Microsoft Limits Internal AI Spending After Worker Uses $28,000

Microsoft is reevaluating how its employees use company resources after a single worker managed to spend $28,000 on AI tokens in just twenty-eight days. While the software giant has spent years encouraging its staff to adopt new tools, this massive bill has forced it to look closer at usage habits. The situation highlights a growing trend where some workers consume vast amounts of computing power without delivering clear business results.

The company monitors internal token consumption to stop unnecessary expenses

The eye-catching figure comes from an internal spreadsheet where workers voluntarily share their compensation and spending data. Out of the hundreds of employees who logged their usage, the median cost was roughly $300 a month. Some departments saw higher bills, with workers in security and cloud divisions routinely spending thousands of dollars. Since the company employs over 223,000 people globally, these numbers only offer a small glimpse into the overall cost of running these internal models.

The massive usage spikes are tied to a practice known internally as tokenmaxxing. This happens when employees intentionally use large amounts of artificial intelligence resources to appear highly active on company tracking dashboards. It became a strange status symbol among engineers trying to show how much they engage with the new systems.

In response to the growing bills, leadership is stepping in to change the culture. An internal memo from an executive recently told staff that optimizing for high token usage is not the goal. Instead, the company wants its teams to focus on generating real value for customers.

Going forward, it plans to track token consumption more carefully to ensure the spending actually leads to meaningful outcomes rather than just padded stats. By tying usage directly to results, the tech giant hopes to keep its computing costs reasonable while still encouraging innovation.