Related ToolsChatgpt

Microsoft Reports 20M+ Paid Copilot Users, Pushes Back on Low-Adoption Narrative

Microsoft Copilot
Image: Microsoft

20 million. That's the paid user count Microsoft is now citing for Copilot, the AI assistant baked into Microsoft 365 products like Word, Excel, and Teams. The company disclosed the figure on April 29, 2026, alongside claims that engagement - not just seat count - is rising.

The announcement is notable because Copilot has carried a stubborn reputation problem: enterprise IT departments buying licenses that employees ignore. Surveys from Gartner and others over the past year consistently found low active usage rates, with workers either unaware the tool was available or finding it not worth the workflow change. Microsoft is now pushing back on that narrative with actual numbers.

What the company didn't fully detail is how it defines "paid user" - whether that means anyone with a license attached to their account, or people who actively open the product on a regular basis. That distinction matters enormously. A company buying 10,000 Microsoft 365 Copilot seats at $30 per user per month counts as 10,000 paid users whether anyone opens it or not. Microsoft said engagement is growing, but stopped short of sharing daily or monthly active user figures.

Still, 20 million paid users is a real number. For context, Microsoft 365 has roughly 400 million paid seats globally. That puts Copilot penetration at around 5% of the installed base - modest, but not negligible for a product that costs an extra $30/month on top of an existing subscription and requires behavior change from people who've worked a certain way for decades.

The pressure on Microsoft to show results here is real. Investors have watched the company pour billions into AI infrastructure and OpenAI partnership costs. A credible user number helps the story, even if the engagement data behind it remains fuzzy. For businesses weighing whether to roll out ChatGPT or Copilot across their teams, the fact that 20 million users are paying for the latter suggests the product has at least cleared the threshold of "people will pay for this" - which, a year ago, wasn't obvious.