Mercor CEO: LLM inference now costs 3x salaries, path to 10% GDP growth
gabriel1 · x · 2026-09-14
Mercor CEO Brendan Foody says his company now spends 3x as much on LLM inference as on employee salaries, with inference spend additive to headcount rather than replacing it.
His argument: within 5 years, companies will spend as much on inference as they do on knowledge-worker compensation today ($40T/year), and if wages stay flat while firms profitably absorb that spend like AI-native companies do, the economy would need $40T/year of additional final output — on the order of 10% GDP growth.
The quote-poster adds that soon every piece of corporate information — internal messages, analytics, tickets, meeting transcriptions — will pass through AI, and habits are the only blocker.
More from AGI Musings
- AI safety debate: uncertainty doesn't automatically privilege catastrophe scenarios — inductionheads · 2026-09-14
- The orthogonality thesis debate: where alignment skeptics and MIRI actually diverge — EigenGender · 2026-09-14
- Anthropic Researcher Quits Over AI Risks as Doomsday Debate Snowballs Across OpenAI, Meta, Google — RebeccaBellan · 2026-09-14
- Alignment Debate: Indifferent AI Is Fatal, or Will ASI Love Us by Accident? — EigenGender · 2026-09-14
- Reporter Claims Subject 'Rejected AI Slowdown' — Transcript Says Otherwise — tszzl · 2026-09-14
- C. Thi Nguyen's 'Value Capture': how quantified metrics replace our richer values — round · 2026-09-14