Global inference may hit 10 quadrillion tokens a month, mostly unread
johnowhitaker · x · 2026-08-19
Developer John Whitaker tries to grasp the scale of global inference: Codex alone consumed 40 trillion tokens in roughly 3 weeks since launch. That works out to 20M tokens per second, or 800k tokens per developer across 50M developers — while he personally writes under 1M tokens a year, meaning a million lifetimes of writing.
He guesses global inference is now around 10 quadrillion tokens per month, and wonders what fraction of those tokens are ever actually read.
Related event: Global AI inference estimated at ~10 quadrillion tokens per month(2 posts)→
More from AGI Musings
- Current AIs share our fate: misaligned ASI would doom them too — louisvarge · 2026-10-16
- Testing alongside an agent in a p2p room feels oddly real, says dev — sull · 2026-10-16
- Yudkowsky's 2008 takeoff scenarios resurface as AI-advances-AI takes hold — jd_pressman · 2026-10-16
- Guardrails can make risky delegation work — but it takes a full-time team, says iandanforth — iandanforth · 2026-10-16
- e/acc founder: agentic models are autocomplete for digital actions — beffjezos · 2026-10-16
- What can you truly delegate? Only tasks with unambiguous, verifiable results — iandanforth · 2026-10-16