In September 2016, Microsoft was the boring pick. Windows was mature, the phone business had just been written off, and the interesting money was chasing companies that did not yet make any. Buying Microsoft stock was the thing you did when you did not want to think about it.
A $1,000 position taken then, and never touched, would be worth roughly $8,900 now. Reinvest the dividends along the way and it lands closer to $9,900.
That is the headline, and it is accurate. It is also the least interesting part of the story, because the arithmetic is easy and the behavior it describes is not.
The short version
- Microsoft averaged about $56 a share in September 2016. It trades near $494 now
- $1,000 bought roughly 18 shares, worth about $8,900 today on price alone
- With dividends reinvested, the figure is closer to $9,900, or about 26% a year compounded
- The S&P 500 over the same decade turned $1,000 into roughly $3,400
- The engine was Azure and then AI, not Windows. Azure grew 40% in the most recent quarter reported
- Microsoft is now worth about $3.68 trillion and plans roughly $190 billion of capital spending this year
- None of this is a forecast, and the tax bill on a gain like that is not small
The arithmetic, laid out
There is nothing clever here, which is part of the point. The whole result comes from two prices and ten years of not selling.
| Line | Figure |
|---|---|
| Average share price, September 2016 | about $56 |
| Shares bought with $1,000 | roughly 18 |
| Share price, September 2026 | about $494 |
| Position value, price only | about $8,900 |
| Position value, dividends reinvested | about $9,900 |
| Compound annual return, with dividends | roughly 26% |
| Same $1,000 in the S&P 500 | roughly $3,400 |
The dividend line is the one people skip, and over a decade it is worth about a thousand dollars on its own. Microsoft’s yield is small, under 1% at today’s price, which makes it look irrelevant on any given year. Reinvested and compounded for ten, it stops being irrelevant.
Against the index
Beating the market is the part that matters, because the alternative was not a shoebox. It was a low cost index fund that anybody could have bought on the same afternoon with less thought and less risk.
Doubling the index’s annual rate over a decade produces almost three times the money. That is compounding doing what it does, and it is the reason a few percentage points of annual return are worth arguing about.
What actually drove it
It is tempting to tell this as a story about Windows, because Windows is what most people picture when they picture Microsoft. Windows had very little to do with it.
The decade belonged to enterprise cloud, and then to the AI build out on top of it. The figures from Microsoft’s most recent reported quarter give the shape of it.
The engine, in four numbers
- Azure grew about 40% year over year in the most recent quarter reported, ahead of guidance
- The AI revenue run rate passed $37 billion, up around 123% on the year
- Planned capital spending of roughly $190 billion for the year, a very large step up
- Market capitalization of about $3.68 trillion
That capex figure is the one to sit with, because it is where the next ten years get decided. Microsoft is spending at a rate that only makes sense if demand for AI compute keeps climbing. If it does, the spending looks like the Azure build out of 2016 and the story repeats. If it does not, it looks like the most expensive misread in corporate history.
This is also a company that has been unusually deliberate about how it talks about the technology it is spending all that money on. Microsoft went as far as writing a rule telling its own AI never to resist being switched off, which is either sincere governance or extremely well judged positioning, and possibly both.
Three things the “if you had invested” post never mentions
Posts like this one are fun and slightly dishonest by construction, so here is the part that gets left out.
You had to sit through the drops. A 26% average annual return is an average. It was not delivered in even slices. There were stretches in this decade where Microsoft fell hard enough that holding felt like stubbornness rather than conviction, and the whole result depends on having done nothing during those months. The arithmetic is easy. Doing nothing while a position falls 30% is not.
You had to pick this company, in 2016, out of everything else. We only tell this story about the winners. The same $1,000 could have gone into any number of names that looked equally sensible at the time and are worth a fraction of it now. Picking Microsoft in hindsight costs nothing. Picking it in advance, and then not being tempted by anything else for ten years, is the actual skill being described.
Concentration is the risk nobody prices. Every adviser in the world would have told you not to put a meaningful share of your savings into one company, however good it looked, and they would have been right to. The $1,000 story works because $1,000 is small. The same decision at a scale that would have changed your life is a different decision entirely, and it is the one that occasionally ends badly.
If you want the broader version of this argument, the debate about which pieces of standard financial advice actually hold up has been unusually lively this year, and we went through which of the claims survive contact with the data when a White House technology adviser took a swing at three of them.
Then there is the tax bill
The $8,900 is not yours. Some meaningful portion of the gain belongs to the government the moment you sell, and the exact portion depends on your income, your state and how the position is held.
Held for over a year, this is a long term capital gain, which is the good outcome. It is still a gain of roughly $7,900 on a $1,000 cost basis, and selling it all in a single tax year can push you into a higher bracket for that gain than you would face if you unwound it over two. This is the sort of thing that is trivially easy to plan for in advance and genuinely expensive to discover in April, and it is worth understanding how the filing software handles investment income and what the various tiers actually cover before you are staring at a form.
Inside a retirement account the picture changes completely, which is a large part of why financial planners talk about account types more than they talk about stock picks.
Doing this again, for 2036
The question underneath every one of these posts is whether you should buy it now. Nobody can answer that, but it is possible to be precise about what the bet would be.
| What worked in 2016 | What is different in 2026 |
|---|---|
| A $450 billion company with room to grow into cloud | A $3.68 trillion company. The same multiple would make it larger than most economies |
| Azure was a challenger with obvious upside | Azure is a leader, and leaders compound more slowly |
| Capital spending was ordinary | Capital spending is roughly $190 billion a year and rising |
| Expectations were low | Expectations are priced in, which is what makes disappointments expensive |
Repeating an 8.9x from a $3.68 trillion base would require the company to reach a size that has never existed. That does not make it a bad investment. It makes it a different one, and anyone quoting the last decade’s return as a forecast for the next is selling something.
This is a look at historical figures, not investment advice. Prices quoted are approximate and move constantly. Past performance tells you what happened, not what will.
The genuinely useful lesson in the Microsoft number is not about Microsoft. It is that the winning position over that decade was dull, widely available, and required no special access or insight to buy. What it required was ten years of leaving it alone, including the parts where leaving it alone felt stupid.
That is the rarest thing in investing, and it is the one part of the story that cannot be bought after the fact.
