Enterprises Blowing Through Full-Year AI Token Budgets in Two Months
Multiple large organizations have consumed their entire 2026 AI token allocations within the first two months of the year, according to Wharton professor Ethan Mollick, who gathered reports from enterprise customers navigating the issue. Organizations have no established processes to budget, forecast, or govern token consumption, and are veering between policies of maximum AI use and reverting to local models depending on which manager sets direction. LangChain's simultaneous launch of an LLM Gateway product — specifically marketed as protection against agents burning through $10,000 overnight — signals vendor recognition of the same governance gap.
Why It Matters
Token cost governance is the fastest-emerging AI operations problem in enterprises: a purchasing and budgeting challenge that currently has no standard framework, forcing ad-hoc engineering solutions or flat access restrictions that reduce competitive advantage.