Coverage update · Semicap · ASML Q3 2026
The Shortage Runs Through 2028
ASML beat the top of its own margin guidance, narrowed its full-year guidance upward, and said plainly that it will not be able to build enough machines to meet demand in either 2027 or 2028. The shares fell, and that decision is now ours to make rather than describe.
On October 14 ASML reported third-quarter net sales of €11.66 billion and a gross margin of 57.5% — above the top of its own 55–57% guidance. It narrowed full-year sales upward to €44–45 billion, from €43–45 billion and guided the December quarter to €14.2–15.2 billion, which would be the largest quarter in the company’s history. Then it said the thing that matters more than any of those numbers: its planned supply will not fully meet customer demand in 2027 or in 2028, and it is investigating capacity increases for 2028 beyond the 30% it has already committed to.1
The stock fell on the report and has kept falling. It traded near $1,647 on October 19, roughly 8% below where it sat when we last published work on it.3
That matters more than usual, because in July’s initiation we wrote a specific sentence: we would add on a semiconductor-wide pullback into $1,550–$1,650, and we would not chase. The stock entered that range this week. Because the single share held in July has since been sold, acting on that condition now would be re-entering the position rather than adding to it. Either way a published condition that goes live stops being analysis and becomes a decision, and our recent record is poor on exactly that transition.
1. What the quarter actually said
Three things, in descending order of durability.
The margin cleared its own ceiling. 57.5% against a 55–57% guide, up from 54.0% in the June quarter. Installed-base management revenue — service and upgrades on machines already in the field, the most predictable line in the business — came in at €3.3 billion, about €400 million above guidance.
The year was narrowed upward. Full-year sales of €44–45 billion, against €43–45 billion previously — the low end raised, the top unchanged — with a fourth-quarter guide of €14.2–15.2 billion. Systems shipped in the quarter: 121, of which 113 were new.
And the constraint moved from demand to supply. Management said planned capacity will not fully meet demand across both 2027 and 2028. For the sole supplier of extreme-ultraviolet lithography, that statement supports the demand outlook and suggests capacity rather than customer appetite is the binding constraint. What it implies for pricing and margin depends on product mix, contract terms and the cost of adding that capacity — none of which the release settles.
Chart 1: The Quarter Against Its Own Guidance
Sales mid-band, margin above the top of the range, full year raised.
One caution on the headline profit figure: net income of €4.96 billion included roughly €500 million from revaluing a stake in Mistral AI. That is a mark on an investment rather than cash from selling machines, and it should be set aside when judging operating performance.1
2. The shares went the other way
Nothing in the release obviously explains it, and the week that followed was the worst for the Nasdaq since June, driven by a repricing of the credit financing artificial-intelligence build-outs rather than by any change in orders — a subject we take up in this week’s outlook. ASML is among the most rate-sensitive assets in that complex: a business whose valuation depends heavily on cash flows many years out, discounted at a 30-year Treasury yield that touched a cycle high of 5.31% on October 5.
So the stock is down for reasons that have little to do with lithography and much to do with the discount rate and the mood around anything AI-adjacent. That is precisely the circumstance our July note described when it set an add range.
Chart 2: The Published Add Range, Now Live
We said we would add into $1,550–$1,650 and would not chase. The stock is there.
The target rests on a scenario tree across 2027–28 earnings, weighted and discounted; three updates since July have landed within 1.5% of each other because the operating assumptions moved very little. What has moved is the discount rate, which is the open question the risk section takes up rather than this chart.
3. The decision, and our record on these
Reaffirming accumulate at a target of about $2,000 is the easy half. The hard half is what we do when a published condition goes live, and the honest answer is that we have recently done nothing.
In August a buy gate opened on a power utility we had researched, published and authorized. No order was placed. We reported that failure in three consecutive notes before retiring the framework rather than publishing the same unexercised authorization a fourth time. A partial-trim order on another holding has now gone unplaced across three notes as well.
The range stands as published: $1,550–$1,650, on a semiconductor-wide pullback, without chasing. The stock is at the upper edge of it now. With no position currently held, a purchase here would be a re-entry at roughly the level we named three months ago.
Under our own trading policy we cannot act on a covered name inside the window around publishing a note on it, so we cannot place an order for five days after this note appears. That is a constraint we wrote for ourselves to stop research and trading from contaminating each other, and it cuts against us here.
What we will do is report the outcome either way in the next note: whether an order was entered, at what level, and if nothing was done, that too. The test is whether we apply the framework consistently once the window closes, not whether an order is placed regardless of how the facts then look.
| Test | Supports accumulate | Challenges it |
|---|---|---|
| Q4 results, January 27 | Net sales inside the €14.2–15.2B guide with gross margin at or above 56% — a point below the ~57% guided, which is the slippage we are prepared to absorb | Sales below €14.2B or margin below 56%, which would weaken confidence that the margin step up is durable rather than proving Q3 was the peak |
| The 2027 supply statement | Repeated or strengthened at the Q4 call | Any softening of the language about demand exceeding supply, which is the core of the pricing case |
| Hyperscaler 2027 capital spending, reported in the coming weeks | Guidance above 2026 levels | The largest buyers guiding 2027 spending flat or lower together — lithography orders follow that aggregate with a long lag, so one company moving is not enough |
| The long bond | Thirty-year yield holding below October’s 5.31% cycle high | A sustained move above it, which compresses the multiple on exactly this kind of long-duration cash flow |
An add range set in July was set at a different long-bond yield, and a monopoly is still a long-duration asset being repriced by rates rather than by its order book — the range may simply be too high for the regime we are now in. “Supply will not meet demand” is a statement about intentions two years out from a company with every incentive to make it, and the one hard number we do not have is bookings, which ASML stopped disclosing. And while the author does not own this name today, he did at the initiation and may buy it again once the window closes, so a reaffirmed positive rating is a conclusion to read with that in mind.
Sources
- ASML Q3 2026 results, released October 14, 2026: net sales €11.66 billion, gross margin 57.5%, net income €4.96 billion including approximately €500 million from the revaluation of a Mistral AI stake; 121 lithography systems sold, of which 113 new; net system sales €8.4 billion, comprising €4.1 billion EUV (including one High-NA system) and €4.3 billion non-EUV; installed base management sales €3.3 billion, about €400 million above guidance; Q4 guidance of €14.2–15.2 billion at a gross margin around 57%; full-year 2026 guidance raised to €44–45 billion at a gross margin around 56%; management statement that planned supply for 2027 and 2028 will not fully meet customer demand, with capacity increases for 2028 beyond the previously communicated 30% under investigation (Form 6-K press release; earnings call transcript). ↩
- Prior guidance, against which the quarter is measured: Q3 net sales €11.0–12.0 billion and gross margin 55–57%; full-year €43–45 billion — both issued with Q2 results on July 15, 2026. Rating, target and the $1,550–1,650 add range per our July 15 initiation, where the wording was that we would add on a semiconductor-wide pullback into that range and would not chase.
- Price of about $1,647 is October 19 intraday via the house quote service, which is levels-only and can lag. Ten-year and thirty-year Treasury context, including the October 5 cycle high, per the sourcing in this week’s outlook. ↩
- Next scheduled event: ASML Q4 2026 results, January 27, 2027, per the company's investor calendar. The Capital Markets Day at which ASML says it will update its longer-term model is June 10, 2027.
Methodology: company figures are quoted from ASML's own release and are stated in euros as reported; the share price is in dollars. Tests are graded against their published wording. Nothing here is a price target for any individual.
Nothing on this page is investment advice. We work in scenarios and trip-wires, not predictions. See disclaimer.
Disclosure: the author holds no position in ASML. A single share held at the time of the July initiation was sold before the research window for this note, so the accumulate rating below is not a view on a security currently owned. Readers should note that the author does hold other semiconductor and AI-infrastructure names discussed in general terms here. Under our own trading policy, no order in a covered name may be placed in the window around publication of a note on it. This note is research and opinion, published to all readers alike, and is not investment advice for any person. bpleon research received no compensation from ASML. See disclaimer.