The headline: strong growth, short supply
Microsoft's fiscal 2027 first-quarter results covered the quarter ended September 30, 2026. Revenue was $92.3 billion, up 19%. Microsoft Cloud revenue was $63.9 billion, up 30%, or 31% in constant currency. Azure revenue rose 48%. CFO Amy Hood said Azure came in ahead of expectations "even with the 3P capacity delays." She credited Foundry and better monetization of CPU capacity.
The delay is the less-noticed part. Hood said capital spending of $45.2 billion came in as it did "as some third-party capacity deliveries we expected in Q1 were delayed." For the next quarter she said some third-party "GPU as a service" contracts had slipped to the second half of the fiscal year. She also said customer demand continues to exceed available supply.
Microsoft did not name the suppliers. It did not say how much capacity is delayed or which customers or workloads are affected.
Reported results versus forecasts
Mixing these up is the easiest mistake to make with this story.
| Item | Type | Figure |
|---|---|---|
| Azure revenue growth, Q1 FY27 | Reported | 48% year over year |
| Azure revenue growth, Q2 FY27 | Forecast | 49% to 50% in constant currency |
| Capex, Q1 FY27 | Reported | $45.2 billion |
| Capex, Q2 FY27 | Forecast | Over $50 billion |
| Capex, calendar 2026 | Forecast | About $170 billion |
Hood tied the roughly $170 billion calendar-year capex forecast to the third-party delays. She also expects spending to accelerate in the second half of the fiscal year. She named CPUs, which include higher memory and storage costs, GPUs, and the third-party deliveries that shifted into that period.
What Microsoft says it is doing about it
CEO Satya Nadella described the infrastructure as a "fungible fleet." In his description, one pool of capacity can be allocated to the highest-value work across compute, data, Foundry and Copilot. He gave these operating measures:
- Data centers: Microsoft added 24 across five continents in the quarter. The company did not give locations or megawatt figures.
- Dock-to-live time: For new GB300 systems in its biggest regions, the time to bring them online fell about 20% quarter over quarter. Those regions hold most of Microsoft's GPU deployments. This measures speed to deployment, not a 20% jump in total capacity. Microsoft gave no baseline duration.
- Copilot throughput: Nadella said it rose more than 50% quarter over quarter. That is not the same as more Copilot users or more revenue, and Microsoft did not update seat counts here.
Efficiency is the other lever. Hood said engineering work across the GPU and CPU fleet let Microsoft put more capacity toward Foundry. She said Foundry revenue more than tripled year over year. She also said the work found underused CPU capacity that could be monetized. She was open that these gains are hard to predict. In her words, "we need to find some more" to hit the Q2 guidance, and some supply Microsoft was counting on for Q2 moved into the second half.
Capex accounting
Two figures are easy to confuse.
- Capex was $45.2 billion. Roughly three-quarters went to short-lived assets, mostly CPUs and GPUs.
- Cash paid for property and equipment was $39.4 billion.
Finance leases were $7.4 billion. Microsoft books the full principal value of a finance lease as capex when the lease begins. So lease timing can move reported capex around, and delayed third-party deliveries can push spending into later quarters.
Free cash flow was $18.6 billion. Hood said it reflected higher capital expenditures. Microsoft also said it expects to stay free-cash-flow positive in FY27.
Why this matters to Windows and Azure admins
Capacity has been a pattern, not a one-off. Earlier calls were similar. On the fiscal 2026 first-quarter call, Microsoft said it expected to be capacity constrained through at least the end of that fiscal year. Futurum's analysis of the fiscal Q3 2026 results also described capacity as the primary limiter on Azure growth. The October call suggests the squeeze has not eased, even with faster deployment.
Regional shortages are real, though Microsoft did not link them to this call. A third-party guide from Trusted Tech Team describes a late July 2025 demand spike in East US that left customers hitting AllocationFailed errors for over a week. That is a secondary source, and Microsoft did not tie any specific region to the delays it discussed on the call. It is still a reminder of how capacity limits show up for admins.
Practical steps, drawn from general industry practice rather than Microsoft's statements:
- Reserve or secure capacity early for GPU or large-VM projects, rather than assuming on-demand availability.
- Qualify more than one region and more than one VM size for critical workloads.
- Treat any GPU-as-a-service timeline that depends on third-party capacity as likely to slip.
- Watch Azure Service Health for capacity-related notices.
Balanced view
The optimistic reading is that Azure grew 48% despite the delays. Efficiency work is adding usable capacity, and Q2 guidance is higher still. Hood also said each dollar of capex can support more revenue over time.
The cautious reading is that the Q2 forecast partly depends on efficiency gains Microsoft itself calls unpredictable. Delayed contracts are pushing spending into the second half. The call did not size the shortfall or promise when the delayed capacity will arrive. Microsoft's filings carry the usual risk language about its data centers depending on land, power, networking and server supply. That is general disclosure, not proof of what caused these particular delays.
Bottom line
Microsoft is selling capacity as fast as it can get it. The limit is delivery speed, power and hardware, not demand. For enterprise IT, the practical lesson is to plan for tight Azure GPU supply into the second half of fiscal 2027. Microsoft's own figures show it is building faster, but not yet fast enough to say demand is met.