From tokenmaxxing to budget caps: the two token economics of enterprise AI
WhatTheLJW · x · 2026-08-25
Leonis Capital's essay "The Two Token Economics of AI" traces how enterprise AI token spend flipped from "burn more" to "the bills arrived":
- The burn era: Jensen Huang said a $500K engineer should consume at least $250K of tokens a year, and Nvidia was working toward a $2B annual token budget for its own engineers; "tokenmaxxing" became a named behavior of employees burning tokens to look productive.
- The bill era: Uber blew its entire 2026 AI coding budget by April and capped engineers at $1,500/month, with its COO admitting the link between token spend and output "is not there yet"; Microsoft revoked Claude Code licenses months after issuing them.
The essay examines the economics behind this divergence.
More from Companies & People
- Radical Numerics opens 19 roles including 16 AI jobs to build the largest models in science — exnx · 2026-08-26
- Bloomberg opens applications for its 2027-2028 Data Science PhD Fellowship — mdredze · 2026-08-25
- Decentralized AI: Bittensor Battles for AI Ownership Against Big Tech — bittingthembits · 2026-08-25
- a16z: How AI Shifts Startups from Engineering-Bound to Capital-Bound — a16z · 2026-08-25
- RadicalNumerics Hiring: 20+ Roles in Experimental Biology and AI Model Training — exnx · 2026-08-25
- Goldman Sachs partner warns AI could cause 'cognitive atrophy' in bankers — rohanpaul_ai · 2026-08-25