Microsoft Employees Self-Report AI Spending, and It Has Nothing to Do With Bigger Raises

Microsoft employees have started tracking something new alongside their salaries: how much AI they use.

An internal spreadsheet reviewed by Business Insider shows nearly 600 entries from US employees this year, with workers voluntarily and anonymously sharing their base pay, raises, bonuses, and stock awards. New to this year's document is a column showing dollar amounts of AI usage, pulled from an internal Microsoft tool that lets employees track their own AI spending over a rolling 28-day window.

A Microsoft spokesperson told Business Insider the tracking tool is still in early testing.

The numbers vary wildly

About 350 of the nearly 600 entries included AI usage figures. The median came out to roughly $300 over 28 days, according to Business Insider's analysis.

The range was enormous. In the Azure organization, usage spanned from $1 to $7,500. In Cloud + AI, the spread ran from $1 to $15,000. The single highest figure in the entire spreadsheet, $28,000 in a 28-day span, came from someone in Microsoft's Customer and Partner Solutions organization, where the median was just $134.

CoreAI had the highest median usage at $975, followed by Security at $526 and Microsoft AI at $490. Experiences and Devices, the org behind consumer products like Windows and Surface, had a median of $250.

Business Insider was upfront that this is a small, self-selected sample. Microsoft employed 223,000 people as of June 30. A spreadsheet with 350 AI-usage entries represents a sliver of that workforce, and the data is self-reported, meaning it could contain errors or reflect selection bias, since employees who volunteer pay and usage data may not be representative of the company as a whole.

No payoff, at least not yet

The more striking finding is what Business Insider says it didn't find: any meaningful correlation between AI usage and reward. Employees burning through thousands of dollars a month in AI tools didn't show higher bonus ratings, bigger merit increases, or better odds of promotion than employees barely using the tools at all.

Given the scale of Microsoft's AI bet, this finding stands out. The company has been spending hundreds of billions of dollars on AI infrastructure and has been pushing employees company-wide to fold AI into their daily workflows. If that push were translating into measurably better individual performance, you'd expect it to show up in compensation data somewhere. In this dataset, it doesn't.

The fair read for AI skeptics, and its limits

A reasonable skeptic could look at this and say the AI investment isn't paying off for workers, and maybe not for the company either, if heavy users aren't producing outcomes that get noticed and rewarded. That's a legitimate question worth asking given how much shareholder money is riding on the AI buildout across Microsoft, Meta, Google, OpenAI, and Anthropic.

But the data has real limits. A 28-day usage snapshot doesn't capture long-term productivity effects, and performance reviews at a company the size of Microsoft run on cycles and criteria that have nothing to do with a self-tracked spending tool still in early testing. It's also possible high-usage employees are in roles where AI use is expected as baseline behavior rather than a differentiator, which would flatten any correlation with rewards. Business Insider's own analysis excluded levels and teams with too few entries specifically because the sample gets unreliable at that granularity.

What this actually is

This isn't a scientific study. It's a voluntary pay-transparency spreadsheet that Microsoft employees have run informally for years, now with an AI column bolted on. Nobody at Microsoft required this data collection, and the company isn't vouching for its accuracy.

Still, it's the closest thing to a real internal look at AI adoption gaps inside a trillion-dollar tech company that's betting its future on the technology. The Customer and Partner Solutions org shows the widest spread, from $1 users to a single $28,000 outlier, suggesting either wildly inconsistent adoption or a handful of employees running experiments or side projects that don't reflect typical use.

Microsoft hasn't said whether it plans to use this usage-tracking tool for performance evaluation once it moves past testing. Whether heavy AI use eventually becomes a factor in raises and promotions, or whether the company decides usage volume isn't a useful signal at all, remains an open question with no timeline attached.