The Journal’s Jonathan Weil argued on August 26 that Microsoft’s reporting on capital expenditures, OpenAI, and Azure leaves investors “flying blind.” Paul Thurrott’s follow-up framed the argument as a familiar complaint about percentage growth rates without the underlying Azure revenue figure. The underlying point holds: Microsoft disclosed that Azure and other cloud services grew 43% in the June 2026 quarter, but did not disclose the business’s exact quarterly revenue, operating cost, operating income, or capital expenditures.
For enterprise IT buyers, this is not an abstract shareholder dispute. Microsoft’s ability to sustain Azure pricing, capacity expansion, AI service availability, and Copilot investment depends on an infrastructure business whose individual economics remain hidden inside a much larger reporting segment. The company’s public statements show extraordinary demand and extraordinary spending at the same time; they do not show how much one is paying for the other.
Azure remains buried inside Intelligent Cloud
Microsoft’s fiscal 2026 earnings release reported $39.3 billion in fourth-quarter revenue for its Intelligent Cloud segment, which includes Azure, Windows Server, SQL Server, enterprise support, and other server products. Azure and other cloud services rose 43% year over year, but the company supplied no dollar amount for that category in the release.
Microsoft did say that annual Azure revenue surpassed $100 billion for the first time during fiscal 2026. That is a useful threshold, and it establishes that Azure is now a business of enormous scale. It is also deliberately imprecise: “surpassed $100 billion” could mean barely over the line or substantially above it, and it offers no audited quarterly comparison that outsiders can use to calculate a run rate.
The company’s own metrics page makes the accounting boundary even less clear. “Azure and other cloud services” includes cloud and AI consumption services, GitHub cloud services, Nuance Healthcare cloud services, virtual desktop products, and other cloud offerings. That means the 43% growth figure is not a clean disclosure of Azure proper, even before trying to separate AI workloads from conventional infrastructure, databases, Windows Server migration, or managed services.
Microsoft does provide a Microsoft Cloud gross-margin figure: 66% for fiscal 2026, down from 69% a year earlier. But Microsoft Cloud is itself a mixed measure, including commercial Microsoft 365 cloud, Azure and other cloud services, commercial LinkedIn, and Dynamics 365. It is a useful directional signal that AI infrastructure and usage are pressuring margins. It is not an Azure profit-and-loss statement.
That distinction is at the center of the Journal’s complaint. Microsoft can accurately say its cloud business is growing rapidly and remains highly profitable in aggregate while leaving unanswered whether Azure’s incremental AI revenue is earning an attractive return after the cost of GPUs, data centers, power, networking, depreciation, and long-term capacity commitments.
The capex numbers measure different things
The Journal also focused on the gap between Microsoft’s $115.948 billion in fiscal 2026 additions to property and equipment and the $145.3 billion capital-expenditure figure that includes assets acquired under finance leases. Those are not interchangeable numbers.
The $115.948 billion figure appears in Microsoft’s cash-flow statement as additions to property and equipment. It reflects cash spending recorded through investing activities during the year ended June 30, 2026. The larger $145.3 billion figure, discussed with investors alongside earnings, incorporates infrastructure obtained through finance leases. Such arrangements can put servers, data-center equipment, or facilities into use without identical up-front cash treatment.
Microsoft has told investors that its capital intensity is rising with cloud and AI demand. S&P Global noted after the July earnings report that capital expenditures including finance leases reached $145.3 billion, while the Associated Press and Axios each reported the company’s sharp increase in AI and cloud infrastructure spending. The broad direction is undisputed.
What remains absent is a full bridge that lets readers allocate the $29.4 billion difference between cash additions and the broader capex measure, identify the asset categories involved, and trace them to individual businesses. Microsoft gives analysts enough information to understand that finance leases matter; it does not provide a simple reconciliation that explains how much of the broader total relates to GPUs, data-center construction, networking equipment, or other infrastructure.
Nor does Microsoft disclose what share of those investments supports Azure versus Microsoft 365, Copilot, GitHub, LinkedIn, internal research, or customers whose workloads may be hosted through Azure but whose commercial terms differ dramatically. As spending moves from tens of billions to well above $100 billion annually, that aggregation is increasingly consequential.
The OpenAI disclosure is new — and incomplete
The strongest factual development in Microsoft’s fiscal 2026 annual report is one the broad transparency criticism should not gloss over: the company now identifies OpenAI as a related party and provides figures that it had not previously disclosed in this form.
Microsoft’s July 29 annual report says it holds an approximate 25% interest in OpenAI on an as-converted basis and accounts for that investment under the equity method. It reports $24.1 billion in fiscal 2026 revenue from commercial arrangements with OpenAI, including revenue-sharing payments, as well as $6.0 billion in accounts receivable from OpenAI at June 30. Microsoft also disclosed $13.0 billion in total funding commitments tied to its investment, of which $11.9 billion had been funded.
Those figures matter because they put a number on a relationship that has often been described in vague strategic terms. They also reinforce why separating Azure’s reported growth from the economics of a single major AI partner is difficult. The $24.1 billion figure is not presented as Azure consumption revenue alone; it includes commercial arrangements and revenue-sharing payments. It therefore cannot be used to determine what OpenAI directly pays to run workloads on Azure, what Microsoft returns through contractual revenue sharing, or what the net economic contribution is after infrastructure costs.
Microsoft separately reported that fiscal 2026 net income benefited by about $5.0 billion from net gains on OpenAI investments, largely related to dilution gains from OpenAI’s recapitalization. Its non-GAAP figures exclude gains and losses from OpenAI investments. This is a valid effort to show investors an earnings view without investment-accounting volatility, but it adds another layer to the task of determining how much of Microsoft’s AI story is operating performance and how much is the financial effect of its ownership stake.
The Journal’s criticism is therefore better understood as a demand for economic disaggregation, not an assertion that Microsoft has made no related-party disclosures at all. The company has now disclosed meaningful OpenAI revenue, receivables, ownership, commitments, and investment effects. It has not disclosed the contractual flows, Azure infrastructure costs, or margins necessary to evaluate the relationship as an operating business.
Microsoft’s segment structure follows management, not customer purchasing decisions
Microsoft reports three operating segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. That structure is permitted under U.S. accounting rules when it reflects how the company’s chief operating decision maker reviews performance and allocates resources. It is not designed to match the products customers buy or the questions investors ask.
The consequence is that Windows Server licensing and Azure capacity can sit in the same reported segment even though their margin profiles and capital requirements are very different. A perpetual or subscription software license can scale with relatively little incremental infrastructure. AI model hosting and high-performance GPU capacity cannot. Blending them gives Microsoft a legitimate segment-level view while obscuring the economics of the fastest-growing and most capital-intensive activity within it.
Microsoft’s choice to disclose exact annual revenue for categories such as LinkedIn and Gaming in past reports makes the absence of an exact Azure revenue figure more conspicuous. The company plainly has the internal accounting capability to measure product lines. The question is whether it believes Azure-specific revenue, cost, and margin would be useful to outside investors — or commercially disadvantageous to disclose.
There is a legitimate competitive argument for restraint. Amazon does not disclose AWS operating costs in granular service categories, and Alphabet does not provide a product-by-product profitability breakdown for Google Cloud’s AI infrastructure. Yet both Amazon and Alphabet provide cloud revenue and operating-income figures for their named cloud segments. Microsoft provides neither an Azure-specific revenue total nor an Azure operating-profit figure.
What IT leaders can take from the reporting gap
Enterprise customers should not expect Microsoft’s financial disclosures to reveal the price or capacity outlook for a specific Azure region, GPU family, or Copilot agreement. They do reveal a more practical warning: Azure demand is being served through an unprecedented infrastructure expansion, and Microsoft has not publicly separated the cost of that expansion from the revenue it generates.
That makes corporate-level Azure growth an unreliable substitute for procurement analysis. A CIO evaluating reserved instances, Azure VMware Solution, Fabric capacity, Microsoft 365 Copilot, or AI Foundry consumption should model the organization’s own unit economics: committed spend, egress exposure, storage growth, network charges, required support tiers, and the cost of moving a workload if pricing or capacity changes.
Microsoft’s fiscal 2026 disclosures establish that Azure has crossed $100 billion in annual revenue and that the company is committing at least $145.3 billion in annual capex when finance leases are counted. What they still do not establish is the return Microsoft earns on that infrastructure, the portion attributable to AI, or the extent to which OpenAI-related revenue shapes the picture. Those are the numbers the company will need to disclose before outsiders can judge whether Azure’s AI expansion is producing durable cloud economics or simply consuming capital at historic speed.
