AI for everyone, but at what cost? Why Apple is cutting employee budgets despite encouraging wider use

Apple is considering further cuts to employee AI spending limits as it seeks to rein in costs. The move reflects a wider trend among tech giants, including Microsoft, Tesla and Uber, to tighten budgets despite pushing greater AI adoption.

After initially encouraging employees to experiment with artificial intelligence (AI), technology companies are increasingly tightening spending limits on its use, as the costs of widespread adoption come under greater scrutiny.

Apple is the latest to move in this direction. According to a Bloomberg report, the company is planning to restrict employees' AI usage as part of broader cost-cutting efforts, even as it seeks to make the technology central to its operations.

The iPhone maker has already reduced the daily AI spending allowance for some engineering teams from $300 to $200 and is considering further cuts that could bring the limit down to as little as $100.

$100 daily cap for some, $25 for others

While Apple is pushing to modernise its engineering processes and internal operations using AI, it is also looking to control how much employees spend accessing the technology.

Engineering teams are the latest to face spending restrictions, with several other teams already operating under a daily cap of $100. The company is now considering reducing this allowance to $25 for some teams. Employees who require higher budgets would have to seek exemptions from senior leadership, Bloomberg reported.

Employees fear potential restructuring

Apple employees have access to several AI models, including OpenAI's ChatGPT, Google's Gemini and Anthropic's Claude.

AI token budgets determine how much text and other data a model can process when responding to requests. AI providers generally charge based on the volume of tokens consumed, with more advanced models often costing more. Resource-intensive tasks, such as writing or debugging software code, can consume substantial quantities of tokens, increasing the overall cost.

Prabhakar Jha