A cap that arrived a year after the encouragement
The sequence matters. Through 2025 and into 2026, the industry's internal message was to use more AI, measured in tokens. TechCrunch's write-up of the Bloomberg report describes the period as a wave of "tokenmaxxing": some companies encouraged employees to increase AI usage, often measured in tokens, as a proxy for productivity gains. 9to5Mac notes the same period produced internal leaderboards at some companies ranking employees by token usage.
Apple's reported move is the rationing phase of that cycle. In October, according to Bloomberg's sources, the company lowered the daily AI spending cap for some of its engineers from $300 to $200. More reductions are being planned: some engineering teams could go to $100 a day. Outside engineering, many groups already work under a $100 daily cap, and Apple has considered taking some of them to $25. Staff who genuinely need more can ask senior leadership for an exemption, and some teams are being steered toward cheaper models or models that run on the device in front of them.
None of this has been announced by Apple. It comes from one report, Bloomberg's, citing people familiar with private discussions. Apple had not immediately responded to Reuters' request for comment when the report was published. The company had not publicly confirmed the reported changes as of November 4. Read the figures as reported, not confirmed.
The saving is the part nobody has shown
A lower cap is not the same thing as lower spending. Bloomberg's own sources undercut the arithmetic: few Apple workers were using their full $300 daily allowance. If most engineers were spending, say, $60 a day, then a cap at $200 saves nothing on an ordinary day. It only bites on the heaviest days, for the heaviest users, who are also the people most likely to be granted exemptions.
Bloomberg says plainly that it is unclear how much Apple will actually save. No total for Apple's internal AI bill has been reported, no headcount for the affected teams, and no target figure for the savings. Regardless of the immediate financial impact, the cap could change behavior at the margin. Engineers who know that exceeding their allowance requires senior approval may be more inclined to choose cheaper models or run workloads locally. Whether that produces meaningful savings, or reduces productivity, has not been established.
The context also cuts against reading this as Apple retreating from AI. Apple's fiscal 2026 financial statements show R&D spending reached $45.77 billion, up 32% from $34.55 billion the previous year. Apple attributed the increase primarily to higher infrastructure-related costs, including AI investments, and headcount-related expenses. The spending figures cover September 2025 through September 2026, not the period since the reported employee cap reductions. The money is still going in. What is being rationed is one specific line: what individual employees may spend, per day, on other companies' models.
Apple is not the only one counting
Reuters' version of the report places Apple in a wider turn: technology companies that spent a year encouraging widespread AI use are now scrutinizing what it costs. Reuters notes Bloomberg recently reported that Goldman Sachs restricted access to some AI coding tools after a review of productivity and risk.
There is a useful question underneath the trend, and none of the reports answers it: what did the unrestricted year buy? The companies now imposing caps encouraged the usage they are rationing, and the productivity case was asserted far more often than it was measured. A daily allowance is, among other things, an admission that the earlier policy had no budget because it had no measurement. Apple has not said what its engineers produced with $300 days that they will not produce with $200 ones, and until a company publishes that comparison, every cap announcement is a cost story with the benefit side left blank.
Sources
- [1] Bloomberg report, as syndicated by Reuters, November 4, 2026 (read first-hand at Gate 1: last month Apple cut the daily AI spending allowance for some engineers to $200 from $300; further reductions to $100 are under consideration for some engineering teams; some non-engineering groups already operate under a $100 daily cap that could be reduced to as little as $25; employees needing greater access can seek exemptions from senior executives; the budgets cover tokens for models such as ChatGPT, Claude and Gemini; Apple did not immediately respond to a Reuters request for comment; Bloomberg recently reported that Goldman Sachs restricted access to some AI coding tools following a review of productivity and risk management. All cap figures originate with Bloomberg, citing people familiar with the matter)Read source
- [2] TechCrunch, November 4, 2026 (read first-hand at Gate 1: relays the Bloomberg report; describes the earlier period as a wave of "tokenmaxxing"; TechCrunch has reached out to Apple for comment)Read source
- [3] MacRumors, November 4, 2026 (read first-hand at Gate 1: relays Bloomberg; some engineers limited to $200 per day in October, down from $300; some teams asked to use less expensive models or on-device AI; some employees told Bloomberg few were using their full $300 daily limit, so the savings may be small; CEO John Ternus took over on September 1 — separate organizational context)Read source
- [4] 9to5Mac, November 4, 2026 (read first-hand at Gate 1: direct Bloomberg quotations — the October cut for some engineers, further cuts planned potentially to as little as $100, the existing $100 non-engineering cap with $25 considered for some teams, exemptions from senior leadership, sources anonymous because the discussions are private; "few workers were using their full $300 daily allowances"; Bloomberg notes it remains unclear how much Apple will actually save)Read source
- [5] Apple Inc. — fiscal 2026 Form 10-K, filed on EDGAR, accession 0000320193-26-000026 (read first-hand at Gate 1: research and development expense was $45,765 million in fiscal 2026, up 32% from $34,550 million in fiscal 2025; the increase was primarily due to higher infrastructure-related costs, including investments in artificial intelligence, and headcount-related expenses; fiscal year ended September 26, 2026)Read source