Microsoft Q4 2026: $90B Revenue, Azure Tops $100B, Copilot at 30M Seats

Microsoft reported $90 billion in revenue for the quarter ended June 30, 2026, an 18% jump that pushed its stock up more than 7% in after-hours trading. Azure grew 43% and crossed $100 billion in annual revenue for the first time. Microsoft 365 Copilot now has 30 million paid seats, up 50% quarter-over-quarter, though that figure represents roughly 7% of the estimated 450 million commercial Microsoft 365 customers. Capex came in at $41 billion, up 70% from a year earlier, but strong operating cash flow of $55.44 billion kept investors calm.

What happened

Metric	Result
Q4 FY2026 revenue	$90B (+18% YoY)
Microsoft Cloud revenue	$59.3B (+27% YoY)
Azure revenue growth	+43% YoY
Azure full-year revenue	First time above $100B
M365 Copilot paid seats	30M+ (+50% vs prior quarter)
Operating income (Q4)	$40.6B (+18%)
Net income (Q4)	$35.8B (+31%)
Diluted EPS (Q4)	$4.81 (+32%)
Q4 capex	$41B (+70% YoY)
Q4 operating cash flow	$55.44B (+30% YoY)
Full-year revenue (FY2026)	$331.8B (+18%)
Full-year net income	$133.7B (+31%)
Commercial remaining performance obligation	$678B (+84%)
Anthropic investment gain (Q4)	~$3.2B

Microsoft CEO Satya Nadella pointed to Azure crossing $100 billion in annual revenue and Copilot reaching 30 million paid seats as proof that enterprise customers are committing real money to AI. CFO Amy Hood noted that commercial remaining performance obligations (contracted future revenue not yet recognized) grew 84% to $678 billion, suggesting the pipeline is growing faster than current capacity allows.

Hood also addressed the $41 billion capex number directly. She said about two-thirds of that spending went to short-lived assets like CPUs and GPUs, and that finance leases, which let the company invest in datacenters without full immediate payment, accounted for around $5.6 billion of the total.

Is Microsoft 365 Copilot actually gaining traction?

Thirty million paid Copilot seats sounds large, but context matters. Microsoft’s estimated commercial M365 customer base is around 450 million. That puts Copilot penetration at roughly 7%. The 50% quarter-over-quarter growth is real, but the base is still small relative to the overall install base. Microsoft also charges usage-based fees on top of existing per-seat licenses, which means the revenue per active user is higher than a simple seat count implies, but also that cost-conscious buyers have extra reason to hold back.

For businesses weighing whether to roll out AI integration across their Microsoft stack, the low adoption rate suggests most organizations are still in evaluation or pilot mode, not full deployment.

Why it matters

Azure’s 43% growth and a $678 billion contracted backlog are the two numbers that matter most for the investment thesis around Microsoft’s AI spending. Emarketer analyst Gadjo Sevilla, speaking to The Register, put it plainly: Microsoft spent $35.80 billion on property and equipment in the quarter, more than double the $17.08 billion in the same period a year ago. Full-year capex hit $115.95 billion, up nearly 80% from $64.55 billion in FY2025. Despite that pace, operating cash flow still grew 30%, which is why investors rewarded the stock rather than punishing it.

The concern that AI spending could outrun returns remains real, however. Fitch Ratings warned on Monday that a potential AI market correction has become a major credit risk. The agency said that “revenue uncertainty and the extent to which capital markets and economies have become intertwined with AI have created a vulnerability for credit in the event of a re-evaluation of long-run returns potential.” A sharp, prolonged correction, Fitch added, could have wider market and macroeconomic effects depending on scale and duration.

This echoes broader anxiety after Meta’s recent results disappointed on AI monetization. Microsoft’s numbers look cleaner for now, partly because Azure cloud growth gives it a direct revenue line tied to AI compute demand, rather than relying solely on ad revenue uplift from AI features.

Our take

The Azure story is the strongest part of this report. A business crossing $100 billion in annual revenue with 43% growth is not a speculative bet; it is a functioning, scaling product. The capex picture is more complicated. Spending that grows 80% year-over-year is only sustainable if demand keeps pace, and the $678 billion backlog is the one number that gives investors confidence it will.

Copilot’s 30 million seats are real but modest. Microsoft is charging a premium (seat cost plus usage fees) and getting 7% penetration of its own captive user base after years of pushing the product. That is not a failure, but it is not the landslide the marketing suggests. Businesses evaluating Copilot should look carefully at actual usage data inside their organizations before committing to full rollout at those prices.

The Fitch warning is worth taking seriously. We have covered similar investor anxiety around Google’s $205B capex plans, and the pattern is consistent: infrastructure spending is accelerating faster than revenue from AI-specific products can justify on its own. Right now cloud compute demand is carrying the load. If that growth rate softens, the math gets harder quickly.

What to do about it

1. Audit your actual Copilot or Azure AI usage before your next renewal. Low utilization at premium pricing is a common finding in early enterprise AI deployments.
2. Compare usage-based costs against your expected workload volume. Microsoft’s dual billing model (seat plus consumption) can push real costs well above the headline per-seat price.
3. If you are in early AI evaluation, consider a time-boxed pilot with defined success metrics before expanding seats or committing to long-term Azure AI contracts.
4. Watch Fitch’s credit signals alongside earnings. If AI revenue growth at the hyperscalers slows for two consecutive quarters, pricing and availability in the SMB tier often follows within 12 months.

The bottom line: Microsoft’s cloud business is genuinely strong, but Copilot still needs to prove it can move beyond early adopters and justify its price premium for the average enterprise team.

Source: The Register · AI/ML