Tokens · per user vs. per company · 2026

Who actually burns the tokens?

A person typing into a chatbot moves tokens at human speed. A company running APIs, agents, and pipelines moves them at machine speed, 24/7. On a per-entity basis, it isn't close. Note the log scale — each gridline is 1,000× the last.

Typical individual usera few chats a week

~1M / yr (est.)

Heavy power userdaily, long conversations

~50M / yr (est.)

Solo developer / prosumerbuilding on the API, coding agents

~5B / yr (est.)

A single large enterpriseone of Google's 375 ≥1T-token customers

>1 trillion / yr

One platform, all surfacesGoogle, May 2026 run-rate

~38 quadrillion / yr

tokens / year →

~1,000,000×

A single large enterprise runs on the order of a million times the tokens of a typical individual — and that's one company out of hundreds. That's great news for you: machine-scale demand is exactly what funds the infrastructure and drives the price crashes that make a powerful $20 personal plan possible.

Sources — Google I/O 2026 keynote / Google AI Blog (3.2 quadrillion tokens/month across all surfaces; 375 Cloud customers each >1T tokens in 12 months; 8.5M developers). CEIBS, Apr 2026, citing provider disclosures (Gemini ~14T tokens/day via API, Q4 2025; OpenAI ~8.6T/day, Oct 2025; Microsoft >500T, FY2025). Menlo Ventures, "State of Generative AI in the Enterprise 2025" (enterprise genAI spend $37B, 3.2× YoY).

Method — Individual/solo-dev figures are illustrative order-of-magnitude estimates; enterprise and platform figures are from named disclosures. Log scale: bar length reflects powers of ten, not linear volume.

The honest caveat — In aggregate, consumers are not tiny: billions of users add up, and consumer image/video generation is token-heavy. The gap is per-user and per-workload, not total.