Some Apple engineers have seen their allowance for third-party artificial intelligence tools fall from $300 to $200. Others now have $100, or as little as $25, for their projects.
Bloomberg reported the reductions, which affect spending on external AI services used for work such as coding assistance and finding bugs. This is not a disclosed cut to Apple’s overall research budget or a stated retreat from developing its own models.
The distinction matters for employees. An allowance pays for practical access to outside tools; it does not, by itself, measure what Apple spends on internal artificial intelligence research, infrastructure or products.
A limit many engineers did not reach
The report also contains an inconvenient detail for any simple cost-cutting narrative: many engineers said Apple employees were not coming close to the existing $300 limit. Reducing an unused allocation need not change their daily work at all.
That makes a bureaucratic explanation plausible. Budgets assigned but never spent can be released for other purposes without removing a tool someone actively uses. The report nevertheless describes confusion surrounding the decision and mentions concern about possible layoffs. It supplies no confirmation that layoffs are connected.
Employees can also request a higher allowance on a per-use basis. The smaller default figure therefore does not necessarily impose an absolute ceiling on a project with a demonstrated need.
Where the tools fit
AI coding assistants can accelerate repetitive work and search for defects across large bodies of code. Their usefulness is not uniform. Experienced developers still have to assess the result, particularly when the software concerned is a familiar Apple product rather than an experiment generated from a brief prompt.
Apple has alternatives to buying every interaction from an external provider. It is developing systems around its own Apple Foundation Models. Its Mac computers can also run some models locally, reducing the need to purchase additional tokens for suitable tasks.
Neither option proves that local processing is cheaper for every workload, nor does the budget report establish a company-wide switch in technical strategy. It does show why the cost of third-party access has become a management question rather than an invisible engineering expense.
The reported reductions arrive as companies examine what they gained from early AI experimentation. In Apple’s case, the concrete change is narrower than an AI retreat: a lower standard allowance for outside services, different limits for different engineers, and a route to request more money when an individual use justifies it.
Sources
- Apple employee AI budgets cut to reduce costs
- Apple
- Apple Maps starts blurring US military bases
- AirPods Pro 3 Are Cheaper on Amazon Than at Apple, With a Bigger Discount and 3 Months of Free Apple Perks
- Apple Will Check Public iCloud Links for Harmful Content Starting in iOS 27.3
- Apple beats Q4 estimates on strong iPhone sales but misses on China revenue, stock falls
Photo: ThisisEngineering / Unsplash (Unsplash License)
