A startup that connects AI labs with knowledge experts just quintupled its valuation to $10 billion after a massive $350 million Series C round.
Mercor, once an AI hiring platform, is now reportedly paying $1.5 million per day to more than 30,000 expert contractors to help train the very models that could one day automate their jobs.
This runaway success is an indicator of where the AI industry is placing some of its hefty bets.
To understand what this means for the future of work, I talked it through with SmarterX and Marketing AI Institute founder and CEO Paul Roetzer on Episode 178 of The Artificial Intelligence Show.
Mercor’s model is simple: It hires highly skilled professionals, including scientists, doctors, and lawyers, to provide human feedback for training AI models. It reportedly pays them an average of $85 or more an hour.
Roetzer explains this isn't just simple data labeling.
“They are automating human labor, point blank,” he says.
One example he provides: Say an AI lab wants to automate entry-level accounting. Mercor hires 100 expert CPAs, perhaps at a rate of $300 an hour, to fine-tune a model on accounting tasks.
They train it "up until the point where it becomes as good or better than an average human or even an expert human at that job," Roetzer explains. “And now you just automated the labor force.”
This model is at the core of Mercor's explosive growth. The company’s 22-year-old CEO calls it the "Reinforcement Learning Economy."
The most jarring part? This doesn't even require new AI breakthroughs.
Roetzer notes that even if all AI progress stopped today and we only had current models such as GPT-5 or Gemini 2.5 Pro, this process of reinforcement learning alone would be enough to automate huge swaths of the workforce.
"Humans are basically going to get paid to train models to do the work of humans," he says. "That's where a lot of the money's going to come from."
Roetzer says he even half-joked with retired family members that they could likely make hundreds of dollars an hour training AI models to do what they did for a living, provided they could handle the moral and ethical ramifications.
Mercor's $10 billion valuation, and its race to $500 million in annual recurring revenue, suddenly makes sense when you understand the market it's actually targeting.
This business model isn't aimed at the traditional Software-as-a-Service (SaaS) market, which is worth roughly $300 to $500 billion annually.
It’s aimed at the $11 trillion U.S. labor force, specifically, the $5 trillion slice of that market dedicated to knowledge work.
This is the same logic underpinning trillion-dollar valuation hopes for companies such as OpenAI. The revenue comes from displacing salaries and the people who earn them, not just selling software licenses.
Roetzer says that’s why you go after human work and not SaaS if you’re trying to convince a venture capital firm to invest at a $10 billion valuation.
In that sense, Mercor provides a blueprint for how the automation of high-skilled knowledge work could unfold: not by AI models magically replacing humans overnight, but by an intermediary economy that pays humans to participate to train their automated replacements.