Microsoft Q2 FY2026 Earnings: What the $81 Billion Quarter Means for Your Licensing Costs

Microsoft reported its fiscal Q2 2026 results on 28 January 2026, covering the quarter ended 31 December 2025. The headlines announced record revenue, another quarter of double-digit growth, and Microsoft Cloud crossing $50 billion for the first time. The stock promptly fell 10%, wiping out $357 billion in market capitalisation in Microsoft’s worst single day since March 2020.

Q2 marked the second consecutive quarter where Azure missed expectations, and Wall Street doesn’t give third chances when $80 billion in annual CapEx is on the line. The company is spending money faster than it can demonstrate returns, and nearly half of its future revenue backlog depends on AI startups that have never turned a profit. For enterprise customers, this quarter confirms what many have suspected: the bill for Microsoft’s AI ambitions is coming due, and you will be the one paying it.

Key points:

• Microsoft Cloud revenue crossed $50 billion per quarter for the first time, growing 26% year over year

• Azure growth slowed from 40% to 39%, breaking three quarters of sequential acceleration

• Capital expenditure hit $37.5 billion in a single quarter, up 66% year over year, with an annualised run rate approaching $100 billion

• 45% of Microsoft’s $625 billion contract backlog is tied to OpenAI, a company with no profitability track record

• M365 Copilot reached 15 million paid seats, representing just 3.3% penetration of Microsoft’s 450 million commercial M365 user base

• M365 Commercial revenue grew 17% while seat growth was just 6%, confirming Microsoft’s strategy of extracting more from existing customers

The Numbers Behind the Headlines

Let us start with what Microsoft actually reported, because the gap between GAAP and non-GAAP figures this quarter is unusually large.

Source: Microsoft Investor Relations Press Release

That 60% GAAP net income growth deserves scrutiny. It includes $7.6 billion in net gains from OpenAI investments. The prior year quarter had $939 million in OpenAI investment losses. This $8.5 billion swing flatters the year-over-year comparison considerably. Strip out the OpenAI accounting and you get the non-GAAP figures: 23% net income growth and 24% EPS growth. Strong results, certainly, but not the moonshot the headline suggests.

Revenue came in at $81.27 billion against analyst expectations of $80.27 billion, and non-GAAP EPS of $4.14 beat the $3.97 consensus. Microsoft also exceeded its own guidance of $79.5 to $80.6 billion. By any normal measure, this was a beat. The market sold anyway.

Where the Money Comes From

Microsoft reports three segments.

Source: Microsoft Investor Relations Segment Revenues

Productivity and Business Processes, which houses Microsoft 365 and Dynamics, remains the profit engine. $34.1 billion in revenue generating $20.6 billion in operating income represents a 60% operating margin. Productivity and Business Processes is the segment that pays for Microsoft’s AI experiments.

Productivity and Business Processes is Microsoft’s main profit-generating segment, Intelligent Cloud delivers the highest growth with lower margins, and More Personal Computing shows continued revenue decline.

Intelligent Cloud, which includes Azure, grew fastest at 29% but operates at a lower margin. The $13.9 billion operating income on $32.9 billion revenue works out to 42%. That margin has been compressing as Microsoft pours money into AI infrastructure.

More Personal Computing continues its slow decline. Windows OEM revenue grew just 1% (flat in constant currency), Xbox content and services fell 5%, and the segment as a whole contracted 3%. The one bright spot was Search and news advertising, which grew 10% (9% in constant currency) excluding traffic acquisition costs, per the Metrics page. Bing’s AI integration appears to be paying off, though not enough to offset weakness elsewhere. This is the legacy business that Microsoft is milking while it transitions to cloud and AI.

The $50 Billion Cloud Milestone

Amy Hood, Microsoft’s CFO, highlighted in the earnings press release that “Microsoft Cloud revenue crossed $50 billion this quarter, reflecting the strong demand for our portfolio of services.” The Microsoft News Source announcement confirmed this was the first time the company exceeded $50 billion in cloud revenue in a single quarter.

Microsoft’s cloud gross margin has dropped to 67%, the lowest in approximately three years, representing a 400-500 basis point contraction year over year. Microsoft attributes this to “AI infrastructure investments and growing AI product usage”. In plain terms: AI workloads are more expensive to run than traditional cloud services, and those costs are showing up in the margins.

The commercial remaining performance obligation, which represents contracted future revenue, doubled to $625 billion. That sounds impressive until you examine what is behind it.

RPO Breakdown: Where the $625 Billion Comes From

The diverse customers portion growing 28% is healthy. The problem is that 45% of the total, roughly $281 billion, is concentrated in a single counterparty.

The OpenAI Problem

Here is where the earnings story gets complicated.

Forty-five percent of Microsoft’s $625 billion backlog, approximately $281 billion, is tied to a single customer: OpenAI. That is a company which has never demonstrated sustained profitability and which, as The Register noted, Microsoft has lost its right of first refusal as exclusive compute provider.

The commercial bookings growth of 230% was driven by “large commitments from OpenAI and Anthropic”. Strip out these AI startup commitments and the remaining 55% of the backlog, approximately $344 billion, grew 28% year over year. Still strong, but not the doubling that the headline figure suggests.

45% of Microsoft’s commercial backlog is tied to OpenAI, while the remaining 55% comes from other customers across Microsoft’s portfolio.

Jefferies analyst Brent Thill told SiliconANGLE: “The disclosure that OpenAI is 45% of the backlog goes back to the situation where people are asking, can OpenAI achieve these financial goals to pay Oracle, Microsoft and many of the providers?” Valoir analyst Rebecca Wettemann was blunter: “Investors are losing patience, largely because much of Microsoft’s eventual potential payback is tied to money coming from OpenAI that is mostly hypothetical at this point.”

The Register’s summary captured the core risk: “But as any landlord knows, just because someone signs a lease doesn’t mean they’re good for the rent when the going gets tough.”

Amy Hood defended the backlog quality on the earnings call, describing the 55% not tied to OpenAI as “the breadth of our portfolio, the breadth of customers, across solutions, across Azure, across industries, across geographies.” She added: “I think we have super high confidence in it.” On the GPU spending, she noted that “the majority of the capital that we’re spending today and a lot of the GPUs that we’re buying are already contracted for most of their useful life.”

That addresses utilisation risk but not pricing risk or counterparty risk. If OpenAI stumbles, Microsoft has $281 billion in contracted revenue from a customer that may not be able to pay. Microsoft has also committed $30 billion from Anthropic and invested $5 billion directly in that company. The Anthropic deal includes up to 1 gigawatt of compute capacity deployment. Nvidia separately invested $10 billion in Anthropic, which will run on Nvidia Grace Blackwell and Vera Rubin systems hosted on Azure.

The Azure Slowdown

Azure remains Microsoft’s growth engine, but that engine is slowing.

Azure growth fell from 40% to 39%, breaking three consecutive quarters of sequential acceleration. The guidance for Q3 of 37-38% implies further deceleration.

But let’s be honest about what “deceleration” means at this scale. Azure grew 39% on a base of nearly $100 billion annualised revenue. That is extraordinary growth for a business of this size. The surprise is not that growth slowed by one percentage point; the surprise is that anyone expected 40%+ growth to accelerate indefinitely. Compound growth at these rates eventually hits mathematical limits. The market’s reaction suggests investors had priced in a trajectory that defied gravity.

RBC Capital Markets analyst Rishi Jaluria told CNBC it was “solid, but maybe a little bit light relative to expectations, and it did show a point of deceleration.” He noted the fundamental dilemma: “If you want to get out of being capacity constrained, you have to build more.”

Amy Hood offered an interesting defence to Fortune: Azure growth would have been “well above the 39%” if all new GPUs were allocated exclusively to Azure. Instead, GPU capacity was deliberately split across M365 Copilot, GitHub Copilot, and R&D. Microsoft is prioritising its first-party AI products over raw Azure growth. Whether that trade-off pays off depends entirely on whether those first-party products generate enough revenue to justify the slower Azure growth.

AI contributed 22 to 26 percentage points of Azure’s total growth, meaning non-AI Azure grew somewhere between 13% and 17%. The AI tail is now wagging the Azure dog.

The $100 Billion Question

The capital expenditure figures are where the market’s patience ran out.

Microsoft spent $37.5 billion on capital expenditure in a single quarter, 66% more than the same quarter last year and $3.2 billion above what analysts expected. The first half of fiscal 2026 totalled $72.4 billion, putting Microsoft on track for roughly $100 billion in annual capital expenditure.

Satya Nadella characterised this spending as functioning like R&D investment. He noted on the call: “All up, we added nearly one gigawatt of total capacity this quarter alone.”

Two-thirds of that capex goes to GPUs and CPUs with a six-year useful life. That equipment is not infrastructure that lasts decades. It is equipment that will be obsolete within a product generation or two.

Morgan Stanley analyst Keith Weiss told Fortune: “Capex is growing faster than we expected, and maybe Azure is growing a little bit slower than we expected. That fundamentally comes down to a concern on the ROI on this capex spend over time.” Edward Jones analyst Logan Purk told Yahoo Finance: “The capex increase will only fuel the debate on generating appropriate returns on such a large capital base.”

Amy Hood clarified the capex allocation priorities on the call: M365 Copilot and GitHub Copilot usage and sales acceleration come first, then R&D and product innovation, with the remainder going toward Azure capacity expansion.

Microsoft guided for Q3 capex to be down versus Q2, citing “normal variability from cloud infrastructure build outs and the timing of delivery of finance leases”. That provides some relief, but the overall spending trajectory remains steep.

The 3.3% Copilot Problem

Satya Nadella framed Microsoft’s AI ambitions boldly on the earnings call: “We are only at the beginning phases of AI diffusion and already Microsoft has built an AI business that is larger than some of our biggest franchises.” That is the narrative Microsoft wants you to hear.

The adoption figures complicate that narrative. M365 Copilot, Microsoft’s flagship AI offering and the metric investors watch most closely to gauge returns on infrastructure spending, has a long way to go.

Fifteen million paid Copilot seats sounds impressive until you realise Microsoft has over 450 million commercial M365 users. That works out to roughly 3.3% penetration. After more than a year of aggressive sales push, 96.7% of Microsoft’s commercial user base has not bought Copilot.

The 160% growth rate looks rather different when you consider that Microsoft has 450 million captive M365 users, a motivated sales organisation, and presumably quite a lot of product demos. After all that effort, 96.7% of users have somehow resisted the transformation. One hesitates to suggest the product might not yet justify the price tag, but the market appears to have reached that conclusion independently.

Microsoft highlighted that customers with 35,000 or more Copilot seats tripled, naming Fiserv, ING, NASA, University of Kentucky, University of Manchester, US Department of Interior, and Westpac as examples. The largest named deployment was Publicis with 95,000+ seats covering nearly all employees. These are real deployments, but they remain exceptions rather than the norm.

GitHub Copilot reached 4.7 million paid subscribers, up 75% year over year. That product has arguably demonstrated clearer value to its target audience than M365 Copilot has to general business users.

Microsoft claimed an AI annual revenue run rate of approximately $26 billion. Satya Nadella emphasised on the call that performance should be evaluated across the entire AI enterprise, including M365 Copilot, GitHub Copilot, Dragon Copilot, and Security Copilot, rather than just Azure AI services.

The question for enterprise customers is whether M365 Copilot delivers enough value to justify the $30 per user per month price tag. At SAMexpert, we track this closely. The ROI case remains difficult to prove for most organisations. Microsoft’s own penetration figures suggest the market agrees.

The Licensing Revenue Machine

This segment houses the products that most directly affect enterprise licensing costs.

M365 Commercial cloud revenue grew 17%, but seat growth was only 6%. That means Microsoft extracted 11 percentage points of growth from existing customers through price increases and upselling to E5 subscriptions. The Performance page attributes part of this to “E5 momentum,” which is Microsoft-speak for customers moving from E3 to E5 subscriptions.

The M365 Commercial products line, which includes on-premises and transactional licensing, grew 13% in constant currency. Microsoft attributed the growth to Windows Commercial on-premises components and Office 2024 transactional purchasing. Office 2024 perpetual purchases exceeded Microsoft’s expectations. Despite years of pushing customers toward subscriptions, demand for perpetual licensing remains.

Dynamics 365 grew 19%, outpacing the broader segment. All workloads contributed according to Microsoft. For organisations running Dynamics, this growth rate typically translates into renewal pressure and upselling conversations.

Microsoft Fabric and Data Platform

Microsoft’s data platform push showed strong numbers.

Microsoft Fabric crossed $2 billion in annualised run rate with 31,000+ enterprise clients, growing 60%. For organisations still running Power BI Premium, the migration to Fabric is becoming increasingly difficult to avoid.

Microsoft also announced Agent 365, which extends M365 and Azure governance to agents built on any cloud. Over 80% of Fortune 500 companies have active agents in Copilot Studio or Agent Builder. The agent licensing model combines per-user subscriptions with consumption-based message charges, meaning costs scale with both headcount and usage.

Forward Guidance and the July 2026 Price Increase

Microsoft’s guidance for Q3 FY26 and beyond sets the stage for what enterprise customers will face.

The revenue guidance midpoint of $81.2 billion barely exceeds analyst consensus of $81.19 billion. Microsoft guided Azure growth to 37-38% for Q3, down from 39% this quarter and 40% last quarter. The market heard what Satya Nadella’s optimistic phrasing tried to obscure: continued deceleration. Analyst consensus for Q3 Azure growth was 37.1% according to StreetAccount, so Microsoft is guiding roughly in line with expectations rather than beating them.

The operating margin guidance of 45.1% came in below the 45.5% consensus, reflecting continued pressure from AI infrastructure investments.

Most significantly for enterprise customers, Microsoft confirmed M365 price increases effective July 2026 for commercial Office productivity subscriptions. These increases, announced in December 2025, range from 5% to 33% depending on the SKU:

Microsoft is bundling Defender for Endpoint P1 into E3, Security Copilot allocation into E5, and additional Intune features across E3/E5 to justify the increases. Whether these additions justify the cost depends on whether your organisation was planning to purchase those capabilities anyway.

The frontline worker SKUs face the steepest increases. F1 rising from $2.25 to $3 represents a 33% jump. F3 rising from $8 to $10 is a 25% increase. Organisations with large frontline workforces will feel this disproportionately.

Combined with the Enterprise Agreement (EA) volume discount removal that took effect in November 2025, these price increases continue the pattern of extracting more revenue from existing customers. The 17% M365 cloud revenue growth against 6% seat growth this quarter demonstrates that this strategy is working from Microsoft’s perspective.

The Worst Day Since March 2020

Investors were unimpressed.

Microsoft shares fell 5-7% in after-hours trading despite beating expectations. By Thursday’s open, the stock was down approximately 10%, losing $357 billion in market capitalisation. It was the largest single-day dollar loss in Microsoft’s history and the seventh-largest percentage decline since the company went public in 1986.

Over the prior three months, Microsoft stock had already declined 11% while the S&P 500 rose 1%. The earnings reaction accelerated an existing trend.

Investor concerns centred on three issues:

1. Slowing Azure growth (40% to 39%, with guidance for 37-38%)

2. Capital expenditure significantly exceeding expectations ($37.5B vs $34.3B expected)

3. Concentration risk from AI startup commitments (45% of backlog tied to OpenAI)

The core tension is between massive infrastructure buildout and near-term revenue growth deceleration. Microsoft is spending money at an unprecedented rate on AI infrastructure while the revenue growth that should justify that spending is slowing.

Amy Hood noted on the call that “demand continues to exceed available supply.” Capacity constraints are expected to extend “at least” through FY26, which ends in June 2026. The supply-demand imbalance creates a difficult communication challenge: Microsoft needs to spend aggressively to meet demand, but that spending compresses margins and growth rates in the near term.

The Meta Contrast

The same day Microsoft fell 10%, Meta surged over 10%. The contrast was not lost on observers.

CNBC highlighted that Meta demonstrated its AI investments were already boosting bottom-line performance, while Microsoft could not make the same claim. Meta showed AI driving engagement and advertising revenue today. Microsoft showed AI driving costs today with revenue promised tomorrow.

This comparison illustrates the market’s growing impatience with the “spend now, earn later” narrative. Investors want to see returns, not commitments. Microsoft’s backlog may be impressive on paper, but nearly half of it depends on AI startups that have never demonstrated they can pay their bills at scale.

Analyst Reactions and Price Target Cuts

Wedbush’s Dan Ives summarised the disconnect: “Wall Street wanted to see less capex spending and faster cloud/AI monetisation … and coming out of the gates, it’s the opposite.” Despite the selloff, most analysts remain bullish, though several cut their price targets.

Deutsche Bank cut its target from $630 to $575, and Goldman Sachs cut from $655 to $600, both maintaining Buy ratings. Bernstein kept its $641 target and Outperform rating.

Bernstein’s commentary was instructive: Microsoft is “prioritizing 1st party apps and R&D over near-term Azure growth,” which they called “a hard but necessary decision” before concluding that “Microsoft’s growth engine is in fact getting stronger.” Truist described Q2 as “fundamentally strong beneath the surface.”

The bull case is that Microsoft is making the right long-term investments and accepting short-term pain. The bear case is that those investments may never generate returns proportional to their cost, particularly given the OpenAI concentration risk.

What These Results Mean for Your Licensing Costs

Price Increases Are Locked In

The July 2026 M365 price increases are confirmed. If you are renewing an EA or Microsoft Customer Agreement (MCA) in the coming months, you have a narrow window to lock in current pricing. The question is whether the bundled additions, particularly Defender P1 in E3 and Security Copilot allocation in E5, offset the increases for your specific situation.

For organisations already planning to purchase those security capabilities, the bundling represents genuine value. For organisations that were not planning to purchase them, you are paying for features you may not use.

Copilot Pressure Will Intensify

Microsoft is prioritising Copilot sales over Azure capacity expansion. Your account team will push Copilot harder in every conversation. The 3.3% penetration rate tells Microsoft the product is not selling itself, which typically leads to more aggressive bundling, promotional pricing, and pressure tactics.

If you are considering Copilot, evaluate it on its merits rather than promotional pressure. The economics of Copilot remain difficult to justify for most organisations. Microsoft’s own adoption numbers suggest the market agrees.

E5 Upselling Continues

The gap between seat growth (6%) and revenue growth (17%) confirms that Microsoft’s primary growth strategy is extracting more revenue from existing customers. E5 upselling is the main mechanism. Expect continued pressure to move from E3 to E5, particularly as Microsoft adds Security Copilot allocation to E5.

Evaluate E5 based on which components you will actually use. The headline feature bundle often includes capabilities that overlap with existing investments or that your organisation will never deploy.

Frontline Workers Face the Steepest Increases

F1 and F3 price increases of 25-33% disproportionately affect organisations with large frontline workforces. If you have significant F-series licensing, review whether all those users genuinely need the capabilities included, or whether some could be served by lower-cost alternatives.

Azure Spending Will Remain High

Despite the growth deceleration, Azure spending across the customer base continues to surge. At SAMexpert, we observe Azure expenses increasing 30-35% year over year across our client base. The combination of consumption growth, AI workload adoption, and price adjustments means Azure budgets require constant attention.

If you have a MACC (Microsoft Azure Consumption Commitment), review your consumption trajectory against your commitment. Underutilisation means you’ve prepaid for capacity you’re not using. Overutilisation means you’re paying list prices on the overage.

🖐 Prepare for upcoming Microsoft EA renewals and pricing changes. Learn more: Microsoft Enterprise Agreement Negotiation.

The Bigger Picture

For enterprise customers, the practical question is different from the investor question. You are not making an investment decision about Microsoft stock. You are making a purchasing decision about Microsoft products. The financial pressures Microsoft faces translate directly into the pricing, bundling, and sales tactics you will encounter.

Sales reps no longer have scorecards based on individual product sales. Their targets are now based on year-over-year growth in total customer spend. Your rep is not trying to sell you the right product for your needs; they are trying to increase what you spent last year. Understanding this dynamic helps explain the bundling, the upselling, and the resistance you encounter when trying to reduce scope.

If you need help with an upcoming Microsoft renewal or want an independent assessment of your licensing position, get in touch. We do not sell Microsoft licences or cloud services, so our advice is independent.