MSCompany report
Microsoft Corporation MSFT
The bet you're really making is that companies everywhere keep moving their software and data onto Microsoft's cloud, and now rent its AI to run on top. You're betting the orders already signed, $678 billion of them, turn into real sales as the new data centers switch on. Right now it is going well: sales up 18% and profit up 31% last year, though the cash left over after building those data centers actually shrank. You pay about 28 times last year's earnings, below the historical average and a little under its rivals.
Key data
MSFT · price with moving averages
Source: market data.
The business
Microsoft sells three things that all lead to the same place. It rents computing power and AI services by the hour on Azure, its cloud. It sells Office, now Microsoft 365, and its business apps as yearly subscriptions to nearly every large company on earth. And it owns Windows, LinkedIn, Xbox and the Copilot assistants stitched across all of it. The money is made twice over: a company signs a multi-year cloud contract, then spends more each year as its data and workloads deepen, which is expensive and painful to unwind once the business runs on it. That switching cost is the moat, and it is why a customer who joins rarely leaves. Beginning in fiscal 2027 the company will report as just two segments, Agents and Infra and Devices and Consumer, a tell that management now views the whole firm as AI infrastructure with everything else attached. The concrete thing a customer holds is an Azure bill and an Office login the entire organization already depends on.
The numbers
Growth is accelerating off an already enormous base, which is rare at this size.
| Quarter | Revenue | Net income | Diluted EPS |
|---|---|---|---|
| Q3 FY25 (Mar '25) | $70.1B | $25.8B | $3.46 |
| Q1 FY26 (Sep '25) | $77.7B | $27.7B | $3.72 |
| Q2 FY26 (Dec '25) | $81.3B | $38.5B | $5.16 |
| Q3 FY26 (Mar '26) | $82.9B | $31.8B | $4.27 |
| Q4 FY26 (Jun '26) | $90.0B | $35.8B | $4.80 |
Revenue climbed every quarter to a record $90.0B in the June quarter. The December quarter's $5.16 per share was elevated versus the $4.27 to $4.80 range of the surrounding quarters. Earnings have landed above Wall Street's estimate for four quarters running.
| Fiscal year | Revenue | Net income | Diluted EPS |
|---|---|---|---|
| FY2022 | $198.3B | $72.7B | $9.65 |
| FY2023 | $211.9B | $72.4B | $9.68 |
| FY2024 | $245.1B | $88.1B | $11.80 |
| FY2025 | $281.7B | $101.8B | $13.64 |
| FY2026 | $331.8B | $133.7B | $17.95 |
Earnings per share compounded about 17% a year since FY2022, and faster lately: FY26 revenue rose 18% and net income 31%. The cash tells the harder story. Operating cash flow rose 34% to $182.9B, yet capital spending jumped 80% to $115.9B, so free cash flow actually fell to $67.0B from $71.6B a year earlier. A concern flagged days ago, that free cash flow keeps sliding while capex keeps climbing, held true this year: the buildout is still eating more cash than it frees.
| Cloud engine, FY26 | Growth (y/y) |
|---|---|
| Microsoft Cloud revenue | +27% |
| Azure and other cloud | +41% |
| M365 Commercial cloud | +17% |
| M365 Consumer cloud | +28% |
The order book is the real tell. Commercial remaining performance obligation grew 84% to $678 billion, roughly four times a single year's revenue and far ahead of the 18% top line. That gap is the earliest sign that demand already signed has not yet reached the income statement. The thing this memo believes the market underweights is that this backlog converts and re-accelerates cloud revenue as capacity comes online; the print that settles it is Microsoft Cloud growth holding in the high-20s or better as capex peaks.
Management
Insiders are net sellers by a wide margin: 15 sales worth $66.2M over the past year against a single $2.0M purchase. CEO Nadella sold $28.9M across two transactions on September 1, and the vendor feed does not carry 10b5-1 status, so plan status is not disclosed on these; read the Form 4 footnotes before drawing intent from the timing. Capital allocation is disciplined and expensive: the company bought back $22.3B of stock in FY26 at roughly 28 times earnings, a full price to pay for its own shares. Returns on capital stay elite, with return on equity near 33% and return on invested capital near 21%, so even the AI buildout is being funded off a business that still earns far above its cost of capital. Full CEO compensation detail is not in this data.
How it fails or surprises you
The backlog lands and cloud re-accelerates (right tail). $678B of signed commitments, up 84%, sit well ahead of 18% revenue. If Azure capacity comes online and converts even partway, cloud growth reaccelerates past the high-20s and the 28x multiple looks cheap in hindsight. The market prices the trailing number, not full conversion. Watch Microsoft Cloud revenue growth over the next two quarters.
The buildout stops paying in cash. Capex jumped 80% to $115.9B and free cash flow fell despite a 34% rise in operating cash. This is the fact the case for owning it explains least well. If cloud growth cools while depreciation from these assets floods the P&L, both margins and free cash compress at once. Watch capex still rising year over year alongside a falling Microsoft Cloud gross margin.
Seats stop growing as AI reprices Office. M365 Commercial cloud grew 17% on more seats at higher prices. If Copilot adoption plateaus or AI lets customers do more with fewer seats, that engine decelerates quietly. Watch Microsoft 365 commercial seat growth, disclosed each quarter, rolling toward the mid-single digits.
Closing thoughts
The question is whether the $678B backlog converts fast enough to justify the spending. Watch Microsoft Cloud revenue growth: if it holds in the high-20s while capex peaks this year, the buildout is working and 28 times is cheap. If it slides toward 20% while spending keeps climbing, demand is not keeping up and the multiple will compress. The deeper bet is that AI is real infrastructure, not a cycle that reverses, a $3.7 trillion company compounding earnings at 17% needs that to be true. The balance sheet can survive being wrong, but the upside is bigger: the backlog is signed commitments, not forecasts, so if it converts, cloud re-accelerates and earnings estimates rise. The risk is another year of falling free cash with rising capex; that would force the market to reprice earnings quality, not just growth.
The bet is still that companies keep moving their software and data onto Microsoft's cloud and now rent its AI to run on top, and the orders already signed turn into real sales as the new data centers switch on. It breaks if free cash flow keeps falling while capex keeps rising and cloud growth cools at the same time. The one pair to watch is Microsoft Cloud revenue growth against capital spending: if the first is decelerating a year from now while the second is still climbing, the buildout is not paying and the premium is no longer earned.
Methodology
Sector frame: enterprise software and cloud infrastructure, judged on repeatable operating earnings with the AI capex cycle normalized against depreciation.
Data gaps: Azure is management commentary, not an audited segment revenue line; the Q4 FY26 quarter is derived from the FY26 10-K less the first three filed quarters; the pack carries no clean near-year forward EPS, so the valuation box shows trailing multiples; full CEO compensation and 10b5-1 status on recent insider sales are not in this data.
Bundle: as-filed fiscal 2026 results, period ended June 30, 2026, income, cash-flow, and balance-sheet figures taken as reported; capex includes finance leases.
Sources: SEC-filed fiscal 2026 10-K (filed July 29, 2026) and the September 2, 2026 8-K segment presentation; price, range, and multiples are market data as of September 6, 2026.
Fact check: All numerical claims reconciled to filed XBRL and 10-K text. Q4 FY26 revenue ($90.0B), net income ($35.8B), and EPS ($4.80) derived from FY26 annual totals less Q1-Q3 filed quarters. FCF figures ($67.0B FY26, $71.6B FY25) derived as operating cash flow less capex from filed cash-flow statements. Capex growth (80%) and operating cash flow growth (34%) calculated from filed annual figures. Valuation claim revised: current P/E of 27.8x is at the 25th percentile of 12-year history (below historical average), not the middle. All other claims verified against as-filed data. Final analysis verified as of Sep 6, 2026.
Bid Cap
Daily ideas, a 390-name database, and a model long/short book from an investor who mostly covers financials. $70 a month or $700 a year.
Subscribe on Substack


