Mercor Raises $750 Million at $20 Billion to Sell Frontier-Lab Training Work to Enterprises Building Their Own Models
AI training-data company Mercor said on November 10 it raised a $750 million Series D at a $20 billion valuation, from investors including Benchmark, General Catalyst, Apollo-managed funds, Nvidia and Salesforce. The round included an employee tender of undisclosed size. CEO Brendan Foody says the money will expand evaluation and post-training work for enterprises that want to fine-tune open models on their own workflows, a bet that companies will train their own models rather than only rent frontier ones.
Mercor made its name selling human expertise to the frontier labs. Its new round is a bet that the next buyer is everyone else: enterprises that start from an open model and train it on their own work, and need the same graders and practice environments the labs use.
The round#
Mercor said on November 10 that it raised a $750 million Series D valuing the company at $20 billion. Co-founder and CEO Brendan Foody wrote that the round included an employee tender, and investors including Benchmark, Felicis, Fundomo, funds managed by affiliates of Apollo, G Squared, General Catalyst, Glade Brook Capital, Nvidia, Salesforce, Sapphire Ventures, StepStone Group, XTX Ventures and Zoom. Mercor did not say how much of the $750 million went to employees selling shares rather than to the company.
The valuation matches the figure Bloomberg reported Mercor was discussing in July, when TechCrunch noted it would double the $10 billion of the company's $350 million Series C the previous October. Foody said on X at the time that annualised revenue had passed $2 billion; Mercor gave no new revenue figure with the round.
What the money is for#
Foody's pitch is that enterprises will increasingly "own their intelligence": take a capable open model and post-train it for their own tasks with a fraction of frontier-lab resources. Mercor says it builds custom evaluations from a company's actual workflows, and practice environments where agents can learn them. It bought Deeptune, which recreates enterprise applications for agents to train in, earlier this year. Smaller companies are building similar simulated workplaces for agent training.
Mercor says more than five million experts in its network are paid to train frontier models on their knowledge, and that it has opened offices in New York and London. The funding will go into evaluation and post-training infrastructure and research, and into more agent environments, Foody wrote.
Foody also says Mercor already spends three times as much on inference as on employee salaries, and that if that ratio spreads across the economy it "points to something closer to double-digit GDP growth over the next five years." That is his projection, not a forecast Mercor supports with data in the post.
Limits#
- The round size, valuation, investor list and network figures are Mercor's own. The other coverage we found repeats Mercor's release; we did not find independent reporting on the round's terms.
- The split between new money and the employee tender, the lead investor and any revenue figures were not disclosed.
- CNBC aired an interview with Foody about the round; we did not watch it and do not quote it.