivoted from matching hires to routing experts into lab workflows, charging hourly finder's fees. Seed ($3.6M, 2023) and Series A ($32M at $250M, 2024) led to a $100M Series B at $2B in February 2025, when ARR stood at $75M.
Growth then accelerated: $350M Series C at $10B in October 2025 (Felicis leading, with Benchmark, General Catalyst and Robinhood Ventures), $1.5M+ paid daily to 30,000+ experts earning $85+/hour on average, and OpenAI and Google DeepMind among the largest customers after they reportedly cut ties with data-labeling rival Scale AI. Gross annualized revenue crossed $500M in September 2025, $1B around February 2026, and $2B by June 2026.
Mercor's branded search tells the same story: roughly 165K monthly branded searches as of early 2026, adding about 36,000 per month, which made it the #4 fastest-growing AI company in Analyze AI's ranking. The case shows a marketplace finding a much more valuable buyer for the same asset than the one it was built for.
What has to be true
- Frontier labs' bottleneck shifted from compute to data quality, so expert-graded feedback became a must-pay input.
- The recruiting marketplace gave Mercor the vetting, payments and matching infrastructure that an expert-data business needed on day one.
- RLHF and evals require scarce, credentialed judgment - doctors, lawyers, PhDs - which commands high hourly rates and resists automation.
- A handful of foundation-model customers created enormous revenue concentration, which fueled both hypergrowth and dependency risk.
What can be applied
The same asset can be sold to very different buyers: the expert network built for hiring was worth far more as training data - the pivot worked because founders watched what customers paid for.
Aftermath
As of July 8, 2026, Mercor reported $2B in gross annualized revenue as of June 2026 - doubling in the four months after its $1B milestone - with most revenue coming from AI foundation-model labs, and it was expanding its talent network, matching systems and automation while planning to return to an AI-powered recruiting marketplace.
Sources
- Mercor quintuples valuation to $10B with $350M Series C
- Mercor, an AI recruiting startup founded by 21-year-olds, raises $100M at $2B valuation
- These AI startups are growing revenue at faster and faster rates
- The 50 Fastest-Growing AI Companies in 2026
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card