Mercor Raises $750M Series D for Enterprise AI Evals

Recreating enterprise software gives Mercor somewhere to put an AI agent before a customer hands it a live assignment. The company is financing more of that work with a $750M Series D announced November 10, 2026, at a $20B valuation, as its business expands from supplying human expertise to AI labs into helping enterprises train and evaluate their own systems.

The financing included an employee tender, according to Mercor's announcement. The company did not disclose how the announced amount divides between primary capital and employee liquidity, leaving the amount available for new operating investment unspecified.

The commercial opportunity runs through a demanding customer relationship. Enterprises need agents that can work within their own policies, software and standards, and Mercor is assembling the expertise, evaluations and practice environments to help them get there. That gives the people who understand a business a consequential role in deciding what its software should learn.

What Mercor's Series D finances

The disclosed participants include Benchmark, Felicis, Fundomo, funds managed by affiliates of Apollo, G Squared, General Catalyst, Glade Brook Capital, NVIDIA, Salesforce, Sapphire Ventures, StepStone Group, XTX Ventures and Zoom. Mercor names no lead for the Series D. Reuters' November 10 report independently corroborates the amount, valuation and focus on professional AI tasks.

Mercor says the investment will support evaluation and post-training infrastructure, research, and additional environments in which agents can learn complex workflows. Those activities connect the work of deciding whether a system succeeds with the work of improving it. For a buyer, the useful result would be a clearer route from a disappointing agent response to a specific change in training or implementation.

The previous October 2025 Series C raised $350M at a $10B valuation, led by Felicis with Benchmark, General Catalyst and Robinhood Ventures participating. The new valuation is twice that earlier figure. The latest tender component also makes a simple sum of announced rounds an unreliable substitute for a reconciled total of primary capital raised.

From finding workers to understanding their work

Founded in 2023, Mercor grew out of a hiring business. Its founding team comprises Brendan Foody, Adarsh Hiremath and Surya Midha; the current company mission page identifies Foody and Hiremath as co-CEOs. San Francisco remains headquarters, with offices in New York and London.

Felicis' July 2026 profile describes how recruiting engineers exposed the founders to demand for expert-generated AI training data. Their infrastructure for finding and assessing people became a way to supply knowledge that model developers needed. Hiremath's work spans engineering, product and the enterprise business, according to that investor interview.

An enterprise brings another layer of specificity. A professional may understand the field but still need to learn how a particular employer handles exceptions, approvals and quality. Turning those practices into useful tests asks something of the customer as well as the supplier: people have to explain decisions they may previously have carried out without writing every rule down.

That is an operating commitment with a budget attached. The employees best equipped to describe a difficult workflow often have plenty of other work waiting for them. Mercor's proposition depends in part on making their participation useful enough that the organization keeps supplying that knowledge as the system changes.

Why Deeptune belongs in the funding story

Mercor announced its Deeptune acquisition in July and describes it as completed in the Series D statement. Deeptune built simulated enterprise applications for agents to practice professional tasks. Bringing those environments into Mercor connects the place where an agent works with the task it receives and the criteria used to evaluate the result.

The distinction matters because a polished answer and a completed workflow place different demands on software. Inside a workflow, an agent may have to navigate an application, use the available information, respect a process and leave a result another person can review. A useful practice environment gives developers a way to examine those steps before expanding the assignment.

This is also where the scope of a supplier relationship can grow. A customer that begins by purchasing an evaluation may need help interpreting the failures, changing the system and measuring it again. Mercor can pursue that wider relationship as it builds more of the surrounding infrastructure. Whether customers buy that combination will depend on the quality and relevance of the work delivered.

Enterprise adoption gives the experts another customer

Mercor's enterprise offering combines business expertise, forward-deployed engineers and a platform for context, integrations and evaluation. Its published process includes identifying potential workflows, deploying agents and continuously assessing performance. This places Mercor in conversations about implementation as well as model capability.

For enterprise teams choosing a supplier, that breadth introduces practical purchasing questions. They need to understand who defines success, how evaluations represent the work, how results can be inspected and how improvements enter production. These are analytical implications of the offering, rather than claims about outcomes at an unnamed customer.

Competition for that relationship can come from several directions: the organization's own engineering and subject-matter teams, implementation partners, and providers of evaluation or training infrastructure. Mercor's integrated approach makes the boundaries between those jobs commercially significant. The buyer still needs a clear account of which responsibilities stay with its own people.

The Series D supports a company trying to carry expert knowledge further into the customer's operation. As the practice environments become more detailed, somebody inside the enterprise will still have to decide which exceptions matter, which approvals survive automation and which result is good enough to hand to the next person. Those decisions give Mercor's growing infrastructure the work it has to represent.