Apple Slashes Engineers' Daily AI Budget From $300 to $200

By Nathan Brooks

Apple cut its engineering teams' daily AI spending cap from $300 to $200 in October, Bloomberg reported, and deeper reductions are already on the way. Sources told Bloomberg some engineering allowances will drop to $100 a day. For employees outside engineering, Apple is weighing a limit as low as $25.

That's a one-third cut overnight for the teams doing the company's most AI-intensive work — and up to a 92% squeeze for everyone else if the $25 figure sticks. Put in per-year terms: an engineer's theoretical annual AI budget falls from roughly $109,500 to $73,000, then to about $36,500 under the planned floor.

The retreat marks a sharp turn for a company that has spent years embedding AI into its engineering processes. It's also the clearest signal yet that the industry's free-spending phase is over.

What happened to 'tokenmaxxing'?

Earlier this year, a wave of what insiders call "tokenmaxxing" swept through corporate America. Companies pushed employees to use AI as heavily as possible, chasing loosely defined efficiency gains.

Then the bills arrived. Token costs mounted, and the enthusiasm faded fast. Many companies concluded that rationing usage was simply the more financially prudent strategy, according to the report.

The pattern is a familiar one for anyone who has lived through earlier enterprise tech cycles: unlimited pilots, then finance catches up, then caps.

What does this change for practitioners?

If you run engineering or AI tooling at an enterprise, Apple's move tells you the audit era has started. The companies that once encouraged maximum consumption are now budgeting AI like cloud spend or travel expenses — per user, per day, with a hard ceiling.

The numbers matter because they set a reference point. A $200-per-engineer-per-day cap from one of the world's most cash-rich companies signals that even deep-pocketed buyers don't see unbounded usage paying for itself.

Bloomberg's report frames the cutbacks around two connected pressures:

• Soaring token bills. Heavy usage drove costs up faster than the loosely defined efficiency gains justified.
• Unclear productivity returns. The gap between what AI costs to build and run, and the benefits it actually delivers, continues to weigh on enterprise adoption decisions.

Why the productivity question still has no answer

The second point is the harder one. Companies still can't reliably measure what heavy AI use produces in return. That uncertainty, Bloomberg notes, keeps weighing on adoption decisions across the enterprise market.

Until someone cracks the measurement problem — tying token spend to shipped code, resolved tickets or measurable output — expect more caps, not fewer. Apple's tiered structure, with engineers getting up to eight times the allowance of other staff, suggests the company is already allocating budget where it believes usage pays off.

For vendors, the message is tougher: customers are learning to consume less, not more. For teams building internal AI workflows, it means justifying every query.

Apple hasn't said whether the cuts will hit product features that depend on the same infrastructure. But with further reductions planned, don't be surprised if other large enterprises copy the per-day ration model in the coming quarters.