What it does · how it got here · where it operates
Let me start where I always start: can I explain how this company makes money to a ten-year-old? With Microsoft, I can. It sells the software the world's offices run on — the documents, the spreadsheets, the email, the meetings — and, increasingly, it rents out the computing power other companies build their own businesses upon. People pay every single month, whether the economy is booming or limping. That is a beautiful thing to own.
It was started in 1975 by two young men, Bill Gates and Paul Allen, with a single idea: a computer on every desk. For twenty-five years that meant Windows and Office — a near-monopoly so durable it printed money. Then the world moved to the internet, and for a while Microsoft looked like yesterday's company. What happened next is the part I admire most: under a new manager, Satya Nadella, it reinvented itself around the cloud — renting computing by the hour through Azure — and turned one-time software sales into recurring subscriptions. By the year ended June 2025 it booked $281.7 billion in revenue and $101.8 billion in profit — and is worth about $2.7 trillion in the market. It is rare to watch a giant change its spots so well.
"The single most important decision in evaluating a business is pricing power. Microsoft can raise the price of Office, and almost nobody leaves." — the test I apply here
Today it operates in over 190 countries with roughly ~228,000 employees. It has no stores to speak of — the product is digital, delivered over the wire — so it carries almost no inventory and earns extraordinary margins (gross margin 68.8%). Note the date and the price: at $367.34, the stock sits roughly a quarter below the ~$497 it fetched a year ago. The business kept growing; the price came down. That combination is always worth a closer look.
| Founded | 1975 · Redmond, Washington (USA) · IPO 1986 |
| Sector / Industry | Technology · Software — Infrastructure |
| Employees | ~228,000 (approx.) |
| Countries of operation | 190+ |
| Revenue (FY2025) | $281.7 B |
| Market capitalisation | ~$2.73 T |
How the business actually works — and whether you'd understand it
"Never invest in a business you cannot understand." It is the first rule, and the one most often broken. So let me walk this business from end to end — the way I'd explain a cork company: it rents and plants the land, strips the bark from the oaks, mills it into stoppers, and sells them to the people who bottle wine. Simple enough for a child. Here is Microsoft's chain:
This is the honest part. I avoided technology for most of my life precisely because I could not say where it would stand in ten years — and I once misjudged IBM badly for exactly that reason. Microsoft passes my franchise test: a product that is wanted, that customers believe has no close substitute, and whose price isn't set by a regulator. But if you cannot explain Azure to a friend, that isn't a failing; it's a signal to stay within your own circle.
Revenue segments · geographic mix · the size of the pond
A business is only as good as the engine underneath it. Microsoft runs on three engines (FY2025), and what I like is that they pull in the same direction — each makes the others harder to leave.
This matters because a great company in a shrinking pond eventually drowns; a great company in a growing one rides the current. Microsoft sells into two of the largest, fastest-growing markets on earth — and takes a leading share of both. Its customers are mostly other businesses: from a two-person startup on Azure to the world's largest banks and governments, all on multi-year contracts they renew almost without thinking. That is exactly the customer I want — one who keeps paying, and barely notices a price increase.
The shape of the fortress · how durable · where it could crack
I think of a great business as an economic castle protected by a wide, deep moat. Microsoft's is among the widest in the world, filled from three springs that feed one another:
Switching costs. Ripping out Office or Azure means migrating data, retraining people, rebuilding integrations. Most firms simply won't — the same high-durability moat as the Apple ecosystem.
Network effects. Teams, LinkedIn and the developer ecosystem grow more valuable with every user that joins.
Scale & intangibles. Data-centre spending few rivals can match, plus a brand institutions trust with their crown jewels.
How durable? Very — the franchise should hold for a decade or more. The danger isn't that it vanishes; it's that artificial intelligence rewrites the rules of productivity and cloud. For now Microsoft sits on the winning side of that wave — but I'll be watching the trend, because moats are eroded long before they fall, and gross margins are the first place erosion shows up.
Who runs it · their record · whose money is on the line
I want businesses run by honest, able people who think like owners. So before another ratio, I ask: who steers this ship, and do they have their own money aboard?
What I weigh most is the behaviour of those in charge: do they buy back stock when it's cheap and pause when it's dear? Do they speak plainly to owners? Microsoft's record — a steadily rising dividend at a modest 21% payout, leaving plenty to reinvest, alongside consistent buybacks — is the behaviour of managers who respect the people they work for. [Ownership detail to be completed once the ownership endpoints are pulled.]
Sourced from your endpoints · TTM unless noted · linked to the data pages
| Metric | Value | Read |
|---|---|---|
| Revenue growth (5-yr CAGR) | ~14.5% | ▲ solid |
| Net income (FY2016 → FY2025) | $20.5B → $101.8B | ▲ ~5× |
| Gross margin | 68.8% | ▲ exceptional |
| Operating margin | 46.8% | ▲ expanding |
| Return on equity (ROE) | 33.1% | ▲ thoroughbred |
| Return on invested capital (ROIC) | 21.3% | ▲ above cost of capital |
| Net debt / EBITDA | 0.12x | ▼ fortress (Altman-Z 7.9) |
| Interest coverage | 52.7x | ▲ no leverage risk |
| Free cash flow / yield | ~$73B · 2.7% | ◆ capex-heavy (see valuation) |
| Capex / revenue | ~30% | ▲ the AI build |
These are the financials of a thoroughbred. A company earning 33% on equity with almost no debt is, quite simply, a money-making machine — and an Altman-Z of 7.9 means it could weather almost any storm. There is one asterisk I'll return to: free cash flow (~$73B) is far below reported profit (~$102B+), because Microsoft is pouring roughly 30 cents of every revenue dollar into new data centres for AI. Whether that is brilliance or an arms race is the whole valuation question.
Two honest yardsticks · DCF vs. the Street · margin of safety
| Yardstick | Today | ~1-yr-ago | Forward | Read |
|---|---|---|---|---|
| P/E — reported earnings | 21.8x | ~36x | ~22x (FY26) · ~19x (FY27) | reasonable for the growth |
| Price / Owner Earnings ◆ | ~30x | — | — | the honest middle |
| P/FCF — free cash flow | 37.4x | — | — | harshest — all capex counted |
| EV/EBITDA | 13.7x | — | — | fair |
| PEG (trailing) | 0.73 | — | — | growth-adjusted, attractive |
A company spending ~30% of revenue on data centres deserves a careful hand. Free cash flow subtracts every dollar of that capex — as if it were all just to keep the lights on. But a large share isn't maintenance; it is growth capex, building capacity for revenue that doesn't exist yet. Owner earnings — Buffett's measure, which leaves the growth spending out — lands near $12 a share, putting the stock at ~30×: full, but not absurd for this quality and high-teens growth. For an AI-era business, owner earnings is the fairer lens than free cash flow — and the whole question is whether that growth capex earns a high return. If it does, ~30× will look cheap in hindsight.
Regulation · litigation · business risk — with sources
The live overhang is legal: multiple firms have filed a securities fraud class action covering buyers between 1 May 2025 and 28 Jan 2026 — the likely reason the stock fell from ~$497 to ~$367. For an owner, the deeper question is the ~30%-of-revenue capex: if AI/cloud returns disappoint, free cash flow stays suppressed and today's price looks dear. These are real risks, but — short of the lawsuit revealing genuine wrongdoing — they read as pressure on returns, not destruction of the franchise.
Dear shareholder — having walked through Microsoft from top to bottom, let me set down plainly what I think, the way I would if we owned it together and met once a year over a Cherry Coke.
First, the business. This is the kind of enterprise I dream of owning forever: it sells the tools the working world cannot put down, and it collects the rent every single month. The moat is among the widest I know — switching costs, network effects, and a scale few can dream of matching, the three feeding one another. The people are first-class: Mr. Nadella performed one of the great turnarounds of our age, and he and Mrs. Hood return capital with a discipline I admire — a dividend lifted year after year at a modest 21% payout, with plenty kept back to reinvest. And the numbers are a thoroughbred's: 33% on equity, 69-cent gross margins, an Altman-Z of nearly eight, and so little debt that a recession would barely make it blink. On the quality of this company, I have not a single quarrel.
So where is the rub? Two places. The first is honest accounting of cash. Reported profit is over $100 billion, but free cash flow is nearer $73 billion, because Microsoft is plowing close to thirty cents of every revenue dollar into data centres for artificial intelligence. Now, one must be fair to the company: much of that is growth capex — money spent to build tomorrow's revenue, not to keep today's lights on — so the harsh thirty-seven-times-cash figure overstates the burden. The honest middle is owner earnings, which counts only the capex needed to maintain the business: by that measure the stock trades near thirty times. So — cheap at twenty-two times earnings, fair at thirty times owner earnings, dear at thirty-seven times cash. Which proves real depends entirely on whether that vast growth spending earns a fat return, or merely keeps Microsoft in an arms race where everyone builds and no one profits. I do not yet know, and a man should be honest about what he does not know.
The second is price and the cushion behind it. A year ago this sold for $497 and thirty-six times earnings — a price that, as I've long warned, can undo a decade of fine business performance. Today, at $367, much of that froth is gone, and the fear of a lawsuit has done me the favor of marking it down further. That is the right direction. But a careful man notices that a conservative appraisal lands near $300, while Wall Street dreams of $552 — and when the estimates of value sit that far apart, the margin of safety is thin. For a business this certain I want a 20–30% cushion, and I do not quite have it here.
What, then, will I do? Were Microsoft already ours, I would not dream of selling a jewel over a few points of multiple — wonderful businesses are far too scarce to part with lightly. But with fresh money today, I would keep watching, finger on the trigger. The price has come to me; it has not yet come far enough. Let Mr. Market keep fretting about lawsuits and capex, and somewhere in the low $300s I would begin to buy in earnest — and below that, where the cushion truly opens, I would back up the truck. Patience, here, is not idleness. It is the position.