Microsoft closed its second fiscal quarter of 2026 with figures that, on paper, align with the script the company has been advocating for months: more cloud, more Artificial Intelligence, and more forward-looking contracts. However, the stock market reaction was the opposite of what many expected. After reporting results above forecasts, the stock fell around 6% in after-hours trading, reflecting a concern that has already taken hold in the sector: AI demand is growing faster than the capacity to serve it, and the cost to catch up is enormous.
The Numbers: a clear “beat,” with cloud over $50 billion
In the quarter (ended on December 31, 2025), Microsoft reported $81.3 billion in revenue, a +17% year-over-year. Earnings per share also increased significantly: in GAAP accounting, diluted EPS was $5.16, while in non-GAAP terms, it was $4.14, surpassing the consensus which hovered around $3.92–$3.93.
The strongest momentum is in the cloud. Microsoft Cloud generated $51.5 billion, a +26% increase, and the company highlighted another key metric that acts as a thermometer for future demand: pending performance obligations (Commercial RPO) grew 110% to $625 billion.
Azure and the “capacity equation”: 39% growth isn’t enough if the market wanted more
On the technological front, the focus was once again on Azure. Microsoft indicated that Azure and other cloud services grew by 39% year over year (a 38% increase at constant currency). While this figure is powerful, it was scrutinized because the market has been interpreting Azure as the primary indicator of actual monetization of AI in the cloud: not just “using AI,” but how much revenue it generates and how quickly the infrastructure scales to support it.
The Intelligent Cloud segment contributed $32.9 billion (+29%). Meanwhile, Productivity and Business Processes, which includes Microsoft 365 and LinkedIn, brought in $34.1 billion (+16%). Conversely, More Personal Computing (Windows, gaming, devices, and search advertising) declined by 3%, to $14.25 billion, reminding us that not all business moves at the pace of the cloud.
The reason for the setback: AI is not just software; it’s capex, energy, and chips
The market’s interpretation focused less on “whether Microsoft is growing” and more on how much it costs to grow during the AI era. In 2026, large-scale AI is resembling less traditional software (with very high margins and almost limitless growth) and more an infrastructure industry: electricity capacity, data centers, networks, GPUs, HBM memory, rack availability… and, crucially, capex.
Reuters summarized the clash of expectations: Microsoft narrowly exceeded Azure’s growth forecast, but concerns shifted to spending and capacity, a combination that might slow down the “visible” growth rate even if demand remains robust.
Practically, the market is penalizing three ideas:
1) Short-term capacity bottlenecks.
When a company reports such a massive future pipeline (RPO), investors see not just “secured sales”: they also wonder whether the provider can deploy infrastructure fast enough to fulfill those commitments without delays.
2) Margin pressure from accelerated investments.
Serving AI requires expanding data centers, acquiring accelerators, resizing internal networks, and boosting storage. This “digital factory” has returns, but they are not immediate. The market fears upcoming quarters where growth is visible but accompanied by shrinking margins.
3) Lack of a clear signal of immediate acceleration.
With Microsoft and other giants investing at high speed, any message sounding like “we’re doing well, but need time to scale capacity” tends to cause volatility.
The metric that explains everything: $625 billion in RPO
In a tech sector context, it’s worth focusing on RPO because it links business and architecture. $625 billion in committed but not yet recognized revenue isn’t just contracts; it’s future infrastructure load. A growing RPO by 110% suggests that the cloud consumption “pipeline” is there, but converting it into recognized revenue depends on the company’s ability to provision compute, networking, and storage almost industrially.
In other words: the quarter confirms that Microsoft’s cloud is in full traction, but also that the next phase of AI will largely be capacity engineering. And that type of engineering isn’t solved by a software update; it requires physical work, energy, logistics, and a supply chain still under tension in many segments.
A “good quarter” in a market already pricing the next battle
The result leaves an industry-like conclusion: Microsoft seems to be executing well on the present — revenue, EPS, cloud — but the market has moved ahead to the next question, which is purely technological: how quickly can Azure scale to avoid leaving demand on the table?
In 2026, the competitive edge is no longer just about who has the best model or the best copiloto. It’s also about who can guarantee stable capacity for companies that want production AI, not just pilot projects. In this context, a “very good” quarter can still be punished if the market perceives the ceiling isn’t demand’s limit… but the ability to serve it.
Frequently Asked Questions
What does it mean that Microsoft has $625 billion in RPO?
It indicates a portfolio of committed, yet-to-be-recognized revenues; in cloud, it’s a sign of future demand but also of pressure to expand infrastructure and meet deadlines.
Why might a 39% increase in Azure not convince the market?
Because investors compare the figure with consensus expectations and, crucially, with the costs to sustain that pace. If growth depends on more capacity, capex and margins impact valuation in the short term.
What impact does AI have on Microsoft’s data center spending?
AI demands more acceleration, networking, and energy, driving up capex and potentially squeezing margins short-term, even as revenues grow.
Which part of Microsoft’s business is most sensitive to AI demand?
Primarily Azure and cloud services within Intelligent Cloud, as they absorb compute and monetize AI consumption at scale.