AI Training-Data Startup Mercor Raises $750 Million at $20 Billion Valuation

Mercor, a San Francisco startup, announced on November 10, 2026 that it had raised $750 million in a Series D funding round, valuing the company at $20 billion.

Mercor does not build AI models itself. Instead, it recruits vetted professionals in fields such as law, medicine, finance and software engineering, then pays them to produce and review material that AI labs use to fine-tune and evaluate their models, including work such as reinforcement learning from human feedback and model grading against expert standards.

The company has not publicly named all of its clients, but multiple reports describe its customer base as including some of the largest AI developers, which rely on Mercor to source the kind of specialist, professional-grade knowledge that general web data cannot provide.

Investors in the new round include chipmaker Nvidia and venture firm Benchmark, according to multiple reports, alongside other financial and corporate backers. The $20 billion valuation is roughly double what Mercor reached in its previous funding round, though reports differ on exactly how much time passed between the two raises.

Part of the $750 million also funded a tender offer, letting some employees and early investors sell existing shares, rather than all of the money going directly into the company as new capital. Mercor has not disclosed how the total splits between the tender offer and primary funding.

The round adds to a string of large investments this year in startups that supply the human-generated data increasingly seen as a bottleneck in training advanced AI systems, as demand grows for specialist expertise that cannot be scraped from the open web.