Mercor has raised $750 million in a Series D that values the AI training company at $20 billion, doubling its valuation from last year's funding round. The money will fund a push beyond frontier AI labs and into companies building agents for everyday business tasks.
Nvidia, Salesforce and Zoom joined an investor list that also includes Benchmark, Felicis, General Catalyst, Sapphire Ventures, XTX Ventures and funds managed by Apollo affiliates. The financing also included an employee tender, according to Mercor.
Major backers join the round
The presence of Nvidia and Salesforce brings two major parts of the AI stack around Mercor: computing infrastructure and enterprise software. But Mercor disclosed no individual investment amounts or lead investor, so the size of each backer's commitment remains unclear.
From frontier labs to enterprises
Mercor built its business connecting expert workers with labs that need specialist knowledge to train and evaluate models. It says its network now exceeds five million experts, and its capabilities are being adopted by big technology companies, open-model developers and Fortune 500 customers.
A model may perform well on a public benchmark yet struggle with an insurer's claims process or a bank's internal policy. Mercor's approach tests agents against a company's actual workflows, systems and quality standards, then uses the results to guide additional training.
Training agents for real work
The company plans to invest in post-training research and realistic practice environments. Its acquisition of Deeptune earlier this year brought technology for recreating hundreds of enterprise applications, letting agents practice professional workflows before deployment.
Mercor has also opened offices in New York and London, expanding beyond San Francisco. Its bet is that as companies deploy more AI agents, they will pay for the expert knowledge and testing needed to prove those agents can perform real work reliably.
Get stories like this in your inbox every week. Subscribe to The Week in Perspective.

