Every large language model needs to learn from somewhere, and increasingly, that "somewhere" is a paid network of human experts evaluating and correcting its outputs. Mercor built a business around being the broker for that network — and did it so effectively that its valuation grew 40 times over in a single year.
Two Businesses in One Company
On its surface, Mercor looks like a modern hiring platform: it automates resume parsing, matches candidates to roles semantically, and even conducts initial interviews through an AI-driven virtual avatar. That's a legitimate, functional product. But it's not where most of the company's revenue actually comes from — the larger business is a marketplace connecting AI labs with narrow domain experts for model training and evaluation work.
The Elite Staffing Agency for AI Training
Think of Mercor's core business as a highly specialized staffing agency, except the client isn't a corporation hiring a permanent employee — it's an AI lab like OpenAI, Anthropic, Meta, or Google DeepMind that needs, say, 50 pediatricians to label a medical training dataset for a few weeks. Mercor finds those pediatricians, vets their credentials, onboards them into the project, and manages payment — at a scale and speed that would be nearly impossible for an AI lab's internal recruiting team to replicate.
The caliber of talent flowing through this pipeline is notably high-end: former Goldman Sachs bankers and McKinsey consultants are among those filling out evaluation forms, writing detailed reports, and grading model responses, training AI systems to competently handle specialized, closed-domain professional knowledge.
An Economics Story That Sounds Impossible Until You See the Numbers
Mercor's business model involves paying out more than $2 million per day to the experts on its platform — a burn rate that would sink most companies. Mercor stays profitable anyway, because the AI labs paying for that expertise pay Mercor even more than Mercor pays out. That spread, multiplied across a rapidly growing base of AI lab clients desperate for high-quality training data, is the engine behind the company's valuation.
Extraordinarily Young Founders, Extraordinarily Fast Growth
Mercor's founders were 21 and 22 years old when they started the company — making them among the youngest builders of billion-dollar startups in modern venture history. The growth trajectory that followed is almost as remarkable as the founders' age: Mercor's valuation moved from $250 million in September 2024 to $10 billion in October 2025, a 40x increase in just 12 months.
What Mercor Reveals About the AI Value Chain
Mercor's rise illustrates a broader truth about where money actually flows in the current AI boom: the most profitable position isn't always the one building a headline-grabbing model — it can be the one solving the unglamorous, high-friction problem of finding and paying the right specialized humans to make those models actually usable in professional domains. As long as frontier AI labs keep needing expert human judgment to refine their models, a well-run marketplace connecting supply and demand for that judgment has a durable, high-margin business underneath it.
FAQ
What does Mercor actually do? Runs a hiring platform on the surface, but its core revenue comes from a marketplace of experts who train and evaluate AI models for major labs.
How much does Mercor pay experts? More than $2 million per day, while remaining profitable because AI labs pay even more.
How old were Mercor's founders when they started? 21 and 22 years old.
How much is Mercor worth? $10 billion as of October 2025, up from $250 million in September 2024.
Source: CrackTheDeck Research.