Companies are cutting AI spending after token costs exploded in early 2025, spurring a new market for cost-tracking tools and a Linux Foundation standards body.

Major tech companies have exceeded their 2026 AI budgets months in: Uber spent its entire year's coding budget by April; Microsoft revoked Claude Code licenses from developers; a Priceline employee reported a Cursor contract renewal came back 4–5× more expensive.

Per-developer token consumption has risen about 18.6× in nine months, driven by agentic features (AI agents that autonomously execute tasks). A March Faros survey of 20,000 developers found engineers using the most tokens were about twice as productive but spent 10× the tokens; Jellyfish found similar patterns but noted most companies cannot yet measure the business value of that spending.

The Linux Foundation this week unveiled the Tokenomics Foundation, a new standards body (modeled after FinOps for cloud cost discipline) to create common definitions and metrics for token usage and billing. The foundation plans a formal launch in July and will define metrics including cost-per-intelligence and tokens-per-watt.