Apple is reportedly cutting the daily amount its engineering teams can spend on AI. According to Bloomberg, the company lowered the cap from $300 to $200 a day in October, and sources told the outlet that some allowances could fall to $100. Outside engineering, Apple is reportedly weighing a daily limit as low as $25. TechCrunch has reached out to Apple for comment, and the company has not yet responded in the material provided.
The change matters because Apple has worked to integrate AI into its engineering processes. A company with that kind of commitment does not usually tighten spending unless the numbers stopped making sense. The report points to a broader pattern: earlier this year, companies pushed staff to use AI as much as possible in the name of vaguely defined efficiency gains, and the frenzy faded once the bills arrived.
That sequence should worry anyone who approves technology budgets. The report describes a gap between what AI costs to build and run and the often unclear productivity gains it is supposed to produce. A spending cap is a blunt tool, but it is at least an admission that usage was not being measured against results. Rationing tokens is what a company does when it has no clear way to say which AI use is paying off.
My view is that companies should stop treating AI usage as a proxy for progress. Before any team gets a larger daily allowance, it should be able to show what the spending produced: shipped work, reduced cycle times, or fewer defects, tracked against the same baseline before the tools arrived. Where a team cannot show that, the cap should stay low. Apple's cuts, as described, suggest the company is moving in that direction, but the report does not say what metrics, if any, drove the decision.
The lesson is not that AI is worthless. It is that enthusiasm and cost discipline have to arrive together. A budget that tracks token consumption without tracking outcomes is only half a budget.