Two years ago this company sold its chips for less than they cost to make. In the quarter to July it kept 85 cents of every revenue dollar as gross profit — with the same factories and the same costs.
A swing of 97 percentage points with the cost base unchanged. That cannot come from efficiency, from scale, or from a better product. There is only one thing it can be: the price went up.
The shares rose sixteenfold in a year. The question this article is about is whether what happened is a moat appearing, or a shortage being mistaken for one.
One of these closes itself
| A shortage | A moat | |
|---|---|---|
| What it is | Demand above capacity, right now | A reason rivals cannot supply |
| What high prices do to it | End it. The profits fund the new capacity | Nothing. There is no capacity to add |
| How long it lasts | Until the industry finishes building | Until the reason stops being true |
| What the margin does after | Returns, often below where it started | Holds |
| In our library | Gross margin −12% → 85% on unchanged costs | 67.7% gross margin, sustained, with 77% of sales on nodes with few rivals |
Read the second row twice, because it is the mechanism nobody plans for. A shortage is self-destroying. The extraordinary profits it throws off are precisely the thing that funds the capacity that ends it — and every competitor is reading the same price signal at the same time.
A real moat has the opposite property. High returns at the company that makes every advanced chip's lithography do not attract a rival, because a rival would have to rebuild four industries and then wait twenty years. The profit is safe because it cannot be competed away.
The profits a shortage produces are the money that ends it. That is the sentence that separates the two, and it is true in every commodity ever traded.
The contracts deserve to be taken seriously
The company's own answer to all of this is a new kind of long-term contract — backed by customers' cash, worth at least $94 billion at floor prices. That is not a brush-off, and our report said so: it deserves to be taken seriously.
If customers will commit cash years ahead at a floor, then some of the pricing is structural rather than cyclical, and the business has genuinely changed shape.
These contracts have never been tested in a falling market.
A floor price only matters when the spot price is below it. Until that happens, a long-term contract and a strong spot market are indistinguishable — the customer pays either way and everyone is happy.
What you want to know is what the customer does when the floor is 30% above what they could buy at today. That is the only test, it has not been run, and every claim about these contracts is a prediction about it.
Three defensible multiples, twenty-two times apart
| Valued on | Multiple | What it assumes |
|---|---|---|
| Next year's expected earnings | 8× | The shortage is the new normal |
| The average earnings of a whole cycle | ~30× | Good years and bad years both happen |
| The last cycle's earnings | 178× | The bad years look like the last bad years |
There is no arithmetic that settles this. Every number is correct. Choosing between them is the entire investment decision, and the one that makes the shares look cheap is the one that assumes the best year in a decade is the baseline.
Graham's rule for any cyclical was to use the earnings of a whole cycle, and it exists because the eight-times number is always available and always seductive at exactly the wrong moment.
Four questions that tell them apart
Split the margin improvement. Falling unit costs are durable and come from the company. Rising prices on unchanged costs come from the market, and the market takes them back. A 97-point swing on flat costs needs no further investigation.
Capital spending plans across all competitors are public. If the industry is adding capacity, you have the end date for the shortage roughly in hand — and it is usually two to three years out, which is comfortably inside most people's holding period.
Not whether it should, whether it has. A moat that has never been tested is a hypothesis. Look for the last bad year in the industry and check what this company's margin did in it.
Work out mid-cycle earnings before you look at the current multiple, so the eight-times figure does not anchor you. If it is still attractive on the average year, you have found something. If it is only attractive on the best year, you have found a shortage.
We scored this 5.4 and declined it — not because we think the business is bad, and not because we are confident the contracts will fail. We declined it because the range of defensible values spans twenty-two times and we cannot narrow it honestly.
That is a legitimate answer. A position you can only justify by picking the most generous of three correct numbers is a position you do not understand well enough to size.
One whose margin moved 97 points on price, and two whose margins have simply held.
Including the five claims to a moat, tested one by one.
Frequently asked
What is the difference between a shortage and a moat?
A shortage is a gap between demand and capacity, and it closes — because the high prices it produces are themselves the signal that brings new capacity. A moat is a reason competitors cannot supply at all. The test is simple: ask what happens when the industry finishes building. A shortage ends there; a moat is unaffected.
How can a gross margin go from negative to 85%?
By price alone. SanDisk sold memory below cost two years ago and kept 85 cents of every revenue dollar in the quarter to July — with the same factories and substantially the same costs. A 97-point swing with unchanged costs cannot come from efficiency, scale or product. It is the price of the commodity moving, and commodity prices move both ways.
Do long-term contracts create a moat?
They create a cushion, and they are worth taking seriously — SanDisk's are backed by customers' cash and worth at least $94 billion at floor prices. But a contract is only as good as the counterparty's willingness to honour it when the spot price falls below the floor, and these have never been tested in a falling market. A promise made in a shortage is tested in a glut.
How should you value a company in a shortage?
Not on the shortage year. The same company was 8 times next year's expected earnings, about 30 times the average earnings of a whole cycle, and 178 times the last cycle's — a 22-fold spread between three defensible numbers. Which one you use is the entire investment decision, and the cheapest one is the one that assumes the good times are the normal times.