The $500M Lesson: Your AI Bill Is a Strategy Problem, Not a Tech Problem

Last month a company spent half a billion dollars on AI in a single month. Not over a year. Thirty days. An AI consultant told Axios it happened because nobody had capped what employees could run on their licenses. The model did not fail. The buyer did. Thousands of people got an open tab, and a chunk of that money went to autonomous agents grinding through tasks a human could finish in seconds. Some of it went to people asking a frontier model to check the weather.

The lesson everyone wants to take from this is that AI is too expensive. That is the wrong lesson.

AI is a meter, not a seat

SaaS trained us to think in flat per-seat subscriptions. AI breaks that. The cost scales with behavior, and agentic behavior scales fast. You would never hand every employee an uncapped corporate card and walk away. Token access is the same thing. The discipline is not technical, it is managerial.

Decide what AI is actually for in your business. Tie a budget to outcomes, not enthusiasm. Meter it. The founders who win are not the ones who spend the most on AI, or the least. They are the ones who know their cost per outcome and can defend it to a board.