Microsoft Azure 48% Growth as AI Spending Climbs
Key Takeaways
- Microsoft reported first-quarter fiscal 2027 revenue of $92.3 billion, up 19% year over year, for the quarter to September 30, 2026.
- Microsoft Cloud revenue rose 30% to $63.9 billion, with Azure up 48% and guided to 49–50% growth in the second quarter.
- Capital expenditure was $45.2 billion in the quarter, up 30%, and is guided to exceed $50 billion in the second quarter, taking calendar-year spend to around $170 billion.
- Free cash flow fell 27% to $18.6 billion, as AI infrastructure spending outpaced cash generation in the quarter.
- Copilot daily users rose nearly 50% quarter on quarter, but Microsoft did not update its paid seat count this quarter.
Microsoft’s latest results tell a clear story about the economics of AI at scale: demand is strong, growth is fast, and the bill for building the capacity to meet it is enormous. For the first quarter of fiscal 2027, the quarter to September 30, 2026, reported on October 28, 2026, Microsoft reported revenue of $92.3 billion, up 19% year over year. Microsoft Cloud revenue rose 30% to $63.9 billion, with Azure up 48% — the headline figure that captures how central cloud and AI infrastructure have become to the company’s growth.
The results were also the first under a new reporting structure announced in September, with three previous segments recast into two. “Agents and Infra” brings Azure, Microsoft 365, Dynamics and server licensing together and grew 25% to $76.8 billion, while “Devices and Consumer” covers Windows OEM, Xbox and search advertising and fell 3% to $15.5 billion as Windows OEM and devices revenue dropped 22%. Readers comparing these figures with older reports should note the recasting, rather than treating the segment totals as directly comparable with the old structure.
Azure, Copilot and the shape of demand
Azure’s 48% growth, with guidance of 49–50% in the second quarter, came despite some third-party capacity deliveries slipping into the second half. Management said customer demand still exceeds available supply, and Microsoft added 24 data centres across five continents in the quarter. That supply constraint is a useful reality check on AI narratives: growth is not limited only by customer appetite, but also by how fast physical infrastructure can be built and powered.
Copilot’s trajectory is more mixed. Daily users rose nearly 50% quarter on quarter, and the number of customers with more than 50,000 seats grew more than eight times year over year, with HMRC named among them alongside Capgemini, HCLTech and Hitachi. Microsoft 365 commercial seat growth, however, slowed to 5%, the lowest in five quarters, which management partly blamed on weaker sales of E3, its mid-tier suite, and standalone licences. With seat growth slowing, more of Copilot’s growth will need to come from consumption — a shift reflected in last month’s move of Copilot’s agentic features, such as Code and Autopilot, onto usage-based Copilot Credits. Customers consuming credits grew more than 75% quarter on quarter, but management conceded that consumption outside coding will build more slowly.
One figure Microsoft did not update is important for honest reporting. The company did not give a new paid seat count this quarter. The last figure widely cited, more than 30 million paid seats, was a July figure, not a current-quarter figure, and it should not be presented as the latest total. The daily-user and large-customer growth figures are the current-quarter evidence; the seat count is not.
The spending side: capex, cash flow and funding
Capital expenditure was $45.2 billion in the quarter, up 30%, and will exceed $50 billion in the second quarter, taking calendar-year spend to around $170 billion. Free cash flow fell 27% to $18.6 billion, and chief financial officer Amy Hood said Microsoft will look at additional funding sources, having not yet tapped the debt markets. Headcount is down 2% year over year, with second-quarter cost guidance including an estimate for the voluntary exit offer given to around 60,000 staff.
Those numbers frame the central tension in Microsoft’s AI strategy. The company is spending at a pace that depresses free cash flow in the short term, on the expectation that AI demand will justify the capacity over time. That expectation may prove correct, but it is not guaranteed, and the funding comment shows management is already thinking about how to finance the build if cash generation does not keep pace. This article is not investment advice, and it does not predict Microsoft’s share price or the return on its AI spending.
Microsoft also pointed to enterprise breadth as a strategic choice. Chief executive Satya Nadella pointed to Rolls-Royce running its engine maintenance supply chain on Foundry agents as an example, and said he would rather build a broad customer base than rely on a few customers buying large amounts of capacity. Nationwide Building Society is among the early customers testing Perception, Microsoft’s security tool for AI agents, which enforces identity, permissions and monitoring outside the model itself. Atos is also among more than 4,000 customers to have used Decision-1, a new Microsoft model for structured decision-making, which Microsoft claims beats large language models on speed and quality — a company claim, not an independent finding.
What to watch next
Three questions will shape how this quarter is judged. First, can Azure sustain growth near 50% as capacity comes online, or will third-party delivery slips and power constraints slow it? Second, can Copilot consumption outside coding build fast enough to offset slowing seat growth? Third, how will Microsoft fund a calendar-year spend of around $170 billion if free cash flow remains under pressure?
There is also a UK-specific loose end. Nothing was said on the call about where the capacity planned for Nscale’s Loughton site — the centrepiece of Microsoft’s $30 billion UK investment pledge — will now go, and it is not clear whether that capacity will move to another UK site or simply arrive later than planned. The Competition and Markets Authority will also have a say in Microsoft’s UK position, with a final decision due by February on whether to designate it with strategic market status, which could bring conditions on how it sells Copilot and licenses its software.
For enterprise readers, the cost question is the most practical one. Claude Enterprise’s cost tools show how vendors are trying to make AI spending visible and manageable, while General Compute’s inference cloud illustrates the infrastructure race from the supplier side. Microsoft’s quarter sits at the intersection of both: a vendor spending heavily to supply AI, and customers trying to measure what they get for it.
Limitations and unanswered questions
The figures in this article are reported results and management guidance, not independent audits. The commercial remaining performance obligation figure of $677 billion, up 73%, was given in the task brief and was not re-verified from the fetched summary; it is therefore not relied on in the body of this article. The July paid-seat figure is labelled as a July figure for the same reason. No investment conclusion is drawn, and the UK capacity and regulatory questions above remain unresolved.
