Cisco and Microsoft both kept growing after 2000. Their shareholders waited 17 and 26 years to get back to even.
That is the sentence to keep. Not "the bubble burst" — the companies did fine. Microsoft is one of the great businesses of the century and anyone who bought it in 2000 spent a quarter of a century getting their money back.
The punishment for overpaying is almost never a collapse. It is time, and time is the one thing you cannot get a refund on.
What $190 was actually assuming
Our October report on Palantir found, on everything measurable, one of the best businesses we have studied. Revenue grew 93% in the quarter, and the growth rate had risen for eight quarters running — something almost no company of that size has ever done. It keeps 85 cents of every dollar as gross profit, needs almost no capital, owes nothing, and customers spend 157% of what they spent a year earlier.
We passed. At $190 the shares cost about 73 times sales.
| To earn this, from $190 | Revenue must compound at | Which means sales grow |
|---|---|---|
| 10% a year for a decade | ~32% a year | 16.1× |
| 7% a year | ~27% a year | 11.0× |
| 5% a year | ~24% a year | 8.6× |
Run the price backwards and the argument changes shape. It is no longer 'will this company do well?' but 'has a company this size ever done that?'
What breaking even actually costs
"Waited 17 years to break even" sounds survivable. Price it.
A government bond paying 5.12% turns £1 into £2.34 over 17 years and £3.66 over 26. So the Cisco shareholder who got back to even did not lose nothing — they gave up more than doubling their money, risk-free, while being entirely right about the company.
That is the real bill, and it never appears on a statement. There is no line item for the return you did not take.
| Years to break even | Your return | What safe money did | The gap |
|---|---|---|---|
| 17 (Cisco) | 0.0% a year | 2.34× | −134% |
| 26 (Microsoft) | 0.0% a year | 3.66× | −266% |
| 10 | 0.0% a year | 1.65× | −65% |
And being too careful has a bill too
In June we called Microsoft a wonderful business at a fair price and then asked for a bargain at $310. Two days later the shares touched $349.20 — inside our own estimate of value — and we were still waiting. They were $497.93 when we wrote the follow-up.
So this article is not an argument for never paying up. Demanding a bargain is itself a decision with a cost, and it is paid just as silently.
The difference is the size of the mistake. Being 15% too strict costs you a good investment. Being 73 times sales too generous costs you a decade.
A price is a prediction with a number attached
Our report named a price: begin near $110. From $190 that is a 42% fall with nothing whatever going wrong at the company — same growth, same margins, same customers, a different multiple.
That is what a valuation discipline looks like in practice, and it is uncomfortable precisely because it has no connection to the news. You are not waiting for the business to stumble. You are waiting for other people to want it less.
Sometimes they never do. That is the risk you accept in exchange for not spending seventeen years getting back to where you started.
Three habits
Take today's price as given and work out what the business must do for you to earn a decent return. Write that number down as a sentence — 'revenue must grow sixteenfold in ten years' — and then ask how often that has happened.
A decade of flat performance is not a neutral outcome while a bond pays 5.12%. Multiply it out before you buy; the number is usually larger than people expect and it changes how patient you feel.
Keep two lists. Businesses you admire, and businesses you would buy today. Most of the best things we have examined sit on the first list only, and writing a starting price next to each one is what turns admiration into a plan.
We have twice declined to judge the most prominent name in this market with a verdict of 'too hard'. It has gone up. That is the cost of a discipline, and a discipline you abandon when it is expensive was never one.
One we passed on at 73 times sales, one we were too strict about, and one we have twice declined to judge.
Including the price run backwards, and our own call graded in public.
Frequently asked
What happens if you overpay for a good company?
You usually still own a good company — and you wait. Cisco and Microsoft both kept growing after 2000, and their shareholders took 17 and 26 years respectively to get back to even. Nothing went wrong with either business. The entire loss was in the price paid, and it was settled in time rather than in a bankruptcy.
Is Palantir overvalued at 73 times sales?
Our October 2026 report concluded yes at $190, while calling it one of the best businesses we have studied — 93% revenue growth, 85 cents of gross profit per dollar, no debt, net dollar retention of 157%. The objection is arithmetic: at that price, revenue has to compound at roughly 32% a year for a decade just for a buyer to earn 10% a year. That is a sixteenfold increase in sales.
How do you work out what a share price is assuming?
Run it backwards. Instead of forecasting and comparing to the price, take the price as given and solve for the growth required to produce an acceptable return. It converts an argument about opinion into a question about plausibility: not "will this company do well?" but "has any company of this size ever grown sixteenfold in ten years?"
What is the real cost of waiting to break even?
Whatever safe money would have earned meanwhile. At a 5.12% Treasury, 17 years turns £1 into about £2.34 and 26 years into about £3.66. So a shareholder who merely broke even over those periods did not lose nothing — they gave up more than doubling, risk-free, while being completely right about the business.