Microsoft reports 48% Azure growth as AI demand faces the returns test

A brushed silver cloud sculpture rests on miniature dark green server cabinets against a sage background.

Microsoft’s latest results show rapid Azure growth and rising operating profit. They also demand care over what the numbers measure: cloud revenue, future contractual commitments and returns from an AI investment are different kinds of evidence.

Microsoft reported a 48 per cent increase in Azure revenue in results released on 28 October, bringing a substantial new data point to the debate about demand for AI infrastructure. Total quarterly revenue reached $92.3 billion, up 19 per cent, while operating income rose 21 per cent to $45.9 billion. The figures cover the three months to 30 September, Microsoft’s first quarter of fiscal 2027.

For readers trying to judge the economics of AI, the central challenge is separating a supplier’s growth from its customers’ returns. A cloud business can sell more capacity while individual buyers are still working out which applications justify their expenditure. The results strengthen the evidence of demand at Microsoft. They cannot, by themselves, establish the productivity gains achieved by every company buying its services.

Read the cloud figures at their proper scale

In its quarterly release, Microsoft put Cloud revenue at $63.9 billion, up 30 per cent, and commercial remaining performance obligations at $677 billion, up 73 per cent. Reported net income increased 32 per cent; the increase was 18 per cent on the company’s adjusted measure excluding the impact of its OpenAI investments. These measures describe different parts of the business and should not be added together or treated as interchangeable evidence of AI revenue.

Azure’s growth rate is the clearest headline about the cloud platform in this release, but it is not a standalone measure of AI sales. The broader Cloud total is another defined reporting measure, not an additional pool of revenue to stack on top of the company total. Preserving those distinctions prevents a familiar mistake in coverage of large technology groups: counting the same economic activity more than once because several attractive figures appear in the announcement.

Future contractual obligations need similar care. They concern revenue associated with commitments that remain to be fulfilled, rather than money already recognised in the quarter. A large figure can support an argument about future demand without showing when the related services will be delivered, how much infrastructure will be needed or what profit will remain. For a business building capacity ahead of use, the timetable is part of the economics.

Profit growth needs its own explanation

The gap between reported and adjusted earnings growth is a reminder that an AI relationship can affect a technology company through more than product sales. Investment-related effects and operating performance belong in different parts of the analysis. Readers who want to understand how the underlying business performed should examine the reconciliation rather than assume the larger percentage describes an acceleration in customer spending.

Neither measure should be discarded simply because the other exists. Reported earnings follow the applicable accounting treatment, while an adjusted number offers a selected view with stated exclusions. The useful journalistic task is to explain the difference and retain the definition. Presenting one as the uniquely authentic result would conceal information that the comparison is meant to expose.

There is a further distinction between profit and the cash needed to expand a cloud business. Equipment purchases and the accounting charges associated with using that equipment do not necessarily fall in the same period. A profitable quarter therefore does not settle whether a particular expansion programme is generating cash at that moment. It is possible to recognise strong operating performance while still asking how quickly new capacity will earn back its cost.

The next test lies beyond demand

For customers, the practical question is the value of completed work. An AI service that helps prepare a document may still require checking, correction and integration into an existing process. A cloud supplier’s revenue records the service sold; a customer’s assessment needs to include the surrounding work. Conflating these perspectives turns evidence of adoption into an unsupported claim about savings.

Microsoft’s release supplies a strong account of growth, but the more revealing sequence will connect demand to delivery, delivery to sustained margins and customer use to measurable outcomes. Future disclosures that clarify those relationships would help distinguish durable economic improvement from expansion alone. The market now has another substantial quarterly result. The argument about AI’s returns still requires more than a revenue growth rate.

Image: AI-generated conceptual illustration for Nuvastra.

Previous
Previous

Oracle puts a cash-flow timetable on its AI infrastructure expansion

Next
Next

GLM-5.3 completes simulated cyberattacks as researchers find instruction failures