Three years ago, Microsoft was a mature software giant with respectable but predictable margins, a cloud business growing fast enough to keep investors interested, and a bold bet on a small AI research lab that most analysts treated as a science project. Then the science project started paying rent. Today, the company’s operating margin sits above 46%, a level that most large-scale technology businesses never reach, and the stock trades roughly 10% below its 52-week high. Those two facts sit in uncomfortable proximity: a business performing at a historic peak, priced as though the market isn’t sure it believes what it’s reading.
That tension is the story. Not whether Microsoft is a good company, which the filings settle quickly, but whether the price today is asking for something the math can actually deliver, or something more fragile.
In one breath
The 10-K filed July 29, 2026 shows Microsoft earning more than 40 cents of net profit from every dollar of revenue, a margin the company has never sustained before. The stock’s current price-to-earnings ratio sits in the lower third of Microsoft’s own decade-long range, meaning the market is paying less of a premium for these results than it has paid for weaker ones. The open question is whether a margin this high is a new floor or a temporary ceiling, because the answer changes the math on every dollar the stock costs today.
How a software company became a margin machine
Microsoft’s operating margin history reads like a slow climb interrupted by one sharp jump. For several fiscal years running up to FY2026, net margins held in a tight band as the company absorbed the cost of building Azure data centers and funding the OpenAI partnership it announced in January 2023. The logic then was straightforward: spend now, earn later. What the fiscal year 2026 filing shows is that “later” arrived faster than almost anyone modeled.
The FY2026 10-K reports a net margin of 40.3% on $331.8 billion in revenue, a four-percentage-point jump in a single year. Think of it this way: for every hundred dollars a business customer spent with Microsoft last fiscal year, the company kept roughly $40 after every bill was paid. A year earlier, it kept $36. That four-dollar difference, multiplied across a revenue base the size of a mid-sized country’s GDP, is what produced the earnings surprise that sent the stock up sharply after the July 29 earnings call.
The engine behind it is Azure. Futurum Group noted that Azure crossed $100 billion in annual revenue for the first time in FY2026, growing at a remarkable clip in the fourth quarter alone. Cloud businesses carry high fixed costs early and high margins later, once those costs are covered. Microsoft appears to have crossed that threshold.
What the price is actually asking for
Here is where the audit gets interesting. At the current price, Microsoft trades at a meaningful discount to the decade-median multiple the market has historically been willing to pay for its earnings. Our data shows today’s price-to-earnings ratio sits in the lower third of Microsoft’s own ten-year range, meaning the market is applying a below-average premium to above-average results. The gap between the current price and what that decade-median multiple would imply is not a valuation target; it is a description of how much the market is discounting the current results relative to its own history.
The discount exists because the market is asking a specific question: can a 40%-plus operating margin survive the capital expenditure cycle that produced it? Microsoft guided on the July 29 earnings call for capital expenditures exceeding $50 billion in the first quarter of FY2027 alone, with full-year spending projected to dwarf anything in the company’s history. To put that in graspable terms: Microsoft is planning to spend, in a single year, more than half of what the entire company earned in revenue just four years ago. That is an extraordinary commitment, and it is the structural assumption the current stock price is quietly stress-testing.
Will Rhind, CEO of GraniteShares, told Futurum Group in early August 2026 that Microsoft had “finally answered the question the market has been asking for 18 months” about AI returns, crediting a stock re-rating to visible AI revenue. That’s the bullish read. The bearish read lives in the same filing: roughly two-thirds of the most recent quarter’s capex went to short-lived assets, CPUs and GPUs that depreciate quickly and must be replaced. A business that must continuously spend at this rate to maintain its growth rate is a different animal than one that built its infrastructure and now harvests it.
The one number that breaks the thesis
Morningstar Senior Equity Analyst Dan Romanoff maintained a $600 fair value estimate as of July 30, 2026, treating the capex cycle as manageable. Sell-side consensus, per Stock Analysis data, clusters around consensus estimates well above the current level, with forward earnings estimates that assume margins hold near current levels even as capex rises. That requires Azure revenue growth to continue outrunning infrastructure costs.
The filed evidence that supports that assumption is the commercial remaining performance obligation figure: $678 billion as of Q4 FY2026, up 84% year-over-year, per the 10-K. A remaining performance obligation is essentially a signed contract for future revenue, work promised but not yet delivered. At $678 billion, Microsoft’s reservation book is larger than the GDP of Switzerland. About 45% of it is tied to OpenAI commitments, which concentrates both the upside and the counterparty risk in one relationship.
The number that breaks the thesis is simpler: if Azure growth decelerates sharply from its current pace, the revenue acceleration that justifies the massive capex commitment disappears, and the margin math reverses. Microsoft guided for continued strong Azure growth in Q1 FY2027, per the earnings call, and noted demand continues to outpace available capacity. That guidance is the load-bearing wall. If it holds, the gap between the current multiple and the decade median would be arithmetically explained by the filed results rather than by a structural deterioration in the business, a distinction readers can weigh against their own frameworks. If it cracks, the capex commitments become a weight rather than an investment.
Reading the numbers
- Net margin, FY2026: 40.3%. This is what’s left of each dollar of sales after every cost, tax, and interest payment. Microsoft’s prior four-year range was 34% to 37%. The jump to 40% means the AI infrastructure spending is now generating returns faster than it’s generating costs. A household analogy: imagine a rental property that cost $200,000 to renovate now generating $40,000 a year in rent instead of the previous $36,000. The renovation is paying off.
- P/E ratio: 32.9 times, versus a decade median of 35 times (our data). The price-to-earnings ratio is the price tag per dollar of annual profit. At 33 times, investors are paying $33 for every $1 of Microsoft’s filed earnings. The fact that this sits in the 33rd percentile of Microsoft’s own history means the market is applying a below-average premium to above-average results. A store selling last year’s bestselling product at a discount is either clearing inventory ahead of a better model or repricing to reflect uncertainty about future demand, the filed data does not resolve which interpretation is correct.
- Commercial RPO: $678 billion, up 84% year-over-year. This is the pile of signed contracts for work not yet delivered, per the 10-K. It functions like a restaurant’s reservation book: the seats are booked, the revenue just hasn’t been cooked yet. At $678 billion, Microsoft’s reservation book is larger than the GDP of Switzerland. About 45% of it is tied to OpenAI commitments, which concentrates both the upside and the counterparty risk in one relationship.
- Q4 FY2026 revenue beat: $90.0 billion actual versus roughly $87.6 billion expected. Per Microsoft’s press release, the company delivered $2.4 billion more than consensus in a single quarter. For context, $2.4 billion is roughly what Microsoft’s entire LinkedIn business generates in a quarter. Beating by a full business unit’s worth of revenue is not a rounding error.
Sources
- Microsoft 10-K, fiscal year ended June 30, 2026 (SEC EDGAR)
- Microsoft Q4 FY2026 earnings call and investor materials
- Microsoft Q4 FY2026 press release and webcast
- Futurum Group: Microsoft Q4 FY2026 analysis
- Morningstar: Microsoft stock after earnings (Dan Romanoff, July 30, 2026)
- Stock Analysis: Microsoft consensus estimates
- Microsoft blog: OpenAI partnership announcement, January 23, 2023
- Microsoft 10-K, fiscal year ended June 30, 2024 (SEC EDGAR)
