AI spending meets a tighter budget
Apple’s internal controls on AI spending are tightening as its research and development costs climb. The company spent $45.8 billion on R&D in fiscal 2026, up 32% from the prior year, with higher infrastructure and AI expenses contributing to the increase. That figure covers a broad area of work, so it doesn’t show how much Apple spends on employee AI tools alone. It does, however, put the new daily limits in context: the company is asking staff to use AI while scrutinizing the costs that come with that push.
Bloomberg reports that Apple lowered some engineers’ daily AI allowances from $300 to $200 last month. The company is weighing a further cut to $100 a day for certain engineering groups. Non-engineering teams currently have a $100 daily limit, which Apple is considering reducing to as little as $25. Employees who need additional capacity can request an exception from senior executives, leaving a route for work that exceeds the standard cap.
These are reported spending limits, not a companywide ban on AI use. Their practical effect will depend on how often employees reach them, which tools and tasks count against the allowance, and how exceptions are handled. The available account doesn’t answer those questions, but they’ll determine whether the caps act as a light check on routine use or a real constraint on more intensive work.
Usage limits put a price on experimentation
The caps apply to tokens, the units of text and other data that AI models process. Providers generally charge according to token volume, and more capable models often cost more to run. Software development and other coding tasks can consume substantial amounts, making engineering use a particular cost consideration. A short exchange may be relatively inexpensive, while repeated prompts, long inputs, or complex coding work can use far more capacity.
That pricing model makes a daily allowance easy to understand as a budgeting tool, but it also puts a meter on experimentation. Employees may need to try several prompts or compare outputs before a tool is useful for a task. When each attempt draws from a fixed allowance, workers have a reason to be selective. That can reduce avoidable use; it can also make people hesitate before trying a tool on work where the value isn’t yet clear.
Apple is asking staff to use AI more while keeping the bill down, according to Bloomberg. That creates a practical tension: broad access may help reshape engineering and internal processes, but unrestricted use makes costs harder to control. The new limits suggest the company is trying to measure that trade-off more closely. This is more significant than it looks: a spending cap can influence which tasks employees choose to hand to AI, even when the tool itself remains available.
Employee reactions are mixed. Some reportedly consider the former $300 daily engineering allowance more than most workers needed. Others see the reductions as part of a wider cost-cutting effort and worry that it could eventually extend to jobs. The concerns aren’t evidence that the caps will lead to staffing changes; they show that employees are reading the policy against a broader organizational backdrop.
Those concerns come as Apple reviews how it operates. Bloomberg reported in late September that new CEO John Ternus, who took over on September 1, is seeking a faster, leaner organization with fewer management layers. Apple has also brought in outside consultants to assess operations and potential efficiencies. The reported AI limits fit that focus on controlling costs, though the available information doesn’t establish that the limits are part of a single companywide program.
Cost scrutiny spreads across tech
Apple’s approach fits a wider shift among technology companies: employee access to AI tools is increasingly paired with spending controls. Uber capped monthly token spending at $1,500 per employee in June. Meta said that month it planned to limit employee AI use, and Tesla set a $200 weekly cap in July. Those examples show that companies are testing ways to make AI costs visible at the employee level, though the limits differ in both scope and timing.
Similar systems typically try to balance access with accountability. A cap can help managers see where demand is concentrated and give employees room to use tools within a defined budget. But a budget alone doesn’t show whether the work produced is useful. If you're working in this space, the distinction matters: a low bill can reflect careful use, but it can also reflect employees avoiding tasks that might have benefited from AI. The quality of the work and the amount spent need to be considered together.
The pressure is visible even at Apple’s scale. The company’s market capitalization was reported at $4.98 trillion, and it has made AI a strategic priority. Yet rising R&D expenses and new usage limits point to a sharper focus on the cost of turning that priority into everyday work. A large market value doesn’t remove the need to account for operating costs, and it doesn’t by itself show whether current AI spending is producing a return.
What the limits may reveal
The next signal to watch is how Apple balances those limits with its push to embed AI in employee workflows. If teams can maintain productivity within tighter allowances, the controls may become part of routine cost management; if not, the company may face pressure to revise where and how staff can use the tools. The key question is whether employees can still reach the models and capacity their work requires, including through the exception process.
Apple’s policy may also show how it judges AI’s internal value. If some groups repeatedly request more capacity, that could help identify where demand is strongest, though spending alone won’t prove that a workflow is more productive. And yet, limits can provide useful information when they’re paired with a clear view of how the tools are used and what work they support. For now, the reported cuts signal closer scrutiny; they don’t establish how much Apple expects to save or what results it expects from employee AI use.
For staff, the immediate change is concrete: some engineers have less daily capacity, and other groups may see their allowances fall as well. For Apple, the test is whether it can keep AI useful in routine work while making its costs easier to track. A tighter cap can set a boundary. The harder question is what happens inside it.
