San Francisco startup Mercor has raised $750 million in its latest funding round, bringing the company’s total value to $20 billion. This latest capital injection comes less than a year after Mercor was valued at $10 billion. The company plans to use the funds to train artificial intelligence systems across every job market.
Big Investors Back the AI Training Platform
The round did not name a lead investor, but the list of participants is extensive. Benchmark, Felicis, General Catalyst, Sapphire Ventures, NVIDIA, Salesforce, and Zoom all took part in the investment. Other backers include funds from Apollo, G Squared, StepStone Group, and XTX Ventures. Part of the capital came from an employee tender offer that allowed staff to sell shares.
Since October 2025, Mercor has doubled its valuation. The company announced the deal on November 10. Investors are betting on a market where AI models are trained on real-world data rather than just public benchmarks. Last week, OpenAI made headlines with its $11 billion raise at a $1.25 trillion valuation, a figure that highlights the massive capital flowing into the sector. This massive influx of cash shows the intense competition among artificial intelligence companies.
Connecting Experts with AI Labs
Founded in 2023 by Brendan Foody, Adarsh Hiremath, and Surya Midha, Mercor began as a recruitment platform. The founders observed that human experts are critical for creating and testing AI models. In just 13 months, Mercor has now doubled its valuation to $20 billion. As Foody puts it, “AI can only be as useful as the human brain behind it.” The company now claims to have a network of over 5 million experts across 800 occupations in the United States. It reports an annualized revenue run rate of more than $2 billion as of July 2026.
Standard benchmarks do not always predict how an AI will perform in a specific business environment. Mercor creates custom tests based on a company’s workflows to demonstrate actual performance. This data is then used to train the model further. Hiremath explains that this is important because “a model can have a high score on a public benchmark and fail.” The company recently acquired Deeptune to help rebuild hundreds of enterprise apps so users can practice with AI agents without risking real systems.