Mercor raises $750M Series D at $20B valuation with employee tender
Mercor, a San Francisco company that uses human experts to train and evaluate AI models, announced a $750 million Series D on November 10, 2026, at a $20 billion valuation. The company says the round included an employee tender offer, but it has not disclosed how much of the total went to that share sale. According to The Next Web, the new valuation is double the $10 billion mark set after a $350 million Series C in October 2025. Thirteen investors or investor groups took part, including Benchmark, Felicis, General Catalyst, NVIDIA, Salesforce and Zoom, and Mercor did not name a lead.
The capital is meant for evaluations, post-training and environments where agents learn multi-step workflows. Earlier in 2026, Mercor acquired Deeptune, a reinforcement-learning environment company that spent two years recreating hundreds of enterprise applications as training settings, according to Unite.ai. Mercor also opened offices in New York and London this year and says its network includes more than five million experts paid to train frontier models. The company runs APEX, a public benchmark series that measures AI on economically valuable work. Mercor has not disclosed revenue.
Co-CEO Adarsh Hiremath framed the pitch around a question every adopter must answer: whether AI works for its own business, with an answer that differs by company. Co-founder Brendan Foody argued that agents will need to be trained and tested on each organization's own work. Source for the announcement: Mercor's Series D blog post and the Business Wire release carried by Yahoo Finance.
Why it matters
For founders building agents, the signal is that buyers want evidence from their own workflows, not just public leaderboard scores. Mercor is pitching that each company needs custom evaluations, which makes task sets, verifiers and training environments a layer that teams can defend. Startups selling agents into enterprises should expect buyers to ask how a product performs on the buyer's own processes.
For investors, a $20 billion mark on a human-data supplier shows that large strategic checks are flowing into expert-driven training and evaluation. NVIDIA, Salesforce and Zoom are backers, and reports describe NVIDIA as a customer too, which raises vendor-neutrality questions for rivals. Operators should weigh whether a supplier's investors compete with them before sending it sensitive workflow data.
Key points
- 1Mercor raised $750 million in a Series D at a $20 billion valuation on November 10, 2026.
- 2The round included an employee tender, but Mercor has not disclosed how much of the $750 million it accounted for.
- 3Thirteen named investors or investor groups took part, and Mercor did not name a lead.
- 4Mercor says its network includes more than five million experts paid to train frontier AI models.
- 5Proceeds will fund evaluations, post-training and environments where agents learn multi-step workflows.
