Mercor lands $750M to scale AI agent evaluation beyond research labs into enterprise

Mercor.io Corp., a provider of human expertise for training and evaluating artificial intelligence models, announced a $750 million funding round that values the company at $20 billion. The fresh capital marks a shift in strategy: moving beyond the frontier AI labs that have been Mercor's primary customers and into the broader enterprise market, where companies increasingly need to deploy and validate AI agents for their own operations.

The company operates a network exceeding 5 million domain experts who contribute to the training and evaluation of AI systems. Until now, most of this work has served leading AI research organizations. The new funding will support Mercor's push to help enterprises across industries develop their own methods for determining whether an AI agent can reliably handle critical business tasks.

Co-Chief Executive Brendan Foody stated that "every job in the economy" will eventually require dedicated AI agents, and organizations will need the ability to train and evaluate these systems using their own operational data. Mercor has spent the past two years establishing the foundational infrastructure for this work alongside frontier labs. The lab business has expanded rapidly; Foody noted in a July post on X that annualized revenue surpassed $2 billion in June.

Co-Chief Executive Adarsh Hiremath emphasized that a model performing well on public benchmarks may still struggle with "the systems, policies and edge cases that define how a specific business runs." To address this gap, Mercor develops custom benchmarks for each enterprise customer, measuring model performance against that organization's specific requirements. Post-training efforts then proceed from these tailored evaluations.

The company's infrastructure includes software acquired from Deeptune Inc., a New York startup that Mercor purchased in July. Deeptune's team had spent the preceding two years building simulated versions of hundreds of enterprise applications to serve as practice environments for AI agents. The new funding will support expanded research capabilities and evaluation infrastructure, including additional enterprise application environments.

Mercor did not identify a lead investor for the Series D round. Corporate investors participating in the funding include Nvidia Corp., Salesforce Inc., and Zoom Communications Inc. Returning backers Benchmark Capital Management Co., Felicis Ventures Management Co., and General Catalyst Group Management also joined the round. Additional investors included Fundomo, G Squared Equity Management LP, Glade Brook Capital Partners, Sapphire Ventures, StepStone Group Inc., XTX Ventures, and funds affiliated with Apollo Global Management Inc.

Mercor's previous funding round occurred in October 2025, when Felicis led a $350 million Series C that placed the company's valuation at $10 billion. Foody indicated that the current round included a tender offer allowing employees to sell existing shares, though the company has not disclosed how the $750 million divides between secondary transactions and new capital.

Jai Das, president, partner and co-founder of Sapphire Ventures, noted that frontier labs have dominated demand for expert-driven training data to date. He anticipates this dynamic will shift as open-source models become sufficiently capable for enterprises to build upon them. Das characterized Mercor's position as having created "something hard to replicate" through several years of collaborative development with leading research organizations.