Mercor’s $10B marketplace pays ex-bankers $200/hour for AI training data

Mercor CEO Brendan Foody revealed that AI labs like OpenAI, Anthropic, and Meta are using his platform to hire former employees from investment banks, consulting firms, and law firms to train AI models on industry workflows. This strategy bypasses companies that are reluctant to share proprietary data that could ultimately be used to automate their own operations, creating a controversial new channel for AI training data.

What you should know: Mercor has built a marketplace connecting AI labs with industry experts who possess deep knowledge of specialized workflows.

• The company pays contractors up to $200 per hour to fill out forms and write reports for AI training purposes.
• Mercor now has tens of thousands of contractors and distributes more than $1.5 million daily to them.
• The startup has reached roughly $500 million in annualized recurring revenue and recently raised funding at a $10 billion valuation.

Why companies resist sharing data: Traditional enterprises are hesitant to provide information that could automate their core business functions.

• “There is an argument that Goldman Sachs doesn’t love the idea of having models that are able to automate their value chain,” Foody explained.
• Companies worry about shifting competitive dynamics when AI labs can replicate their specialized processes.
• This resistance forces AI labs to seek alternative sources of industry expertise through former employees.

The big picture: Mercor represents a new model for extracting industry knowledge that sidesteps corporate gatekeepers.

• Former employees carry valuable institutional knowledge that technically belongs to them rather than their previous employers.
• Some contractors maintain their day jobs while contributing data on the side, though Mercor instructs them not to upload company documents.
• The platform essentially creates a “gig economy” for specialized knowledge work, similar to how Uber transformed transportation.

Competitive landscape: Mercor has benefited from disruptions affecting established players in the AI training data market.

• The company quintupled its value in the past year, though it remains smaller than competitors Surge and Scale AI, both valued above $20 billion.
• Many AI labs shifted away from Scale AI after Meta made a large investment in the company and hired its CEO.
• Early data vendors like Scale AI initially focused on simple labeling tasks in developing countries, while Mercor targets high-skilled U.S. knowledge workers.

What they’re saying: Foody believes this approach will fundamentally reshape professional services industries.

• “Over time, ChatGPT will be better than the best consulting firm, better than the best investment bank, and better than the best law firm,” he predicted.
• “That’s going to transform the economy radically, which will be a broadly positive force that helps to create abundance for everyone.”
• He acknowledged potential ethical concerns but argued that knowledge “in an employee’s head belongs to the employee, and not their company.”

Looking ahead: Mercor plans to expand beyond AI labs to serve traditional industries directly.

• The company expects to partner with law, finance, and medical companies that want help leveraging their data for AI agent training.
• Foody envisions companies will embrace this “new future of work” rather than resist it.
• Some job postings already blur the line between requesting employee knowledge and company data, including seeking startup CTOs who “can authorize access to a substantial, production codebase.”