Mercor announced a $750 million Series D at a $20 billion valuation on 10 November. The company supplies human experts who train and grade AI models, and it says the money will carry that work from the frontier labs to businesses in every industry. The round included a tender offer for employee shares, and the split between that and new capital was not given.
“the systems, policies and edge cases that define how a specific business runs”
Adarsh Hiremath, Mercor co-CEO, to SiliconANGLE · siliconangle.com ↗
No lead investor was named. Nvidia, Salesforce and Zoom put in corporate money, returning investors Benchmark, Felicis and General Catalyst took part, and others include Sapphire Ventures, StepStone, G Squared, XTX Ventures, Glade Brook, Fundomo and Apollo-affiliated funds. The valuation doubles the $10 billion set a year earlier, when a $350 million Series C was led by Felicis.
Mercor draws on a network of more than five million paid experts, and a July post from co-founder Brendan Foody said annualised revenue passed $2 billion in June. It bought Deeptune, a New York startup that spent two years rebuilding hundreds of enterprise apps for agents to practise on, in July, and it has opened offices in New York and London this year, beyond its San Francisco base.
Header graphic of Mercor's own Series D announcement post, 'Organizing human intelligence for the entire economy'
Courtesy of Mercor · Image: Mercor
Foody's argument is that every job will need its own agents, and that firms need a way to train and test them on their own work. For each enterprise customer Mercor builds a custom benchmark from that company's workflows and then post-trains a model on it. Co-founder Adarsh Hiremath says a model can ace a public benchmark and still fail inside a company's own systems. Mercor also says it spends three times as much on inference as on salaries, and that more than 800 US occupations hold workflows no model has been evaluated on.
About Mercor
Mercor is an American AI data startup that provides experts to train AI models and chatbots. It was founded in 2023 by Brendan Foody, Adarsh Hiremath and Surya Midha and is headquartered in San Francisco.
The three Bay Area high-school friends dropped out of college and won Thiel Fellowships; the company began by connecting freelance programmers in India with US firms, using an AI interviewing platform. After its $350 million Series C at a $10 billion valuation in October 2025, its founders, then 22, were described as the youngest self-made billionaires. In late March 2026 a supply-chain attack on the LiteLLM tool reportedly exposed up to 4TB of internal data and contractor details, which led to class actions and a pause in work from clients including Meta.
Sapphire's Jai Das expects the labs to stop being the main buyers as open models improve and enterprises build on them, and Mercor says enterprises can post-train capable open models with a fraction of frontier resources. The new money goes to evaluation and post-training research and to more practice environments for agents.
Sources · 2