Apple has cut daily AI spending caps for some engineers from $300 to $200, and it is weighing deeper reductions that could bring some engineering teams down to $100 a day. Bloomberg reported the changes on November 4, and The Next Web’s account of the story describes the move as part of a wider push to rein in costs. It is one of the clearest signs yet that a major US employer is treating AI tool use as a budget line to police, even as it keeps asking staff to use those tools.
The cut works out to a 33.3% reduction for the affected engineers, according to eWeek. Reporting says Apple made the change in October. Apple has not publicly confirmed the caps or the reasons behind them, and the reporting describes the limits as aimed at employee usage rather than at Apple’s overall AI investment.
What the Apple AI spending cap covers
The budgets are measured in tokens, the units AI models process as they read prompts and generate output. Coding and debugging work consumes many of them, so engineers hit limits faster than most other staff. Outside engineering, employees already work under a $100 daily cap, and Apple has considered lowering that to $25 for some teams.
Employees who need more can request exemptions from senior leaders. Reporting does not say how many people are affected or how many exemptions have been granted. eWeek’s headline asked whether the change will save money at all, and the answer depends on how many people actually hit the ceiling.
Which teams face the next round of cuts
The table below separates what has been reported as done from what is still under consideration.
| Company or group | Reported limit | Status |
|---|---|---|
| Apple, some engineers | $300 cut to $200 per day | Reported as done in October |
| Apple, some engineering teams | $100 per day | Under consideration |
| Apple, non-engineering staff | $100 per day | In place; $25 for some teams under consideration |
| Goldman Sachs | Coding tools limited to about 12,000 engineers | Reported in October internal guidance |
| Accenture | $29 per month | Reports differ; verify before relying on it |
| Uber | $1,500 per month on individual coding tool spending | Reported |
| Tesla | $200 per week | Reported |
The distinction matters. The $100 level for some engineering teams and the $25 level for other teams are described as plans being weighed, not final policy.
Why Apple is tightening AI budgets now
Apple’s stated context is a push to run a leaner company under CEO John Ternus. Technology.org reports that Apple’s R&D spending rose 32% in fiscal 2026, driven by infrastructure and AI costs. That growth explains why finance teams are scrutinizing every category, including the tokens engineers burn through each day.
Reporting also ties the cut to worries about layoffs after Apple cut more than 200 jobs in August. Some staff fear the caps are a prelude to bigger cuts, but no layoffs have been tied to the caps. Apple has separately warned investors in an SEC filing that AI agents could act without permission, which shows the company sees AI tools as both a productivity gain and a risk to manage.
Inc. framed the cut against Apple’s $114.5 billion in revenue, which makes the trade-off easy to see. A profitable company is still counting every token.
Goldman Sachs and the wider rationing trend
Apple is not alone. In October internal guidance, reported by Bloomberg on November 2, Goldman Sachs limited Anthropic’s Claude Code and GitHub’s coding tools to its roughly 12,000 engineers, according to The Decoder. That pulled access from most traders and bankers. Goldman has not disclosed how many of them lost access.
The reversal stands out. In July, Goldman CIO Marco Argenti praised “vibe coding” by non-technical staff, and Goldman’s own in-house AI tools remained broadly available across the firm. Bloomberg’s reporting, as summarized by Quartz, describes the change as made after a productivity and risk review. A Goldman partner had warned in August that AI could degrade bankers’ analytical skills.
Goldman spokesperson Nick Carcaterra said the firm is limiting the tools to people “who are best suited to leverage them for the long term.” That wording treats access as tied to role rather than as a company-wide default.
Other companies are setting limits too. Accenture reportedly stopped some staff from spending more than $29 a month on AI after employees ran up large bills, according to Yahoo Finance. Reports differ on whether $29 is a ceiling or a per-employee funding amount, so treat that figure as unverified until the company confirms it.
Uber capped individual AI coding tool spending at $1,500 per month, Tesla set a $200 weekly limit, and Meta introduced token budgets, according to Technology.org. Revolut reportedly went the other way on cost. It cut AI coding costs about 90% by running an open model on its own servers after some engineer spend reached $1,000 to $10,000 a month across roughly 2,000 engineers. That account comes from a weekly AI news roundup, so it deserves confirmation from primary reporting.
The current wave follows a period reporting calls “tokenmaxxing,” when some companies encouraged heavy AI spending. Nvidia CEO Jensen Huang has said engineers should spend an amount equal to half their annual salary on AI tokens, a benchmark that sits uneasily beside the caps now spreading through corporate IT.
Markets and the cost question for AI vendors
Cost pressure reaches the vendors too. AI stocks fell worldwide this week on worries that Democratic state and federal wins could slow data-center construction, then partially rebounded Friday, according to an Associated Press report carried by The Washington Times.
Enterprise budget caps make that question sharper. If customers ration usage, AI companies must show growth from a smaller base of heavy users. The trend has raised questions about the planned IPOs of Anthropic and OpenAI. For the data-center side of that debate, see OpenAI’s pledge to pay full data-center power costs.
What workers and IT managers can do now
Employees and managers have concrete steps to take, whether or not their company has announced limits.
- If your employer caps AI usage, ask whether your role qualifies for an exemption and who approves it. Reporting on Apple says requests go to senior leaders.
- Confirm which AI tools are approved for work before you use them, especially coding agents and assistants that can read internal code.
- If your tool shows token or usage totals, check them weekly so a long debugging session does not surprise you at the limit.
- IT managers can set per-user AI budgets, define which roles get coding agents, and track token spending before it climbs.
- Ask whether heavy coding workloads could run on a cheaper or self-hosted model, the route Revolut reportedly took.
What to watch next
Three developments will show whether this is temporary belt-tightening or a lasting change. First, watch whether Apple confirms the caps and turns the $100 and $25 levels into policy, or leaves them as options. Second, watch whether Goldman discloses how many traders and bankers lost access, and whether other banks follow. Third, watch whether Accenture clarifies what its $29 figure means.
If Apple’s exemptions prove common, the caps will look more like a budgeting tool than a hard limit. If exemptions stay rare, expect more large employers to copy the model.
