Microsoft spent $37.5 billion on capex in one quarter—more than many entire companies make in revenue—while Azure grew 39% (38% constant currency) and the company returned $12.7 billion to shareholders. The stock still dropped nearly 25% in Q1 2026. Wall Street labeled it the worst quarter since 2008. You watched the forward P/E compress to roughly 21-22x, levels not seen in years. Reality is the punchline.
Consensus piled on the panic. Massive AI infrastructure spend collides with any hint of Azure deceleration. Copilot adoption looks slow at scale. Hyperscalers or custom ASICs will bypass Redmond. The market repriced Microsoft as if the cloud engine had stalled and the capex was a black hole with no near-term payoff. Shares took their biggest hit since the financial crisis.
The numbers tell a different story. In fiscal Q2 2026, Azure and other cloud services revenue rose 39% year-over-year (38% constant currency), with AI contributing an estimated 13-16 points to that growth. Microsoft Cloud revenue hit $51.5 billion, up 26% (24% constant currency). Overall revenue grew 17% to $81.3 billion. Intelligent Cloud delivered 29% growth. That is not a stalled engine. That is a business compounding at scale while front-loading the next infrastructure layer.
You are not buying a pure-play AI name that swings with every GPU shipment. You are buying a diversified cash compounder where Office, LinkedIn, and Dynamics still deliver steady, high-margin revenue to fund the build. The commercial remaining performance obligation—the backlog—surged to $625 billion, up 110%. That gives more than two years of visibility at current run rates, much of it already committed to Azure and related workloads. Demand exceeds supply. Microsoft rations capacity rather than chasing it.
Capex reached $37.5 billion in the quarter, up 66% year-over-year. Roughly two-thirds went to short-lived assets like GPUs and CPUs that turn over quickly. The balance funds long-lived infrastructure that monetizes over 15 years or more. Efficiency gains in the existing cloud fleet already allow reallocation and faster monetization. This spend expands the total addressable market and locks in enterprise workloads for years. Most companies would kill for this kind of "problem."
Copilot has 15 million paid seats on a Microsoft 365 commercial base exceeding 450 million seats—that is roughly 3.3% penetration. Early innings, not failure. Seat adds grew 160% year-over-year. Large Azure and M365 deals drove commercial bookings up 230%. ARPU lift from E5 upsells and agentic features still has plenty of room. You do not need rapid 50% penetration for the math to work. You need steady conversion on a massive installed base while the infrastructure moat widens.
The forward P/E compressed to the low 20s during the selloff. You pay roughly 21-22x for mid-teens EPS growth, superior margins, and a backlog most software peers only dream about. Compare that to the S&P 500 multiple on slower growth. The market handed you a discount on a business that has delivered 15-20%+ growth through every prior cycle.
Microsoft returned $12.7 billion to shareholders via dividends and buybacks in that same quarter. Cash flow from operations hit $35.8 billion, up 60%. The balance sheet funds the build-out comfortably. This is not overextension. It is disciplined capital allocation that rewards owners while positioning for the long game.
Here is the deadpan fact bomb: Microsoft dropped a $625 billion backlog on the table while spending $37.5 billion on AI infrastructure in one quarter—and still grew cloud revenue 26% while returning $12.7 billion to owners. The stock treated it like bad news anyway.
This setup is classic buy-the-weakness at a reset multiple with the growth engine still humming. The diversified base keeps throwing off cash. AI spend expands TAM rather than cannibalizing it. You own the platform layer most enterprises will not rip out.
What kills the thesis: Azure growth decelerates below 35% constant currency in the next two reported quarters (Q3 or Q4 FY26) without a clear AI offset. Microsoft Cloud gross margin falls and stays below 66% for two consecutive quarters. FY2026 total capex exceeds $150 billion with no visible acceleration in AI-related revenue run-rate. Copilot paid seats penetration stalls below 5% by the end of calendar 2026 with no ARPU lift from E5 upsell. OpenAI-related exposure triggers material credit or partnership issues visible in 10-Q disclosures. Any one of those trips the wire and the story changes. Until then, the compression looks like an overreaction to temporary optics.
The consensus crowded one side of the boat. You see the data. Azure at 39%. Cloud at $51.5 billion. Backlog at $625 billion. Shareholder returns of $12.7 billion. Capex funding durable assets on top of efficiency gains. The multiple reset created the margin of safety. This is not blind faith in hype. It is owning a proven compounder at a price that finally reflects temporary indigestion instead of structural strength.
Microsoft is not in a pickle. The market is. Buy the weakness. The infrastructure bet is already generating the numbers that matter, and the valuation has adjusted to give you an entry most cycles never offer.