Microsoft Smashes Estimates — Then the Stock Craters: The Azure “Miss,” $37.5B AI Capex Shock, and the $450 Line in the Sand

Microsoft’s quarter was one of those “beat-and-bleed” prints: results cleared the bar on revenue and EPS, but the stock sank anyway because the market didn’t like the mix, the deceleration optics, and the size of the AI spending bill that’s coming due. On the headline numbers, Microsoft posted revenue of $81.27B (above ~$80.3B expected) and adjusted EPS of $4.14 (above ~$3.9 expected), with GAAP EPS boosted by a sizable OpenAI-related accounting gain that flowed through “other income.” The punchline: the core business is still growing at a healthy clip, but investors are increasingly treating Microsoft like an infrastructure utility with a startup’s capex plan.

Cloud was the focal point, and it’s where the “disappointed” narrative comes from—even though Azure technically edged expectations. Azure and other cloud services grew 39% YoY (38% constant currency), versus estimates roughly around the high-38% range, but it was a touch slower than the prior quarter’s ~40% growth. That small step-down mattered because the stock had rallied into the 200-day moving average and positioning looked optimistic; when you’re priced for perfection, “39% instead of 40%” can get you a -6% candle. Management also reiterated that demand continues to exceed supply, which cuts both ways: it supports backlog and pricing power, but it also implies capacity constraints and timing-driven lumpiness that can create quarter-to-quarter volatility in reported growth.

If you want a cleaner “why cloud disappointed” explanation based on what they said : 1) growth decelerated slightly sequentially, 2) supply/demand timing is creating variability (capacity delivery, when it comes online, and contract mix affect in-period recognition), and 3) margins in cloud are being pressured by AI investments. Amy Hood guided Microsoft Cloud gross margin to about 65% next quarter, down YoY, explicitly due to continued AI investments—so even if cloud revenue is strong, the profitability profile is being diluted while they build out. In other words, the market’s issue wasn’t “Azure is broken,” it was “Azure is strong, but it’s expensive.”

Segment-wise, Productivity & Business Processes was a clear winner. Revenue was $34.1B, ahead of consensus, driven by solid Microsoft 365 performance and accelerating Copilot adoption. They called out 50M paid seats for Microsoft 365 Copilot, and separately cited 4.7M paid Copilot subscribers (+75% YoY) plus strong growth across paid GitHub Copilot—evidence that monetization is finally matching the hype cycle. Microsoft 365 Commercial Cloud grew 17% (14% CC), with ARPU uplift led by E5 and Copilot; commercial paid seats grew to 450M+. LinkedIn was steady (low double-digit growth, driven by Marketing Solutions), and Dynamics 365 remained healthy (high teens growth).

Intelligent Cloud was strong in absolute terms, but it carried the capex/margin overhang. Segment revenue was $32.9B, up ~29% (28% CC), with Azure doing the heavy lifting and on-prem server revenue getting a short-term boost from SQL Server 2025 plus transactional purchasing ahead of expected memory price increases. Management emphasized heterogeneous infrastructure, sovereign/region-specific deployments, and their “tokens per watt per dollar” optimization—basically telling investors they’re building an AI superfactory, not just renting VMs. They also highlighted their silicon roadmap (Maia 200 accelerator coming online, CPU progress with Cobalt 200) and the idea that sovereignty requirements are increasing, which plays to Microsoft’s enterprise and government relationships.

More Personal Computing was the soft spot and the easiest “why the stock is down” soundbite beyond capex. Revenue was $14.3B, essentially in-line but a touch below estimates. Windows OEM and Devices was helped by Windows 10 end-of-support dynamics and some pre-buying ahead of memory price increases, but management expects that benefit to normalize and guided Windows OEM/Devices down in Q3 (low teens decline; Windows OEM down ~10%). Search and news advertising ex-TAC grew ~10% but was slightly below expectations due to execution challenges and normalization of third-party partnership benefits. Gaming was weak, with revenue down and Xbox content/services below expectations due to first-party content timing and platform impacts, plus impairment charges weighing on expenses.

Spending and capex were the big “tell” in this report. Capex in the quarter was $37.5B including finance leases (above expectations), and Hood said roughly two-thirds went to short-lived assets like GPUs and CPUs—essentially a statement that AI capacity is being added as fast as supply allows. Total finance leases were $6.7B (primarily datacenter sites), and cash paid for PP&E was $29.9B, which helps explain why free cash flow was pressured sequentially despite cash flow from operations jumping to $35.8B (+60%) on strong billings and collections. They also guided capex to decrease sequentially next quarter due to timing variability, but emphasized the mix of short-lived assets stays similar as they work to close the demand-supply gap—so this isn’t a capex “peak,” it’s a capex “plateau with spikes.”

On the outlook, Microsoft guided Q3 revenue to $80.65B–$81.75B (15–17% growth) with FX as a tailwind, but they flagged operating margin down slightly YoY given continued AI R&D, compute capacity, and talent investment (opex guided to grow ~10–11%). They guided Azure growth to 37–38% constant currency next quarter—roughly in line with Street—but reminded investors that quarterly variability can be driven by capacity timing and contract mix. They also forecast Microsoft Cloud gross margin around 65% next quarter, down YoY. The forward message is consistent: the demand is there, backlog is massive (commercial RPO up 110% to $625B, with meaningful OpenAI contribution), but near-term margins and cash conversion will be noisy as they invest.

If MSFT has slides back toward ~$450 (near 2026 lows) after failing at the 200-day, that level becomes the “prove-it” zone: bulls want to see support hold while fundamentals stay intact, and bears will press if cloud deceleration optics plus capex intensity keep compressing the multiple. The fundamental counterweight is that Microsoft is simultaneously showing: 1) durable core productivity growth, 2) real Copilot monetization scale, 3) Azure still growing near 40%, and 4) a backlog/contract engine that suggests demand is not the problem—execution and capital intensity are the debate.