Pomegra
Mercor raised $750M in a Series D at a $20B valuation, double its October 2025 mark, as Nvidia, Salesforce and Benchmark back its expert-driven AI training.
Key Takeaways
- Mercor raised $750M at a $20B valuation on November 10, 2026, up from $10B in October 2025.
- Benchmark, Nvidia, Salesforce, Zoom and General Catalyst joined the round. No lead investor was named.
- The round includes an employee tender, and the split between new and secondary capital is undisclosed.
Lead
Mercor, the San Francisco company that supplies human expertise to train and evaluate AI systems, announced a $750M Series D on November 10, 2026. The round values it at $20B. That is twice the $10B it commanded when it raised $350M just over a year earlier, in October 2025.
Participants include Benchmark, Felicis, Fundomo, General Catalyst, G Squared, Glade Brook Capital, Sapphire Ventures, StepStone Group, XTX Ventures and funds managed by affiliates of Apollo. Strategic money came from Nvidia, Salesforce and Zoom. The announcement named no lead investor.
What Does Mercor Actually Sell?
Mercor sells access to professionals who generate and grade the data that AI models learn from. Its network holds more than five million experts, and frontier AI labs, big tech companies, startups and Fortune 500 firms use them to evaluate and post-train models on real professional work. Post-training means further tuning an existing model after its initial training run.
Founded in 2023, the company is run by co-CEOs Brendan Foody and Adarsh Hiremath, with Surya Midha as a third co-founder. It says the new money will go toward infrastructure and research for model evaluations and post-training, plus more simulated environments where AI agents can practice complex workflows.
Those environments are a growing part of the pitch. Mercor acquired Deeptune earlier in 2026, a company that recreated more than 100 enterprise applications for agent training. "Every job in the economy will need its own AI agents, and every company will need a way to train and evaluate them," Foody said.
Why Did the Valuation Double in a Year?
The valuation doubled because labs and enterprises are spending heavily on high-quality human data as models move from answering questions to performing jobs. Public benchmarks no longer separate good models from useful ones. Hiremath put it this way: a model can score well on a public benchmark and still fail on the systems, policies and edge cases that define how a specific business runs.
The doubling also reflects a crowded and richly priced category. Scale AI is valued at roughly $14B, and Surge AI has been reported at more than $25B. Snorkel AI sits near $3.5B and micro1 near $500M. Against that field, $20B puts Mercor in the upper tier but not alone at the top.
Skepticism is warranted on the structure. A round that mixes primary capital with an employee tender and carries no named lead is harder to read than a clean growth raise. Mercor did not disclose revenue in the announcement, nor how much of the $750M is fresh money versus secondary sales.
What Does the Investor List Signal?
The list signals that the buyers and suppliers of AI infrastructure want a stake in the data layer. Nvidia sells the chips that run the models Mercor's experts help refine. Salesforce and Zoom sell software into which enterprise agents are being inserted. Each has a direct interest in agents that work reliably on company-specific workflows.
Benchmark, the earliest institutional backer in the lineup, is a repeat participant. Apollo and StepStone point to growing interest from credit and secondaries investors, who typically value cash flow and tender liquidity over narrative.
Risks and Open Questions
The year was not without friction. A security breach in April 2026 led Meta to freeze AI data work with Mercor, and the company published an incident update in June. Customer concentration among a handful of frontier labs remains a structural exposure, since a lab can shift data work in-house or to a rival with little notice.
Cost structure is another question. Foody has said Mercor spends about three times more on model inference than on employee salaries, a sign that its own operations lean heavily on AI. Contractors also keep a large share of billings, which caps gross margin compared with software.
Outlook
Mercor enters 2027 with fresh capital, a $20B mark and a pitch built around agents that need training on specific jobs. The test is whether demand from AI labs holds as synthetic data and automated evaluation improve, and whether the company can broaden its base beyond a few large buyers. Disclosure of revenue and the primary-versus-secondary split of this round would clarify how much of the valuation rests on proven earnings.
