Token economics model projects $1.14T global API spend by 2030, but only 1% of users pay beyond coding

AccBalanced · x · 2026-11-22

Trader Jared Kubin published a full token-economics spreadsheet modeling where AI revenue can come from: global API spend growing 11.5x from $99B (2026) to $1.14T (2030), anchored to Anthropic's audited 70.6% US revenue share. Key findings: 76% of AI app spend goes to coding tools and chat assistants; sustaining 2030 numbers requires every office worker to pay $13,686/yr while only 1% of money currently comes from non-coding users; a 40% cut to chip forecasts lowers the sellable token cap to $779.8B. Open-weight models win tokens (72% of gateway volume) but only 25% of spend, with closed tokens costing 7-8x more. Includes a drawdown checklist for AI-semis investing.

Original post →

More from Venture

Venture channel →