Earnings Recap — Q4 FY2026
CY Q3 2026 · Reported July 29, 2026 · Beat 7 of last 7 quarters
The Buildout is a website built on agentic AI tools, currently in beta, and could have factual errors.
Microsoft's continued capacity expansion and efficiency gains underscore the sustained demand for AI infrastructure, with Azure growth accelerating to 43% and CapEx expected to grow again in FY27. The shift to usage-based pricing across Copilot and GitHub signals a maturing AI monetization model, while the extension of data center useful life to 25 years will impact depreciation and lease classification, potentially affecting industry benchmarks.
Microsoft reported Q4 FY26 revenue of $90.0B, up 18% YoY, with operating income up 18% and EPS of $4.74 (up 23% adjusted for OpenAI investment). Azure growth accelerated to 43%, driven by efficiency gains and faster capacity deployment, while commercial RPO grew 84% to $678B. Microsoft Cloud revenue was $59.3B, up 27%, and the company added 31 data centers in the quarter, bringing the annual total to 88. M365 Copilot paid seats surpassed 30 million, and GitHub Copilot reached 50 million users, with usage-based billing driving consumption revenue.
Management guided Q1 FY27 revenue of $89.85B–$90.95B (16–17% growth) and Azure growth of ~45% in constant currency, with H1 growth expected to accelerate. Full-year FY27: double-digit revenue and operating income growth, operating margins down less than 1 point, and CapEx growing YoY to ~$175B (including lease reclassification impact from extending data center useful life from 15 to 25 years). They expect M365 Commercial cloud revenue growth to accelerate through the year on premium SKU momentum and usage-based billing, while Windows OEM/Devices revenue is expected to decline in the high teens due to PC market weakness. Demand continues to exceed supply, and management emphasized efficiency gains and capacity expansion as key levers.
“All up, we added another gigawatt of capacity this quarter and remain on track to roughly double our overall capacity in just 2 years.”
on Capacity expansion
“Customer demand continues to exceed available capacity.”
on Azure demand
“We are very, very clear about the architectural sort of design of the platform, which is you've got to keep your harness separate from the model. When the harness will ensure that your memory, your context, all of that is external. That means any given model at any given time is swappable.”
on Model choice and enterprise IP
How material could traction be for open and custom models over the next year or two, and how does Microsoft benefit given its frontier lab exposure?
Satya emphasized that enterprises want to control their own destiny and keep their harness separate from models, making models swappable. Microsoft benefits by providing the platform (Azure) that delivers the right model for the right job, regardless of model choice.
Are we still in the same capacity-constrained environment, or is Microsoft just executing better?
Amy confirmed demand still exceeds supply, but highlighted efficiency gains in CPU/GPU fleets and process improvements that allowed faster monetization of new capacity in the quarter.
How does Microsoft protect itself if there is overcapacity, and how do you manage through hardware price increases?
Amy explained that a large portion of CapEx is short-lived assets (CPUs/GPUs) with flexible timing, and the diverse customer base provides flexibility. On pricing, they focus on efficiency gains and cloud ROI to offset component price increases.