Ing. Calogero Bono
Apple is reducing the amount employees can spend each day on artificial intelligence, according to Bloomberg, with TechCrunch, MacRumors and 9to5Mac picking up the story. In October the daily cap for some engineers dropped from 300 to 200 dollars, and further cuts are reportedly coming. The news matters because it shows how even large companies are starting to face the real cost of AI at work.
A few terms before we get into the facts. Token is the unit AI models use to measure the text they read and produce, and it is what you pay for. Tokenmaxxing refers to the practice of consuming as many as possible, often as a measure of productivity. Spending cap is the daily limit each employee can consume. On-device AI means models that run locally on the device, with no cost for each request to an external service.
What did Apple decide about AI spending limits?
According to TechCrunch, citing Bloomberg, in October Apple lowered the daily cap for engineering teams from 300 to 200 dollars. Further cuts are in preparation and some quotas could fall to 100 dollars a day.
Outside engineering, reports MacRumors, many groups already have a 100 dollar limit. For some teams Apple reportedly considered a cap of just 25 dollars.
How does daily AI use change for teams?
Some teams are encouraged to use less expensive models or models that run locally on the devices. Anyone who needs more budget can ask management for an exemption.
Engineers have the highest quotas because using AI to write code consumes many tokens. Apple employees, reports 9to5Mac, have access to Anthropic's Claude, OpenAI's ChatGPT and Google's Gemini. It is not clear which tools are affected by the new limits.
Why is Apple cutting right now?
The context is the cost cutting pushed by CEO John Ternus, in office since September 1. According to the report, he wants to reduce hierarchies and inefficiencies to do more with fewer people. Apple has also hired outside productivity consultants for some divisions. We wrote about the change at the top in John Ternus at his first quarterly results as CEO.
The cut comes as spending grows. Again according to 9to5Mac, in fiscal year 2026 research and development rose 32% to 45.8 billion dollars, partly for infrastructure and AI. The year's figures are in our article on the 478.8 billion in revenue.
Is the era of tokenmaxxing over?
According to TechCrunch, at the start of the year many companies pushed staff to use AI as much as possible, in the name of loosely defined efficiency gains. The phase reportedly was short, because the bills arrived and rationing seemed more sensible.
9to5Mac recalls that Meta had introduced internal rankings, unofficial ones included, based on token use. Accenture, according to rumors, reportedly reduced its central Claude allocation to 29 dollars a month per employee, with the rest charged to individual projects.
Limits to know
Everything rests on anonymous sources cited by Bloomberg, and Apple has not commented, at least in the sources we have. We do not know how much it will save, because some employees did not even use the full 300 dollar cap. The tools involved and whether the 100 and 25 dollar cuts will become permanent also remain uncertain. Internal reactions are described as mixed.
What changes for those working in digital
For a company, the lesson is practical. AI spending without a cap or a way to measure results grows quickly, and token consumption does not equal productivity. It is worth setting thresholds per team, allowing justified exceptions, and choosing the model based on the task rather than its name. To understand how to keep infrastructure costs under control, you may find our piece on Kubernetes HPA in production useful. Local models are also becoming a realistic option for simple tasks.
Our take
The cut is understandable, but the numbers tell a less dramatic story than the headline. If few people used the full cap, the measure is more a signal of discipline than a real saving. The risk is something else. A single limit for everyone penalizes those who get real value from AI and rewards those who use it little. Exceptions granted by managers help, but add bureaucracy. The real issue remains measuring benefit, not counting tokens.
Frequently asked questions
How much can an Apple engineer spend on AI per day?
It depends on the team. For some engineers the cap dropped from 300 to 200 dollars and could reach 100. Outside engineering, many groups already have 100 dollars, with talk of 25.
Who first reported the news?
The original source is Bloomberg, which cites people familiar with the matter who asked not to be named. TechCrunch, MacRumors and 9to5Mac picked it up on November 4, 2026.
Has Apple confirmed the cuts?
Not in the sources we have. TechCrunch says it contacted Apple for comment, without reporting a reply.
Which AI tools do Apple employees use?
They have access to Claude, ChatGPT, Gemini and other models. Some teams are encouraged to use cheaper models or models that run on the device.
Can the limit be exceeded?
Yes, with a request. Anyone who needs a larger allowance can ask senior management for an exemption.
What to watch in corporate AI costs
The Apple case is a clue, not yet a proven trend. In the coming months, real savings and the effect on work will matter.
- Check whether Apple confirms the cuts and whether new limits beyond 100 dollars arrive.
- Watch whether other large companies copy Accenture and move from central quotas to per-project budgets.
- Check how much AI-related research and development spending weighs in upcoming financial reports.
