Microsoft Stock Earnings Preview: Can Azure Growth Justify $50 Billion in Spending?

Microsoft can report another quarter of strong growth tonight and still leave investors disappointed. The company expects capital spending to exceed $50 billion in three months. 

At that scale, a busy Azure cloud and a popular Copilot are only the beginning. The harder question is whether all that computing power will eventually leave more cash for shareholders.

Let’s break down what analysts expect from Microsoft’s earnings, what has changed since July, and which results would justify greater confidence in MSFT stock. We’ll also work through the cash-flow maths and the scenarios worth preparing for before the numbers arrive.

Microsoft Q1 FY27 Earnings Expectations: Revenue, EPS and Timing

Microsoft reports after the US market closes on Wednesday, October 28, 2026. These are its first-quarter fiscal 2027 results, covering July to September 2026. For Indian readers, the release falls early on October 29 IST, after the US closing bell at 1:30 a.m. The earnings call begins at 5:30 p.m. ET, or 3:00 a.m. IST on October 29.

Here is the starting point for reading the results:

Sources: Microsoft’s September reporting update and GeekWire’s October 27 preview. Revenue and EPS estimates are analyst forecasts; the other rows are company guidance.

The latest revenue estimate is already close to the top of management’s range. Even reaching $90.95 billion would exceed that snapshot by only $200 million, approximately 0.2%. A small revenue beat would therefore offer limited evidence about the returns on Microsoft’s much larger infrastructure commitment.

The December-quarter outlook could be more revealing. Investors need to know whether growth continues as new capacity arrives, and whether the cost of delivering it becomes easier to absorb.

What Has Changed at Microsoft Since Its Last Earnings?

July’s report left a strong business facing a rising investment bill. Four subsequent developments shape this preview.

September 2: Microsoft changed the reporting structure

Microsoft is moving from three reporting segments to two: Agents and Infra, and Devices and Consumer. Its September SEC filing makes this effective from fiscal 2027, so tonight’s results use the new structure.

Agents and Infra brings enterprise software and cloud infrastructure together. Devices and Consumer groups Windows, Xbox and advertising. Microsoft also plans quarterly Azure revenue disclosure, giving readers an actual dollar figure alongside its growth rate. Reuters reported the changes on September 2.

There is a catch. The contents of the product categories have changed too. GitHub’s cloud business and Security Copilot move into Microsoft 365, for example. A stronger Microsoft 365 growth rate can partly reflect the new perimeter; comparing it with an old-definition number would overstate the improvement. 

Think of a supermarket moving products between departments. A department’s revenue can change even before customers spend an extra rupee. Use Microsoft’s restated comparisons before drawing conclusions about acceleration.

September 25: Copilot moved further into completing work

Microsoft introduced a Copilot experience built around Home, Code and Autopilot. The ambition extends from answering questions to creating software and carrying out longer tasks. Some capabilities were entering its Frontier programme or private preview, rather than becoming universally available immediately.

A launch this late in September deserves more attention in the outlook than in explanations of the entire quarter’s revenue. The financial opportunity is discussed below: completing work can create a larger charging opportunity, but also a larger computing bill.

October 26: Anthropic disclosed a bigger Microsoft commitment

Anthropic’s public S-1 disclosed an agreement, amended in February and October, to spend at least $44.9 billion with Microsoft between November 2026 and May 2033, regardless of actual usage. Exceptions include Microsoft’s uncured material breach and undelivered tranches.

That strengthens evidence of future demand.

October 26: Microsoft introduced another voluntary exit offer

Microsoft announced a voluntary exit programme across participating teams in India, the US, Canada and the UK. GeekWire reported approximately 60,000 eligible employees, citing the company.

The offer arrived after the quarter ended. Any recognised charge needs to follow its accounting timing; recurring savings require actual departures and may be offset by replacement costs. Our Microsoft voluntary-exit explainer examines those economics separately.

Azure Growth: What Would Actually Impress Wall Street?

Azure rents computing power, storage and related services to customers. Its revenue grows when customers use more capacity, pay higher effective prices, or change the services they consume.

The revised management benchmark is 44% to 45% growth in constant currency, which removes exchange-rate effects. Under the restated definition, June-quarter growth was 42%, versus the previously reported 43% for “Azure and other cloud services”. The right comparison uses the new definition on both sides.

Even meeting guidance may produce a mixed reaction. GeekWire reported that BNP Paribas analyst Stefan Slowinski saw investors looking for approximately 46%.

The test is how quickly new capacity earns money. In July, management said additional capacity had been monetised quickly. Tonight, listen for whether newly available servers are again translating into customer consumption.

A restaurant analogy helps. A queue outside is encouraging. Adding tables creates value when diners actually occupy them and the extra meals cover the additional rent, staff and equipment. An expensive dining room with empty tables tells a different story.

For Azure, the strongest combination would be growth above the revised range, sustained demand beyond a handful of AI developers, and a December-quarter outlook showing that new capacity supports further expansion.

A capacity-related miss needs a different response from a demand-related miss. Delayed delivery can push revenue into a later quarter. Weaker customer consumption can reduce the return on equipment already installed. Ask management to identify the cause before treating either outcome as temporary.

Can Microsoft’s AI Spending Turn Into More Cash?

This is where the earnings preview becomes more interesting.

Microsoft’s latest full-year figures reveal the tension:

Sources: Microsoft’s FY2026 annual report and earnings release. Changes and free cash flow are calculated from reported figures. Fiscal years end on June 30. [S11] [S13]

That leaves $66.99 billion in calculated free cash flow, down 6.5% for the year.

Here is another way to read those numbers. For every additional dollar of annual revenue, Microsoft generated approximately 53 cents of additional operating profit. Yet calculated free cash flow fell by approximately 9 cents per additional revenue dollar, because cash investment increased faster than operating cash generation.

Those are company-wide comparisons, not a measurement of AI’s standalone profitability or return on capital. They show why strong accounting earnings and weaker cash left after investment can coexist.

The optimistic explanation is straightforward: Microsoft is paying to create capacity before collecting all the revenue it can produce. The difficult possibility is that maintaining its competitive position requires persistently higher spending, leaving less cash than investors expect.

Tonight cannot settle the lifetime return on a data centre. It can show whether the gap is starting to narrow, or whether the expected payoff keeps moving further away.

A lower reported capex figure needs an explanation

Microsoft extended the estimated useful lives of data centres and office buildings from 15 to 25 years, effective from FY2027. Amy Hood said the change shifts more future leases from finance leases into operating leases. Microsoft’s capital-spending measure includes the former and excludes the latter. Its revised calendar-2026 capex expectation of approximately $175 billion therefore did not represent a corresponding reduction in underlying investment plans.

This concerns buildings, rather than a claim that AI chips will remain competitive for 25 years.

A familiar comparison is a business paying for premises through a different rental arrangement. Changing the accounting category does not make the premises free. Equally, the accounting change alone does not establish that the premises are a poor investment.

Use three separate checks:

Operating-lease payments already reduce operating cash flow; subtracting them again would count the same outflow twice. Finance-lease principal is classified within financing cash flows, so inspect it separately.

Do not subtract the entire guided $50 billion-plus capex figure from operating cash flow and label the result free cash flow. That mixes a measure including financed additions with cash spending.

The persuasive outcome is not simply a smaller capex headline. It is productive capacity, improving cash generation and a credible explanation of the obligations needed to sustain both. Compare the September quarter with its year-earlier counterpart and examine a rolling year, because quarterly billing and payment timing can distort the picture.

Copilot’s Real Test Is Revenue per Customer

Microsoft reported more than 30 million paid Microsoft 365 Copilot seats in July. Paid adoption provides a firmer starting point than downloads or expressions of interest.

But a paid seat is not a profit figure. A customer can receive a discount, adopt a bundle, use expensive models intensively or renew on different terms. Multiplying seats by a headline subscription price would conceal those differences.

September’s announcement makes the commercial model especially relevant. Microsoft described fixed subscription access for everyday work alongside usage-based billing for more demanding agent work. An agent is software that can carry out a sequence of tasks; those longer assignments can require much more computing than a short answer.

For a customer, that resembles paying a basic mobile-plan charge plus additional charges for heavier usage. For Microsoft, charging for extra work can help revenue keep pace with the cost of serving it.

The opportunity is attractive: earn more from an existing customer relationship as the product completes more valuable work. The risk is that customers restrict usage when bills become unpredictable, or that computing costs absorb the extra revenue.

Look for evidence connecting paid adoption, revenue per user and delivery costs. Improving ARPU alongside healthy renewals and stable margins would be stronger evidence than a seat-count milestone alone.

Also separate growth generated by customers from growth generated by moving businesses into the Microsoft 365 category. Microsoft’s new presentation can improve visibility, but it cannot answer every question about Copilot’s individual profitability.

How Much Growth Is Already Priced Into MSFT Stock?

Microsoft closed at $531.33 on October 27, down 0.24% that session. Against FY2026 EPS of $17.28 excluding OpenAI investment effects, that represents approximately 30.7 times annual earnings.

The useful question is how much earnings must grow if investors become less willing to pay a premium.

Calculation: reference share price ÷ assumed multiple. These are sensitivity assumptions, not price targets or forecasts. The EPS starting point excludes OpenAI effects but is not fully normalised recurring profit.

For example, 15% EPS growth followed by a 27-times valuation produces an illustrative value of $536.54, only about 1% above the reference price. Earnings can progress substantially while the share price makes little progress if the valuation multiple contracts.

That is why the cash evidence deserves weight. It helps assess whether Microsoft’s premium rests on improving economics or on investors remaining comfortable with a distant payoff.

For the immediate event, Investopedia’s October 26 options snapshot indicated approximately 5% movement in either direction by the end of the week, using the October 23 close. It is an earlier snapshot rather than a fresh October 28 quote. Options indicate expected volatility; they cannot establish direction or limit the eventual move.

Anyone tracking Microsoft stock should therefore keep two questions separate: how much the price might move this week, and how much durable earnings and cash generation the current valuation requires.

Microsoft Earnings Scenarios: How Investors Can Respond

These are analytical scenarios, not predictions. The suggested responses focus on what evidence should change an investment decision.

For existing investors, the central task is deciding whether the original growth and cash assumptions remain credible. For prospective investors, it is deciding whether the price compensates for what still needs to be proved.

An upbeat quarter can strengthen the business case while a sharp price increase reduces the prospective return. A disappointing quarter can create a more attractive price while also exposing a problem that deserves a lower valuation. The direction of the price alone cannot settle either judgment.

Position size belongs in that assessment. If an ordinary earnings move would force an immediate change of plan, the exposure may be too large for the uncertainty being accepted.

How to Read Microsoft’s Earnings Release and Guidance

Start with revenue and operating profit, then move to product growth, cash flow and the outlook. Three traps deserve particular attention.

First, an adjusted EPS beat can still contain unusual benefits. June-quarter EPS was $4.81 under GAAP and $4.74 after Microsoft’s OpenAI adjustment. The quarter also contained an Anthropic investment gain of $3.2 billion. The OpenAI adjustment did not remove that separate gain.

Read the reconciliation rather than assuming “adjusted” means every exceptional item has disappeared. A gain on an investment can increase EPS without showing that customers paid more for Azure or Copilot. Conversely, an investment-related accounting loss need not signal weaker customer demand.

Second, contracted revenue needs a timetable. Microsoft reported commercial remaining performance obligations, or RPO, of $678 billion. RPO is revenue contracted but not yet recognised. On the July call, management said growth excluding OpenAI was 25% and approximately 30% of total commercial RPO would become revenue within twelve months.

A backlog is like a builder’s order book: valuable visibility, but work must still be delivered before the associated revenue is earned. Watch the portion due soon, customer breadth and any change in delivery assumptions. Anthropic’s October disclosure also should not automatically be added to Microsoft’s last reported backlog; the reporting dates and recognition requirements differ.

Third, ask what changes in the next quarter. The most useful management answers will connect capacity delivery to consumption, consumption to profit, and profit to cash. An assurance that AI demand is strong leaves the investment question only partly answered.

Microsoft also guided to a decline of less than one percentage point in full-year FY2027 operating margin. Any revision should be assessed alongside changes in expected growth and spending.

The standard for greater confidence should be demanding. Azure growth must remain healthy after the reporting changes, Copilot must earn its way beyond adoption headlines, and the cash left after infrastructure spending must develop a credible upward path.

Microsoft has a powerful opportunity to serve both the infrastructure and the applications behind enterprise AI. The investment case becomes stronger when each additional wave of growth leaves investors with more than a larger construction bill.