KEY POINTS
- Microsoft posted fiscal Q3 revenue of $82.9 billion, up 18%, with Azure and other cloud services growing 40% on the back of AI workloads.
- AI annualized revenue now stands at roughly $37 billion, up 123% year over year, and management lifted calendar-2026 capex guidance toward $190 billion.
- Traders should watch the June quarter Azure print: management guided 39% to 40% constant-currency growth, the bar that determines whether the AI capex cycle is still clearing returns.
Microsoft delivered fiscal third-quarter revenue of $82.9 billion on Wednesday, up 18% year over year, with Azure and other cloud services growing 40% as AI workloads continued to absorb every gigawatt of capacity the company can bring online. Operating income jumped 20% to $38.4 billion, and diluted earnings per share rose 23% to $4.27, beating the Street and pulling the broader Magnificent Seven cohort higher into Apple's after-the-bell print on Thursday.
The number every desk fixated on was $37 billion. That is Microsoft's annualized AI revenue run rate as of the quarter end, up 123% from a year ago, and it includes both Azure consumption from model builders and Microsoft's own Copilot and OpenAI-powered services. Management's official commentary framed it as proof that demand is still running well ahead of supply, a line CFO Amy Hood has repeated in some form on every call since fiscal 2024.
The Capex Bill Is Now $190 Billion
The cost of carrying that demand is no longer abstract. Microsoft told investors it expects to spend more than $40 billion on capital expenditures in the June quarter alone, and full calendar-year 2026 capex is now tracking toward roughly $190 billion. That figure includes about $25 billion of incremental cost tied to higher component pricing, which is a polite way of saying that high-bandwidth memory and advanced packaging supply remains tight and that Nvidia, SK Hynix, and TSMC continue to capture pricing power further up the stack.
For context, Microsoft's full-year capex was $44.5 billion in fiscal 2024 and roughly $80 billion in fiscal 2025. The trajectory means the company is on pace to spend more on data centers and silicon in a single year than it has spent on R&D across most of its history. That is why the stock did not melt up on the headline beat: the higher the capex slope, the longer the runway investors need to model before AI gross margins reach maturity.
What the Hyperscaler Triangle Says
Wednesday's print sat alongside Meta's, which posted $56.3 billion of revenue (up 33%) and lifted its own 2026 capex range to $125 billion to $145 billion, per Fortune. Add in Amazon's roughly $200 billion of expected 2026 spend at AWS, and the three largest commercial buyers of accelerated compute are now telegraphing more than $500 billion of capex this year — a number Bank of America has cited in lifting its 2026 global semiconductor forecast to $1.3 trillion, with Nvidia, Broadcom, Marvell, and AMD as the top revenue beneficiaries.
That math has two implications for traders. First, the hyperscaler bid for GPUs and custom silicon is not normalizing in 2026; it is expanding. Second, the burden of proof is shifting from supply to demand. After a stretch in which the trade was simply long anything that ships into a data center, the next leg requires that AI revenue at the hyperscalers themselves continues to compound at a multiple of capex growth. Microsoft's 123% AI growth versus roughly 130% capex growth is exactly the kind of ratio bulls need to defend.
Where the Risk Sits
The soft spots in the print were real but not fatal. Constant-currency Azure growth came in at 39%, in line with the high end of guidance but not above it, and management's June-quarter outlook of 39% to 40% leaves little room for upside surprise. Intelligent Cloud segment guidance of $37.95 billion to $38.25 billion for fiscal Q4 implies a deceleration from the 27% to 28% growth pace, which could matter if buy-side models had penciled in re-acceleration on the back of new capacity coming online in Texas, Wisconsin, and the Middle East.
The other tail is the OpenAI relationship. A Wall Street Journal report earlier this week that OpenAI may miss internal revenue and user targets pressured tech sentiment going into the print, and the restructured commercial agreement between the two companies has left some investors uncertain about how Microsoft monetizes inference workloads it does not directly control. Management did not address the topic in detail on the call, which Chris Capossela's team has historically treated as a tell that the agreement is still being negotiated rather than finalized.
What Traders Should Watch
The next catalyst is the June quarter, which will be the first full period reflecting the higher component pricing flagged on this call. Watch three things: whether Azure constant-currency growth holds the 39% to 40% line that management just guided, whether AI gross margins inflect as the Nvidia Blackwell B300 ramp kicks in mid-quarter, and whether capex run-rate pushes through the $40 billion-per-quarter mark. A clean print on all three keeps Microsoft as the cleanest large-cap AI long. Any slip on Azure deceleration would put pressure on the entire hyperscaler complex, including Nvidia heading into its own May earnings date.
