Mercor operates a network of more than 5 million domain experts who provide professional knowledge to train, test, and improve AI models. While the company has previously focused on providing expert-driven training data to major model developers, the new strategic focus is on helping enterprises measure whether AI agents can be trusted with specific jobs. Co-Chief Executive Brendan Foody stated that every job in the economy will eventually require its own AI agents, and companies will need dedicated infrastructure to train and evaluate these systems on their own work.
The practical effect of Mercor’s platform is the development of custom evaluation benchmarks for each enterprise client. Co-Chief Executive Adarsh Hiremath explained that a model may score well on public benchmarks but still fail when applied to the specific systems, policies, and edge cases that define a particular business. By building these customized tests, Mercor provides a foundation for post-training, allowing businesses to refine AI systems using examples and feedback drawn from their own professional workflows.
To support this enterprise expansion, Mercor is investing in research, evaluation systems, and simulated enterprise environments. These environments allow AI agents to practice complex professional tasks within recreated workplace conditions. Part of this post-training infrastructure relies on software from Deeptune Inc., a New York-based startup that Mercor acquired in July 2026. The Deeptune team had previously spent two years recreating hundreds of enterprise applications specifically for AI agent practice.
The $750 million round includes participation from new corporate investors Nvidia Corp., Salesforce Inc., and Zoom Communications Inc. Existing investors such as Benchmark, Felicis, and General Catalyst also participated, alongside funds managed by affiliates of Apollo Global Management. The company did not specify how the total funding amount is split between fresh capital and a tender offer for employee shares. Foody noted that Mercor’s annualized revenue passed $2 billion in June, driven largely by its work with frontier labs.
Sapphire Ventures partner Jai Das observed that while frontier labs have been the primary buyers of expert training data, this dynamic is expected to shift as open models become viable for enterprise applications. The valuation reflects the broader market appetite for AI infrastructure, with AI-driven startups raising $102 billion in the third quarter of 2026, accounting for 64% of global venture capital. Mercor’s next major milestone is the continued deployment of its custom evaluation infrastructure to Fortune 500 enterprises and other organizations seeking to integrate specialized AI agents into their operations.
