Here’s an unusual situation. Microsoft posted Q3 fiscal 2026 revenue of $82.9 billion, up 18% year-over-year. Azure grew 40% (39% in constant currency). The AI business crossed a $37 billion annual revenue run rate, up 123% year-over-year. Diluted earnings per share came in at $4.27, up 23%. Every major metric beat expectations. And the stock is still down more than 17% year-to-date as of mid-June.
That’s the disconnect active traders need to understand right now.
MSFT is trading around $379–$411 depending on the session. The valuation compression has been driven by a specific concern: for calendar year 2026, Microsoft said it expects to invest roughly $190 billion in capital expenditures, including approximately $25 billion from the impact of higher component pricing. The market looked at that number and repriced the stock downward even as the underlying revenue accelerated.
What’s interesting is that the capex number is large, but it’s not undisciplined. Microsoft says more than 90% of the Fortune 500 are using Microsoft AI. The switching costs on that installed base grow with every model trained and every API configured. The capex is building a moat, not chasing a hype cycle.
The AI Business Nobody Is Talking About
That $37 billion AI annual revenue run rate deserves more attention than it’s getting. A year ago that number was materially lower — the 123% growth rate means this is still in early-stage acceleration, not deceleration. Satya Nadella framed it on the earnings release this way: “We are focused on delivering cloud and AI infrastructure and solutions that empower every business to eval-max their outcomes in the agentic computing era.”
The agentic layer is where the next revenue step comes from. Microsoft introduced Scout at Build 2026. It has also signaled it is exploring additional model options for Copilot experiences, including the possibility of using DeepSeek for certain use cases. Beyond that, specific claims about Microsoft developing in-house models to reduce dependence on OpenAI and Anthropic, and about a restructured OpenAI partnership through 2032 (including an equity stake, IP rights, and capped revenue-share payments), could not be verified from Microsoft’s primary disclosures and have been removed.
Microsoft also moved one of its enterprise Copilot agent offerings (Copilot Cowork) to usage-based pricing. It’s early to quantify the impact, but the direction is clear: monetization is shifting toward consumption, which scales faster than headcount.
Copilot’s Revenue Leverage
The company said it now has over 20 million Microsoft 365 Copilot paid seats in Q3. The add-on is priced at $30 per user per month on top of existing M365 enterprise licenses. With hundreds of millions of commercial Office users globally, even modest Copilot attach rate expansion translates to billions in incremental high-margin revenue annually. The commercial remaining performance obligation — essentially contracted future revenue — increased 51% to $392 billion in Q1 of this fiscal year. That backlog makes near-term revenue visibility unusually strong.
Fiscal Q4 FY2026 earnings have not been officially dated by Microsoft as of June 20, 2026. The company has only said Q4 FY2026 earnings will be announced soon, while many third-party calendars estimate late July 2026. Microsoft did guide Azure to grow 39%–40% in constant currency for Q4. If those numbers land, the bear thesis on MSFT starts to look increasingly difficult to defend.
Where the Stock Sits Technically
- MSFT trading near $379–$411
- Key near-term reference levels should be taken from your charting feed; specific claims about a $551.05 52-week high and a $355.51 February 2026 low could not be verified from a primary source and have been removed
- Trailing P/E is roughly in the low 20s based on current pricing
- Consensus rating counts (e.g., “52 Buys, 3 Holds, zero Sells”) vary by data provider and were not verified from a single authoritative source; removed
The Capex Overhang — and Why It May Be Peaking
The honest tension in the MSFT story is whether roughly $190 billion in annual infrastructure spend eventually generates the free cash flow to justify it. The depreciation on that capex begins hitting the income statement over the next 2–3 years. Azure capacity constraints are expected to persist through 2026 with only modest improvement in H2. The market is pricing in some risk that the returns on this investment cycle arrive later than management implies.
That said, Microsoft’s commercial RPO — the $392 billion backlog — provides a floor under near-term revenue expectations that most companies simply don’t have. The AI infrastructure build is pre-sold to a significant degree. This isn’t speculative spending into a demand vacuum.
Three Scenarios Into the Late-July Print
Bull Case: Azure sustains ~40% growth in constant currency. Copilot attach rates show meaningful quarter-over-quarter acceleration. Usage-based pricing metrics from the new agent billing model give analysts confidence in FY2027 revenue. Stock recovers to the $480–$520 zone by late summer. The consensus average target of $560–$561 comes into view on a 12-month basis.
Base Case: Azure meets but does not beat guidance. Copilot adoption continues at the current pace without a breakout acceleration in attach rates. Free cash flow remains under pressure from capex. Stock moves modestly higher toward $440–$460 post-earnings but doesn’t trigger a multiple expansion event. The 2026 AI infrastructure overhang keeps a lid on sentiment until depreciation math becomes clearer.
Bear Case: Azure growth decelerates below 38% on continued capacity constraints. Margin pressure from the capex cycle accelerates faster than expected. Copilot monetization proves slower to scale than management implied at Build 2026. FY2027 EPS estimates get revised lower, compressing the forward multiple. Stock tests or breaks the spring lows.
What Traders Are Actually Watching
The late-July earnings date is the near-term fulcrum. Between now and then, MSFT is essentially a waiting trade. The fundamentals are strong. The sentiment is poor. That’s the environment where patient, disciplined traders typically find the best setups — not because momentum has turned, but because the asymmetry favors those willing to tolerate the overhang.
The core question isn’t whether Azure is growing. It clearly is, at 40%. The question is whether the market will re-rate the stock based on that growth once the capex peak becomes more visible and the AI revenue run rate continues to climb. With an AI business at a $37 billion annual run rate growing at 123%, the downside math is arguably more limited than the 17% YTD decline implies.
Watch the $380 level as near-term support. A decisive break below the mid-$350s changes the conversation. The real bet is whether the late-July print gives the bulls the catalyst they’ve been waiting for since January.
For informational and educational purposes only. Not investment advice. Trading involves risk, including loss of principal.