Microsoft Stock Jumps 9% as Azure Crosses $100 Billion and Q4 Blows Past Estimates

Microsoft reported fiscal Q4 2026 revenue of $90 billion, up 18% year-over-year and well past the $87.62 billion consensus estimate. Adjusted EPS came in at $4.74 (GAAP $4.81, including a $0.07 gain tied to its OpenAI stake), against expectations of $4.24. The stock jumped 9% to $427 in Thursday trading, up from Wednesday’s $390.54 close.

The number that mattered most: Azure crossed $100 billion in annual revenue for the first time in fiscal 2026, with quarterly growth accelerating to 43%, up from 40% the prior quarter. Total Microsoft Cloud revenue hit $59.3 billion for the quarter, up 27%. That acceleration is the opposite of what a lot of analysts were bracing for — cloud growth was supposed to be decelerating by now as the initial AI infrastructure buildout matured.

Capex was the other release valve. Microsoft spent $41 billion including leases, under the $42 billion level the market had been treating as a warning line for AI spending running ahead of returns. CFO Amy Hood told analysts capex will grow further in fiscal 2027, but paired that with a change in depreciation assumptions — extending useful life on buildings to 25 years from 15 — which softens the near-term margin hit from all that data center construction.

Analysts moved fast. Citi raised its price target to $600 from $570, calling the quarter “a solid rebuttal to the bear case” on AI spending sustainability. Wells Fargo went further, to $650 from $625. Barclays was the outlier, trimming to $512 from $545, still bullish but flagging that capex intensity has to translate to durable margin at some point.

The subtext is the AI capex debate that’s been hanging over every hyperscaler earnings call this year: is the spending buying revenue growth, or just keeping pace with rivals in a race nobody can afford to lose. Microsoft’s answer this quarter is that Azure’s growth rate is accelerating, not plateauing, which is the strongest evidence yet that Copilot and enterprise AI workloads are actually landing as paid consumption rather than free-tier experimentation. Whether that holds through a fiscal 2027 with even higher capex is the question the next four quarters have to answer.

Sources

Related reading