Microsoft reported solid fourth-quarter earnings on Wednesday, but investors didn't care about the top-line numbers. They zeroed in on two uncomfortable truths: Azure cloud growth is stalling, and the company is set to spend over $100 billion on AI infrastructure this year. The reaction was brutal. Microsoft shares tumbled 10% on Thursday and continued falling Friday, wiping out $381 billion in market value over two sessions. It marked the company's worst week since March 2020 and the second-largest single-day market cap loss in history, trailing only Nvidia's $589 billion "DeepSeek Shock" drop in January 2025.
The Numbers Behind the Selloff
Microsoft's revenue actually grew 17% to $81.3 billion, and the company crossed the $50 billion quarterly revenue milestone for its cloud business for the first time. Capital expenditures came in at $37.5 billion for the quarter alone, up 66% year-over-year, with the first-half total reaching $72.4 billion. That spending is driven almost entirely by data center expansion and AI chip procurement to serve customers like OpenAI and Anthropic.
But Azure revenue growth, the metric Wall Street watches most closely, came in at 39%, slipping from 40% the prior quarter. The miss was small in absolute terms but devastating in sentiment. Investors had been growing increasingly anxious about whether massive AI infrastructure spending would translate into proportional revenue growth, and the slight deceleration confirmed their worst fears.
The OpenAI Concentration Risk
Perhaps the most unsettling disclosure came buried in Microsoft's financial statements. The company revealed that OpenAI accounts for 45% of its $625 billion revenue backlog, representing $281 billion in committed cloud spending. That figure stems from OpenAI's $250 billion Azure agreement announced in October.
The problem is that OpenAI remains unprofitable. Anthropic disclosed this week that it expects to spend $12 billion on training and $7 billion on inference in 2026 alone, and industry observers expect OpenAI's cost structure to be similar or larger. Investors are now asking an uncomfortable question: what happens to Microsoft's backlog if OpenAI cannot sustain its growth trajectory or fails to achieve profitability before burning through its capital?
Wedbush Securities analyst Dan Ives acknowledged the tension directly, noting that Wall Street wanted to see less capital expenditure and faster cloud monetization, but instead got the opposite. He framed 2026 as the inflection year for Microsoft, arguing the company needs to push forward with its data center buildout as more customers move down the AI path.
The Broader Sentiment Shift
Microsoft's selloff is not happening in isolation. The iShares Expanded Tech-Software Sector ETF dropped 5% on the same day, and ServiceNow shares fell 10% on concerns that AI tools are cannibalizing traditional enterprise software demand. The software sector has declined 25% since the start of 2026, raising questions about whether AI is disrupting the very companies supposed to benefit most from it.
The contrast with Meta underscores how narrow the market's tolerance has become. Meta announced plans to spend between $115 billion and $135 billion on AI in 2026, nearly double its 2025 spending, yet its shares surged 10% after reporting 24% revenue growth. The difference came down to one factor: Meta demonstrated that AI spending was already driving advertising revenue. Microsoft has not yet made that same case convincingly for Azure.
Microsoft CFO Amy Hood acknowledged that Azure would have grown 40% if the company had allocated all new GPU capacity to its cloud business rather than balancing internal AI products like GitHub Copilot and Microsoft 365 Copilot. The admission highlights a strategic tension at the heart of Microsoft's AI strategy: the company is simultaneously building for external cloud customers and consuming massive compute for its own AI products.
What This Means for the AI Investment Thesis
The $381 billion wipeout does not necessarily signal that AI spending is misguided. Microsoft's demand backlog doubled to $625 billion, suggesting genuine enterprise appetite for AI cloud services. The company also disclosed 15 million paying subscribers for Microsoft 365 Copilot at $30 per month per seat, putting the product on track to generate billions in annual revenue.
But the selloff represents a clear shift in Wall Street's patience. For months, investors accepted "building for the future" as justification for enormous capital expenditure programs. Now they are demanding faster evidence that those investments generate returns. Microsoft, with its $3.2 trillion market cap and its deepest exposure to OpenAI's fortunes, became the most visible target for that reckoning.
The coming quarters will determine whether this is a temporary correction or the beginning of a broader repricing of AI infrastructure stocks. Either way, the era of uncritical enthusiasm for AI spending appears to be over.
