Microsoft hits 30M Copilot seats: measure outcome per seat

Microsoft closed its fiscal year today with a number designed to end an argument: Microsoft 365 Copilot now has over 30 million paid seats. Azure crossed $100 billion in annual revenue for the first time. Total FY26 revenue hit $331.8 billion, up 18%. The "is enterprise AI real" debate is over on Microsoft's side of the ledger. The question that's still open — and the one nobody bills you for — is whether the seats you pay for produce anything.

What actually happened

Per Microsoft's Q4 FY26 release, the quarter ended June 30 brought $90.0 billion in revenue (up 18%), $40.6 billion in operating income (up 18%), and Microsoft Cloud revenue of $59.3 billion (up 27%). Azure grew 43%. For the full fiscal year: $331.8 billion revenue, $155.2 billion operating income, and Azure past $100 billion annually.

The Copilot seat count is the line that matters for operators. Thirty million-plus paid seats is up from the roughly 20 million Microsoft reported on its April call, per CNBC — a 50% jump in one quarter. Satya Nadella's framing was explicit about where the pitch is headed: "We are advancing the frontier on the cost-to-outcome curve, ensuring every customer can turn tokens into business results." CFO Amy Hood called it "a strong quarter to close out the fiscal year."

Note what Nadella did not say. Not tokens per dollar. Not benchmark scores. Cost to outcome. Microsoft is selling on the same axis we tell clients to buy on — which is convenient, because they get to define the outcome.

Why the Copilot seat count matters for your business

Thirty million seats means Copilot is now the default line item in the Microsoft 365 renewal your ops manager signs without reading. That's the real mechanism here: AI stops being a decision and becomes an assumption baked into a bundle. Somebody at your company will forward you this headline as proof you're behind.

You're not behind. You're under-measured. The number to track isn't seats purchased — it's outcome per seat, and it's embarrassingly easy to get: pick the three workflows the tool is supposed to compress, time them before rollout, time them ninety days later, and divide the annual license cost by hours actually returned. We've run this with clients and the result is bimodal. A handful of people use it constantly and get real time back. Most seats go dark inside a month and quietly renew forever.

Buy Copilot if the math works — Microsoft's distribution advantage is real and fighting it is usually a waste of your energy. But buy the seats people use, not the seat count that makes the headline true for you too. And keep the workflows that actually matter to your revenue — quoting, fulfillment, support triage — in systems you control, not in a productivity add-on whose pricing and model selection change without your input.

Sources: Microsoft, CNBC.