Apple Lowers Daily AI Token Allowances for Engineers to $200

Apple has cut the internal daily spending caps on artificial intelligence models for some software engineers from $300 to $200, with plans to introduce steeper restrictions across the company, according to Bloomberg reporting detailed by The Next Web (https://thenextweb.com/news/apple-cuts-employee-ai-token-budgets).

The daily limits govern token usage across third-party models available to staff, including Anthropic's Claude, OpenAI's ChatGPT, and Google's Gemini. Token processing costs mount quickly during software drafting and code debugging. Under the new policy, Apple is considering further reductions that could lower allowances for some engineering groups to $100 per day. Non-engineering employees, who largely operate under an existing $100 daily limit, could see caps drop to $25. Staff requiring higher volume can request formal exemptions from senior management. Apple declined to comment on the changes.

Alongside the lower caps, some engineering divisions have been instructed to favor cheaper models or on-device alternatives when possible. The policy has drawn mixed internal responses: while certain staff members noted that typical daily usage rarely approached the earlier $300 ceiling, others view the limits as part of a wider belt-tightening push following the dismissal of more than 200 workers in August.

The budget enforcement follows the appointment of John Ternus as chief executive on Sept. 1. Ternus has flattened management structures, accelerated operational tempos, and brought in third-party management consultants to audit workflow efficiency across selected business units.

The individual spending limits contrast with Apple's surging foundational outlays. In its annual regulatory filing, Apple disclosed that research and development spending climbed 32% during its 2026 fiscal year, propelled by higher infrastructure and AI development costs. Total operating expenses rose 22% to $75.5 billion over the same period.

Enterprise token caps are becoming standard across major technology companies confronting escalating developer compute overhead. Microsoft instituted division-level AI budgets in July, Tesla caps employee usage at $200 per week, Uber limits software engineering AI tools to $1,500 per month, and Accenture has capped Claude access at $29 monthly.