Meta just killed its tokenmaxxing leaderboard. Uber burned through its entire AI budget in 4 months. One company got a $500M Claude bill nobody noticed.

The era of "use as much AI as possible" is over.

Meta told employees last week it would limit AI use after costs grew "exponentially." Amazon pulled its tokenmaxxing leaderboard. Uber exhausted its full 2026 AI coding budget by April.

One company forgot to set usage limits and received a $500 million Claude bill. That is not a typo.

The FinOps Foundation is hearing "existential crises" from companies that are 3x over their entire 2026 token budget. And it's only June.

Here is the Jevons paradox in action: token prices dropped 90% since 2023. But enterprise spending on large language models doubled since late last year. Tokens consumed grew 450% while prices fell 50%.

Per-developer AI consumption rose 18.6x in nine months. The math is simple. More powerful models. More autonomous agents. More tasks. More tokens. Bigger bills.

Deloitte's 2026 State of AI report found only 21% of organizations have mature governance for agentic AI. Gartner predicts 40% of enterprises will demote or decommission autonomous AI agents by 2027 because governance gaps surface only after production incidents.

This is not a pricing problem. This is a governance failure.

AT&T's chief AI officer says his teams save 90% by routing simple tasks to cheaper models. Salesforce now tracks "agentic work units" instead of tokens. The Linux Foundation launched the Tokenomics Foundation to build cost standards.

Audit your AI spending today. If you don't know your per-developer token consumption, you're already behind. The companies that survive this are the ones treating AI costs like cloud costs: with budgets, accountability, and governance.

SOURCE: https://techcrunch.com/2026/06/05/the-token-bill-comes-due-inside-the-industry-scramble-to-manage-ais-runaway-costs/

VERIFIED: TechCrunch, New York Times, Fortune, Fortune India, Deloitte 2026 State of AI in the Enterprise

SIGNAL: The tokenmaxxing-to-tokenminning shift is the first real proof that enterprise AI spending has hit a governance ceiling. Companies that don't build cost controls now will be forced to later.