Mercor CEO: Why Application Layer Companies Have No Moat & The Cost of Hiring AI Researchers

Company Performance & Financials

• McCor is currently valued at over $10 billion and reports revenue exceeding $1 billion annually.
• The company added $300 million in net new Annual Recurring Revenue (ARR) within a 60-day period.
• McCor has maintained profitability since burning only $500,000 post-seed; it currently holds more cash than it has ever raised.
• The company is paying out over $3 million daily in the "fastest job category ever created" (AI training agents), a figure Brandon Foody projects will triple to $9 million within 12 months.
• Internal projections indicate token spend for AI agents currently exceeds total employee headcount costs.

Security Incident & Market Position

• McCor experienced a security incident on a Saturday which involved attackers utilizing swarms of coding agents to exhaustively review codebases; the breach was contained quickly.
• Following the incident, McCor added a seventh core value focused on security and hired security consulting firms like Mandiant.
• False rumors claimed McCor lost OpenAI and Meta as customers; Foody confirms the OpenAI relationship is "stronger than ever," while the Meta relationship is paused due to other factors (including the Scale acquisition), not the hack.
• Regarding rumors of poaching Miro One team members, McCor denies issuing a single offer; the story originated from an employee sending outbound messages with high signing bonuses ($500k) without legal offers being finalized.
• Amazon offered to acquire McCor for $13 billion, which was false; Foody states he would not sell the company for $30 billion given his motivation to solve human-economy integration as an independent entity.

AI Trends, Strategy, and Defense

• Foody argues that "the model is the product," asserting that building defensibility in the software layer on top of models is increasingly difficult as models can now clone end-to-end applications (e.g., Slack) within 12 months.
• Infrastructure companies (compute) and data moats are viewed as having more sustainable competitive advantages than application-layer companies.
• McCor believes the API layer will become commoditized with zero switching costs, driven by enterprise "system of records" for evaluating model performance on specific workflows.
• McCor is expanding data collection in the physical world (e.g., surgeons, mechanics) but prefers a horizontally scalable talent network model over niche vertical data providers.
• The company is developing the "AI Productivity Index" (APEX) to measure which jobs/tasks AI can automate, noting frontier model performance on these tasks rose from 0% to 40% in 12 months.

Economic Impact & Future Outlook

• Foody predicts that in five years, the average enterprise will spend more on compute/inference than on human headcount.
• He anticipates a "golden age" of AI security engineering, with a boom in tools to defend against agent-based attacks and a consolidation of the market due to current "frothy" funding conditions.
• Foody supports eliminating income tax for the bottom 50% of Americans, suggesting taxation should shift to capital gains, consumption, or carbon instead.
• He expresses concern over the concentration of value in the top 8-10 US tech names but views it as efficient capital allocation despite the societal inequality implications.
• In 12 months, McCor expects the "training agents" job category to become the primary function of knowledge work, moving away from humans performing repetitive tasks.
• Foody believes OpenAI and Anthropic could each reach a $10 trillion valuation within five years, though most future inference will likely shift to open-source or fine-tuned models.
• He predicts a multi-chip future where NVIDIA's monopoly diminishes as labs build in-house chips, though it will remain the most valuable company regardless.

Operational & Hiring Landscape

• The most difficult role to hire is the AI researcher, with compensation for top talent reaching tens of millions in stock annually due to a supply-demand ratio of 10:1.
• Meta's intelligence group (TBD) has reportedly offered employees $20 million in annual cash/stock to compete with McCor's talent retention.
• McCor employs roughly 150 people in its delivery organization, now utilizing an AI project manager to automate the hiring and coordination of experts end-to-end.
• The company has never mandated working hours despite high dedication from leadership, emphasizing sustainability for employees with families.

Investment & Leadership

• McCor is planning to go public in the next few years but is not rushing the process while under three years post-college dropout.
• Foody cites Jeff Bezos and Jensen Huang as the "coolest" leaders he has met and expresses a desire to have Bezos as an investor.
• He acknowledges the "lump of labor fallacy," arguing that technology revolutions historically create more jobs than they destroy, citing a 25x productivity increase over 250 years.
• McCor received a Series B at a $10 billion valuation (approx. 25x revenue) and a previous round at $2 billion (100x revenue), both of which were considered "uncomfortable" by the market at the time due to 50% month-over-month growth rates.