Satya Nadella has put a number on Copilot's momentum at every earnings call for two years. What made Wednesday's different was what he said about it on the call:
"The number of conversations per user nearly doubled year-over-year. Average weekly engagement is on par with Outlook and Teams."
Seat counts and daily-user tallies describe how many licences have shipped. Weekly engagement on par with Outlook and Teams describes something closer to habit, Copilot getting opened as often as the two applications every knowledge worker already treats as switched-on-all-day infrastructure. Nadella came back to the same point later on the call, when an analyst asked him directly how usage had changed since Copilot's early pilots. He said:
"What used to be months is days from when a license is bought."
- Microsoft Overhauls Copilot Leadership to Drive Enterprise Adoption and Accelerate Agentic AI in M365
- Microsoft Earnings Analysis: Copilot Takes Off but AI Bills Keep Rising
- Microsoft 365 Copilot Passes 30 Million Paid Seats as Cloud and AI Growth Power Record Quarter
The Number Microsoft Was Trying to Outrun
That comparison would have looked like wishful thinking six months ago. On the January earnings call, Nadella told investors that daily users of the Copilot app had increased nearly three times year-over-year. Weeks later, Sensor Tower data reported by CNBC put a harder number on that same product: around six million daily active users in February 2026, against 440 million for ChatGPT and 82 million for Google Gemini. Only around 3% of commercial Microsoft 365 subscribers had bought the enterprise Copilot add-on at all.
That gap, between the growth rate Nadella was describing and the scale Sensor Tower was measuring, sat behind the leadership reorganisation Microsoft announced in March. Jacob Andreou, previously Microsoft AI's product and growth chief, was installed as EVP of Copilot, taking over consumer and commercial product lines that had until then been run separately. Mustafa Suleyman, who had led Copilot before, moved to focus on frontier model development instead. In his memo to staff, he explained the change:
