A Double in a Quarter That Changes Nothing

Our epi wafer names are up one hundred percent this the past 6 weeks

Here is the alpha, stated plainly before I earn it.

Our epi wafer names are up roughly one hundred percent this quarter, and the move so far reflects only the demand the market can already see. The repricing I actually underwrote has not happened. AI specific chips are still less than half of one percent of global wafer starts, yet they already generate close to a fifth of all semiconductor revenue. That is the widest gap between volume and value in the history of the industry. It sits in plain sight on numbers anyone can pull.

The supply side cannot answer it on any useful clock. A new wafer line takes two to three years to make its first qualified wafer, then twelve to eighteen months more to pass customer qualification. Industry capex peaked in 2023 and has fallen sharply since, and the teams who would need to expand have guided lower, not higher. That means the supply cap through 2027 is already fixed. It cannot be argued with. It can only be waited out.

So the demand is visible, the supply is capped, and the two are set to collide on a known schedule. The suppliers who grow the epi layer sell under multi year contracts. The ones being written now are being written into that cap. When the loose market pricing of the last cycle resets into a tight market across 2027 and 2028, the revenue steps up on capacity that already exists. That reset is the catalyst. It has not fired. The double you are looking at is the market reacting to page one. The repricing is further down the document.

That is the whole edge, and I have now given it to you for free. The rest of this letter is why I am not selling into a double, which is a harder discipline than buying was.

A position that doubles in three months is usually the market telling you to sell. The margin of safety that justified the entry has narrowed. The distance between price and value has closed. Discipline says harvest the crop while it is ripe.

That instinct is correct, and I want to honor it before I explain why I am not acting on it here.

This spring I wrote about the epitaxial wafer layer. It is the thin crystalline surface grown on top of the polished silicon disc, the few microns where every transistor in an advanced chip actually lives. I compared it to the ice on a rink. The concrete slab underneath holds everything up, but the skating happens on the inch of ice laid down on top, and a single defect in that layer makes the rink unusable for a competition. The wafer is the slab. The epi is the ice. The function lives in the layer on top, and only a small handful of companies in the world can grow it to the tolerance that three nanometres and below demand.

I did not write the piece expecting a double in a single quarter. I wrote it because the arithmetic was plain and the market was not looking at it. That disproportion between volume and value tells you, in clean numbers, where the marginal dollar is being earned and where the binding constraint sits.

So what changed to move the price?

Nothing in the supply picture. That is the part worth sitting with. A wafer line still takes two to three years to produce its first qualified wafer. The reactors that grow the epi layer still carry lead times measured in many months. None of that moved. A decision to expand made today does not produce a sellable wafer before the second half of 2027 at the earliest.

What changed is that the market started reading. The value was in the constraint the entire time. The double is not the market pricing the constraint. It is the market noticing the constraint exists.

I have watched this pattern before, in a different corner of the same stack. When the price finally moves on a business the crowd had ignored, it moves quickly, and the speed fools people into thinking the value was created in the move. It was not. The real work happened long before the reframe, in the quiet stretch when the position looked like dead money and the thesis lived only in the arithmetic.

Now to the question a double actually raises. Does the margin of safety still exist?

This is where I try to be honest rather than attached. A double does two things at once. It vindicates the underwriting, and it narrows the cushion. Those are not the same, and confusing them is how a good investor turns a sound thesis into a sentimental one. I sold a battery storage position two years ago when it ran from eight dollars toward twenty, not because I stopped believing in the business, but because the valuation had already reflected what the market once ignored. The crop was ripe. A farmer who refuses to harvest because he is fond of his field is not patient. He is sentimental.

So I ask the only question that matters here. Is the catalyst spent, or is it still ahead?

It is still ahead. The double this quarter is the market reacting to the demand picture it can already see. The event I underwrote is the contract reset that runs through 2027 and into 2028. That reset has not happened yet. The margin of safety narrowed. The catalyst did not fire. Those are different facts, and only the second one governs whether I hold.

So I hold. Not because the position went up. Positions that go up are the easiest ones to talk yourself out of, because a double feels like permission to declare victory and move on. I hold because the reason I bought has not been consumed by the price. The market read the first page. It reacted to what was written there, and what was written there was only the demand. The supply cap, and the repricing it forces, is still further down the document.

I do not know what the names do next quarter. I never do. Volatility after a double is normal, and I would not be surprised by a give-back that makes the last three months look foolish for a season. That is the market having seasons. It changes nothing about the aquifer underneath.

A double is not the end of a thesis when the market has only read the first page.

Neel Khokhani, Founder and CEO Epochal Corporation, @neel_epochal