Microsoft's AI build-out is the most capital-intensive in its history, yet its financial reporting hides the economics of Azure, the platform absorbing most of the spending.
Microsoft's AI build-out is the most capital-intensive in its history, yet its financial reporting hides the economics of Azure, the platform absorbing most of the spending.

Microsoft's AI build-out is the most capital-intensive in its history, yet its financial reporting hides the economics of Azure, the platform absorbing most of the spending.
Microsoft reported $145.3 billion in total capital spending last year, up roughly 80 percent, yet buried Azure's results inside a segment that also holds legacy software.
"Cloud computing, meet cloudy numbers," Jonathan Weil, a Heard on the Street columnist at The Wall Street Journal, wrote in a critique of the company's disclosures.
Microsoft said Azure and other cloud services revenue rose 41 percent in the fiscal year ended June 30, without giving a dollar figure. Chief Executive Satya Nadella said on the July earnings call that Azure's annual revenue topped $100 billion. The company's Intelligent Cloud segment, which includes Windows Server and other legacy products, generated about $138 billion in revenue, up 30 percent, with operating income of about $57 billion.
The opacity matters because Azure is the engine of Microsoft's AI strategy and the destination for most of its data-center spending. Amazon.com reports Amazon Web Services as a stand-alone segment with disclosed margins; Microsoft's second-largest cloud business remains a black box. OpenAI's anticipated initial public offering could force more disclosure, since Microsoft owns 25 percent of the ChatGPT developer and recorded $24.1 billion in revenue from the relationship last year.
Microsoft's capital-expenditure disclosures add to the confusion. Additions to property and equipment on the cash-flow statement totaled $115.9 billion last year, the figure the company used to calculate free cash flow. In an accompanying release, Microsoft reported capex of $145.3 billion including assets acquired under finance leases. The company provides no reconciliation tying the two figures together, even though it discloses finance-lease amounts separately.
The gap matters for a company whose spending is going mainly toward Azure, the network of data centers that underpins much of the modern web and powers AI models. Mixing the costly AI-infrastructure engine with high-margin software masks Azure's true profitability. Microsoft provides granular detail on far smaller businesses — Xbox generated about $22 billion in revenue last year and LinkedIn about $20 billion — while Azure's numbers stay hidden.
Microsoft's relationship with OpenAI is another disclosure gap. The company owns 25 percent of the ChatGPT developer and has identified it as an equity-method investment, making OpenAI a related party. Yet Microsoft's filings included no related-party disclosures regarding OpenAI until the latest annual report, which was light on detail.
In that report, Microsoft said it recorded $24.1 billion of revenue from OpenAI last year, including revenue-sharing payments, along with $6 billion of accounts receivable. It provided no prior-period figures or profit data, leaving investors unable to judge whether the dealings were at arm's length. OpenAI is a major Azure customer, so a registration statement for its IPO could shed light on Azure's true performance.
Microsoft remains a powerhouse, and hopes for OpenAI still run high. But the information gap carries a cost: investors cannot tell whether the AI build-out is generating returns or burning cash, and the discrepancy between the $115.9 billion and $145.3 billion capex figures invites scrutiny. If OpenAI files to go public, the resulting disclosures could pressure Microsoft to open up its own books — or leave investors flying blind for longer.
This article is for informational purposes only and does not constitute investment advice.