Mercor's CEO Says AI Token Spending Has Surpassed Employee Payroll
A striking milestone has emerged from one of Silicon Valley's fastest-growing AI startups. Brendan Foody, CEO of Mercor, revealed during an appearance on the 20VC podcast that his company now spends more money on AI tokens — the computational units that power large language model interactions — than it does on the salaries of its entire workforce. For a $10 billion company with roughly 300 employees as of October 2025, that is a statement that carries enormous weight for the future of business, technology, and labor.
"Right now we're spending more on tokens for our internal agents than we are on employee headcount," Foody said plainly when asked by podcast host Harry Stebbings. "That's correct. It's pretty incredible."
In an era where AI investment is accelerating at breakneck speed, Mercor's spending pattern offers a rare, concrete glimpse into what the AI-first company of the future might actually look like — and the implications are far-reaching.
What Is Mercor and Why Does It Matter?
Founded in 2023, Mercor operates as a specialized platform that connects companies like OpenAI and Anthropic with networks of human experts who help train and evaluate AI models. The company has grown extraordinarily fast, reaching a $10 billion valuation and becoming one of the most talked-about players in the broader AI ecosystem. Its business sits at a unique intersection: it both leverages AI internally and helps other companies build better AI externally.
According to PitchBook data, Mercor had approximately 300 employees as of late 2025. That relatively lean headcount makes Foody's admission even more remarkable — if a 300-person startup is already spending more on AI compute than on human salaries, the financial calculus of building and scaling companies is shifting in ways that were unimaginable just a few years ago.
How Mercor Deploys AI Agents Across Its Operations
Mercor is not simply experimenting with AI as a side feature. Foody described a company where AI agents are deeply embedded into core business functions. According to his podcast appearance, Mercor uses AI agents across a wide variety of operational areas, including:
- Project management: AI systems coordinate workflows, track deliverables, and manage timelines with minimal human oversight.
- Recruiting and candidate evaluation: AI agents screen applicants, conduct preliminary assessments, and rank candidates for human review.
- Accounting and financial operations: Routine financial tasks are automated through intelligent agent pipelines.
- Fraud detection: AI monitors transactions and behavioral patterns to flag anomalies in real time.
Most impressively, Mercor has reportedly conducted more than 5 million AI-assisted interviews. That scale of AI deployment — across mission-critical functions like hiring and financial controls — signals that the company has moved well beyond pilot projects into genuine operational dependency on AI infrastructure.
A Prediction That Should Alarm Every Business Leader
Foody's comments did not stop at describing Mercor's current reality. He went further, suggesting that what his startup is experiencing today could become standard practice across entire industries before long. In his view, the average company could soon be spending more on AI compute than on employee salaries — a prediction that should prompt serious strategic thinking in every boardroom and HR department around the world.
This is not a distant, theoretical scenario. The cost of AI inference — running large language models to generate outputs — has dropped dramatically over the past two years, even as the capabilities of those models have surged. As AI becomes cheaper per token and more capable per dollar, the economic logic of replacing or augmenting human labor with AI agents grows stronger with each passing quarter.
For businesses that have not yet begun modeling AI spending as a significant line item in their operating budgets, Mercor's trajectory is a warning signal. The question is no longer whether AI will change cost structures, but how fast and how dramatically it will do so.
The Broader Debate: Are AI Costs Generating Real Returns?
Not everyone in the technology industry is celebrating the surge in AI spending. While companies like Mercor appear to be integrating AI agents with measurable results, a growing chorus of tech leaders and investors is questioning whether soaring AI infrastructure costs are actually producing commensurate returns on investment.
The concern is legitimate. Enterprises worldwide have poured billions of dollars into AI tools, platforms, and subscriptions over the past two years, yet productivity gains and revenue impact at many organizations have been modest, inconsistent, or difficult to quantify. The gap between AI hype and AI-driven business outcomes remains a persistent challenge — and for companies that are spending heavily on tokens without a clear framework for measuring impact, the risk of misallocated capital is real.
Mercor's case, however, appears to be different in important ways. The company's AI spending is tightly coupled with core revenue-generating and cost-saving operations. When AI agents are running 5 million interviews or handling fraud detection at scale, the ROI case is far more direct than for organizations using AI primarily as a writing assistant or internal search tool.
What This Means for the Future of Work
The broader implications of Mercor's spending inversion — AI costs exceeding human payroll — touch on one of the most debated questions of our time: what happens to human workers in a world where AI can perform an expanding range of cognitive tasks faster, cheaper, and at greater scale?
The honest answer is that we are still in the early stages of this transformation. Mercor itself continues to employ humans, and its entire business model depends on human expertise to train and validate AI systems. But the trajectory is clear. As AI agents become more autonomous, more reliable, and more deeply integrated into business operations, the nature of human work will shift rather than simply disappear — moving toward higher-order judgment, creativity, and oversight roles that AI cannot yet replicate.
For professionals, entrepreneurs, and executives watching this space, Mercor's story is both a case study and a roadmap. AI token spending is no longer a footnote in a tech company's budget — it is becoming the dominant operational cost of the modern AI-native business. Understanding that shift, and preparing for it, may be the single most important strategic priority of the next decade.
Key Takeaways
- Mercor CEO Brendan Foody confirmed his company spends more on AI tokens than on employee salaries.
- Mercor is a $10 billion startup with around 300 employees that helps firms like OpenAI and Anthropic train AI models.
- The company uses AI agents for recruiting, project management, accounting, fraud detection, and has conducted over 5 million AI-assisted interviews.
- Foody predicts that most companies will eventually spend more on AI compute than on human payroll.
- The debate over whether massive AI spending generates real business returns remains unresolved across much of the industry.