Apple is reportedly tightening how much employees can spend each day on AI tools, trimming some teams’ allowances to $200 from $300 even as the company tries to spread these tools across its work. Bloomberg says Apple lowered daily “token” budgets for some engineering teams last month and is considering further cuts for certain groups that could take limits down to $100. Some non-engineering teams are already capped at $100 a day, and those limits could drop as low as $25, though staff can ask senior leaders for exemptions. Tokens are the metered units AI models charge for when they read and generate text, and the budget covers products like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. The move echoes a broader shift across tech: after an early burst of “try AI everywhere,” companies are starting to treat usage like any other operating expense that needs guardrails and proof it’s paying off. A daily token cap acts like a built-in throttle: hit $200 and usage pauses unless someone secures an exception, turning AI from an open-ended experiment into a procurement decision where extra consumption must clear an internal return-on-investment test. When that happens at big customers, it can push AI vendors toward volume discounts, fixed per-seat plans, or stricter controls, rather than relying on simple pay-per-token growth, making revenue lumpier and pricing power more dependent on winning budget approvals than on raw user enthusiasm.