Microsoft is reportedly keeping a closer watch on employee use of artificial intelligence tools after internal data showed significant differences in monthly AI consumption across departments.
The figures were reportedly collected through a voluntary spreadsheet in which employees share information including salaries, bonuses and AI usage. According to reports, around 600 US-based employees contributed to the spreadsheet, with approximately 350 providing details about their AI consumption.
The median reported monthly AI usage was around $300, although spending varied considerably between teams. Employees in Microsoft’s CoreAI division recorded a median usage of about $975, followed by Security at $526, Microsoft AI at $490 and Cloud and AI at $325.
Other reported median figures included approximately $250 for Experiences and Devices, $241 for Azure and $134 for Customer and Partner Solutions.
Individual usage was considerably higher in some cases. One employee in Customer and Partner Solutions reportedly recorded around $28,000 worth of AI tokens over a 28-day period. Other reported high-use figures included approximately $16,000 in CoreAI, $15,000 in Cloud and AI, and $10,000 in Security.
The $28,000 figure could represent an exceptional case or potentially an error, but the data has highlighted the growing cost associated with intensive AI usage inside large technology companies.
Microsoft is reportedly monitoring token consumption more closely to identify unusually high usage. The company had previously encouraged employees to experiment with AI tools, but concerns have emerged around excessive consumption that does not necessarily translate into measurable business value.
The trend has also been described as “tokenmaxxing”, referring to employees intentionally using large quantities of AI tokens to increase their visibility on internal usage dashboards. Similar behaviour has reportedly emerged at other technology companies.
Microsoft leadership has emphasized that employees should focus on using AI to deliver meaningful results for customers and the business rather than simply maximizing token consumption.
The increased monitoring reflects a broader challenge facing technology companies as AI becomes more deeply integrated into everyday work: encouraging experimentation and productivity while keeping rapidly increasing AI infrastructure and usage costs under control.
