Mercor's CEO says it now spends more on AI tokens than employee salaries

The landscape of corporate spending is changing dramatically as Mercor's CEO reveals that his startup is now allocating more resources to AI tokens than to employee salaries. During a recent guest appearance on the '20VC' podcast, CEO Foody shared insights into this striking trend, stating, "Right now we're spending more on tokens for our internal agents than we are on employee head count. That's correct. It's pretty incredible." Founded in 2023, Mercor has rapidly ascended to become a key player in the AI sector, boasting a valuation of $10 billion. The company specializes in assisting organizations like OpenAI and Anthropic in training their AI models through a network of human experts. As of October 2025, it employed approximately 300 people, according to PitchBook data. Foody elaborated on the diverse applications of AI agents within the company, which range from project management and recruitment to accounting and fraud detection. Impressively, Mercor has facilitated over 5 million AI-assisted interviews to date. He believes this shift in spending patterns is indicative of a larger trend that could reshape corporate America in the coming years. "I would bet that in five years, the average enterprise will spend more on compute than on headcount," Foody asserted. His comments emerge as a dialogue continues among business leaders about the tangible returns on investments in AI. Uber's COO, Andrew Macdonald, recently expressed skepticism, noting the lack of a clear correlation between increased AI spending and productivity enhancements. Foody explained that the declining costs of AI technology, combined with enhancements in model performance, are leading to a phenomenon akin to Jevons paradox, where cheaper AI results in a surge in consumption rather than a decrease. He emphasized that Mercor evaluates the effectiveness of various AI models for specific business tasks to ensure optimal value. Looking ahead, Foody anticipates a future where AI becomes a fundamental operational expense for enterprises, potentially rivaling or exceeding the costs associated with human labor. "Humans will still play an important role at the things models can't do," he added. "But I expect that the cost of inference and compute will surpass that of labor."