Apple lowers daily AI spending limits for some engineers, from $300 to $200

What Apple changed

Apple lowered the daily limit on how much some of its engineers can spend on AI models, from $300 to $200. The change took effect last month. The company is also weighing a further drop to $100 for certain engineering teams, according to a Bloomberg report relayed in an ifanr morning briefing.

Outside engineering, many teams already work under a $100 daily cap. Apple has also discussed taking some of those teams down to $25. These are limits per day, so they act as a running brake on usage rather than a single yearly budget line.

Employees who can show a real work need can ask senior leadership for an exemption. The cap is therefore a default, not an absolute wall.

What a token budget actually is

A token is a small chunk of text, roughly a word or a piece of one, that an AI model reads or writes. Companies that use AI models through a paid service are typically charged for the tokens their requests consume. A token budget is simply a spending ceiling on that usage.

Some teams are now being asked to switch to cheaper models, or to local models, meaning models that run on the company's own hardware rather than through a paid outside service. Cheaper models cost less per request but are usually less capable, so teams have to decide which jobs really need the expensive option.

The report adds that Apple is still pushing employees to use AI more deeply in development and in internal operations. The company is not retreating from AI. It is trying to control the price of using it.

Staff are not of one mind

Reactions inside Apple are mixed. Some employees say they rarely hit the old limits, so the lower ceiling changes little for them. Others worry that cutting AI budgets is an early step in a wider push to squeeze costs.

Apple's spokesperson declined to comment, so the company has not explained the reasoning. Nothing in the report says why the cuts were made, or whether they will spread beyond the teams named. Any reading of Apple's motives is therefore inference.

What the facts do show is a clear direction. Limits are tightening at the same time as the company encourages heavier use. That combination pushes people to pick their AI jobs carefully.

A wider squeeze on costs

The same ifanr briefing carried other money-related items. Counterpoint Research's preliminary data put global smartphone shipments in the third quarter of 2026 down 19% from a year earlier. Revenue still grew 3%, because the average wholesale selling price rose 27% to a record $442.

Apple's own phone revenue rose 17% in that quarter, with shipments roughly flat. A Bloomberg Intelligence survey of Chinese consumers found about 69% plan to wait until at least 2027 to replace their phones. AI ranked last among 11 reasons to upgrade, at 8%.

The briefing does not link these figures to Apple's token budgets, and they are separate stories. Still, together they show how hardware and AI spending are both under scrutiny. Buyers and employers are asking what each purchase delivers.

What this means for other teams

Few organizations have Apple's cash, yet many face the same basic issue. AI usage costs rise with each person and each task, and the benefits are harder to measure than the bills. A cap per person is a blunt but easy tool to apply.

The more useful response is to know which tasks truly need the strongest model and which can run on a cheaper one. Teams that can answer that will cope with budget limits better than teams that cannot.