Three 22‑Year‑Old Dropouts Disrupt AI Recruiting, Landing $10 B Valuation in Two Years

Founding and Early Growth

In January 2023 three friends—Brendan Foody, Adarsh Hiremath and Surya Midha—left Harvard, Georgetown and Georgetown respectively and co‑founded Mercor in a 50 m² San Francisco office. Within 1,000 days the startup grew from zero revenue to roughly $500 million in annualized revenue, signing contracts with the world’s top AI labs (OpenAI, Google, Meta, Microsoft, NVIDIA) and overtaking competitor Scale AI during its crisis.

Product and Market Fit

Mercor’s AI recruiting tool lets a client describe a role in natural language (e.g., “full‑time Python engineer with computer‑vision experience”). The system performs semantic search across hundreds of thousands of resumes, GitHub projects and social‑media profiles in seconds, then presents an AI‑generated interview video of the candidate, compressing a weeks‑long hiring cycle into under 24 hours. This speed advantage attracted the first wave of customers.

Funding Milestones

Seed funding of $3.6 million (led by General Catalyst) in 2023 enabled team expansion and model improvements. By January 2024 the company’s annual recurring revenue topped $1 million and it had built a talent pool of 100 000 experts in 25 countries. A $100 million Series B round in early 2025 raised the post‑money valuation to $20 billion, and an eight‑month‑later Series C round of $350 million (led by Felicis Ventures with Benchmark and General Catalyst) pushed the valuation to $10 billion.

From Recruiting to Computation‑Scale Labor Network

In February 2024 Mercor realized its talent network could also provide data‑labeling and model‑evaluation services for large‑model developers. The company hired former Uber CPO Sundeep Jain as its first president and expanded its talent pool to include lawyers, doctors and journalists, reaching 30 000 specialists.

APEX Benchmark

To address the gap between academic AI benchmarks and real‑world economic impact, Mercor released the AI Productivity Index (APEX) in October 2024. A consortium of industry and academic experts—including Larry Summers, Dominic Barton, Cass Sunstein and Eric Topol—designed tasks representing four professions (investment‑banking analyst, large‑law firm associate, strategy‑consulting analyst, general‑practice physician). APEX v1.0 evaluated 21 state‑of‑the‑art models; GPT‑5 achieved the highest overall score of 64.2 %, while the best open‑source model (Qwen‑3) scored 59.8 % and ranked seventh. Scores varied by domain, with law at 70.5 % and investment banking at 59.7 %.

All models required substantial human supervision; none could complete the tasks unsupervised. The authors estimate that fully autonomous models could unlock thousands of billions of dollars for the U.S. economy, provided reliability and safety are ensured.

Work Culture and Organizational Challenges

Mercor adopted a “six‑day work week” culture, initially operating seven days a week. Founders argue that the intense pace is necessary to compete with AI giants such as OpenAI and Anthropic. Employees within a half‑mile of the office receive a $10 000 housing stipend. In October 2025 the company faced a lawsuit from Scale AI alleging theft of trade secrets, and later that month COO Surya Midha stepped down to become board chair, prompting investor scrutiny.

Outlook

Despite legal and governance hurdles, Mercor’s rapid ascent—from a dorm‑room startup to a $10 billion AI infrastructure player—demonstrates the potential of combining human expertise with AI‑driven talent matching and evaluation. The company continues to expand its global talent network, refine the APEX benchmark, and position itself as a critical layer of the emerging AI economy.