Mercor Raises $750M at $20B to Scale Enterprise AI Training

By Grace Kim

Mercor raised $750 million in a Series D on November 10, 2026, valuing the AI training company at $20 billion. The company said the financing included an employee tender and participation from NVIDIA, Salesforce, Zoom and other investors.

Benchmark, Felicis, Fundomo, funds managed by affiliates of Apollo, G Squared, General Catalyst, Glade Brook Capital, Sapphire Ventures, StepStone Group and XTX Ventures also participated, according to Mercor’s announcement.

What will the $750 million fund?

Mercor said the capital will accelerate infrastructure and research for model evaluations and post-training, which it defines as additional training of an existing model. The money will also support more environments where agents can learn complex workflows.

Mercor opened New York and London offices in 2026, extending its footprint beyond San Francisco. The company said the expansion brings it closer to enterprise and laboratory customers and helps it recruit employees.

In plain terms, the round expands both Mercor’s training capacity and its operating footprint.

How does Mercor turn expert work into training data?

Over the past two years, Mercor has built infrastructure to train and evaluate frontier models on real-world work with human expertise. The company now extends that setup to enterprises. Its evaluations aim to establish what agents can do, what they cost and how their performance can improve.

Mercor’s custom evaluations use an employer’s actual workflows to test models against company requirements. They also identify where additional training or practice in realistic environments could improve results.

Mercor reports that its network includes more than five million experts paid to train frontier AI models. In the United States, more than 800 distinct occupations contain workflows against which models have yet to be evaluated.

Mercor’s enterprise playbook starts with a capable open model. The company then builds public evaluations, uses post-training to reduce costs and trains models for customers. It separates general task knowledge from employer-specific execution under a company’s systems, policies and quality standards.

Mercor describes post-training as requiring a fraction of the resources needed at the frontier. Its frontier-lab and enterprise work share the same expert network, evaluation methods and post-training infrastructure, allowing lessons from each side to inform the other.

In plain terms, Mercor is selling employer-specific practice, not generic AI knowledge.

What does the Deeptune acquisition add?

Earlier in 2026, Mercor acquired Deeptune, a reinforcement-learning environment company. Mercor’s July 9, 2026 announcement outlined the planned acquisition. Deeptune had spent two years recreating hundreds of enterprise applications, from spreadsheets to Salesforce, to build agent-training environments where agents can practice professional workflows.

Each environment has three components: software where work occurs, tasks that define what the agent must accomplish, and verifiers that determine whether it succeeded. Mercor’s experts create the tasks and verifiers, translating human judgment into learning signals.

The July announcement said APEX measures model performance on economically valuable tasks across thousands of real-world workflows. Deeptune supplies the software layer needed to turn that expertise into training environments.

In plain terms, Deeptune adds the simulated workplace that turns expert judgment into agent training.

Why does inference spending run three times above payroll?

Mercor reports that it spends three times as much on inference—the computing used to run models—as on employee salaries. The company says this spending pays for itself by making employees better at their jobs.

Co-founder and chief executive Brendan Foody said that if the ratio holds as AI spreads through the economy, it points to “double-digit growth in gross domestic product over the next five years,” nearly three times the rate he described as customary.

The new capital supports what Mercor calls a single enterprise proposition: “build the environment, supply the experts and measure whether an agent is ready.” The practical test is reliable performance inside a customer’s own workflows, where general task knowledge alone is insufficient.

In plain terms, Mercor says better workers justify the computing bill, while reliable in-workflow agents determine whether the enterprise model scales.

The $750 million expands Mercor’s reach; its next test is whether the shared expert, evaluation and simulation infrastructure can produce repeatable performance inside enterprise systems.