TL;DR: ASML's October 14 results highlight revenue from its installed fleet as well as sales of new lithography systems. Productivity upgrades can add useful capacity without another complete machine, but service revenue is not a direct measure of additional wafers. Read the financial category, the technical intervention, and the customer's realized output separately. The distinction matters when assessing both the quarter's margin and the industry's capacity plans.
The question raised by ASML's latest quarter is more interesting than how many new machines left the factory. Customers also buy work that improves machines already installed. An upgrade can change productivity; maintenance can preserve availability; other services can support continuing operation. Their economic value is real, while their contributions to output differ.
ASML reported third-quarter sales of €11.662 billion, including €3.289 billion of Installed Base Management sales, versus €9.326 billion and €2.762 billion respectively in the prior quarter. Its release defines Installed Base Management as net service and field-option sales and attributes stronger-than-expected performance in that category to demand for productivity upgrades. These are company-reported, unaudited quarterly figures as of the October 28 cutoff. ASML's Q3 2026 release
The review below combines a simple financial reconciliation with an explicitly fictional fab upgrade decision. It does not estimate an undisclosed segment margin, forecast a share price, or claim that the invented tool performance describes an ASML product.
The first bridge: absolute growth and share of revenue
Start with the reported category rather than treating every euro as a new-system sale. Installed Base Management increased by €527 million between the two quarters. Total sales increased by €2.336 billion. Both grew, but they did not grow at the same percentage rate.
The installed-base category represented approximately 28.2 percent of third-quarter sales, compared with about 29.6 percent in the previous quarter. Its absolute revenue rose while its share of the larger total fell. That combination is entirely coherent. A mix percentage and a growth amount answer different questions.
Subtracting the category from total sales gives €8.373 billion for the remainder in the third quarter and €6.564 billion in the second. This arithmetic is useful for reconciliation. It should not be promoted into a new official segment or a claim about its standalone profitability. The subtraction preserves the reported boundary; it does not create disclosures the company did not provide.
An investor can now describe the quarter more accurately: the installed fleet generated more reported revenue, while other sales grew faster in aggregate. The explanation still needs product mix, timing, and management's discussion. A rising category does not by itself establish which intervention generated every additional euro.
The financial label contains several economic jobs
A service category can include activities that maintain existing performance and activities that improve it. A repair restoring an unavailable machine has a different physical interpretation from an upgrade that increases its processing rate. Both can be valuable to the customer and contribute to the same reported category.
ASML's customer-support material describes maintenance planning, diagnostics, performance optimization, upgrades, and training. It provides a useful picture of the operating relationship around an installed machine. It does not give an outside reader a complete revenue or margin split among those activities. ASML: Customer support
The analysis should therefore avoid a convenient but unsupported translation: installed-base sales grew by a certain percentage, so wafer capacity grew by that percentage. Revenue reflects prices, service mix, timing, and other commercial factors as well as physical work. The physical outcome requires its own evidence.
The same caution runs in the other direction. A productivity improvement can create substantial customer value without appearing as a new lithography-system unit. Counting only complete systems shipped would miss that channel. The correct model includes the installed fleet while respecting what the public figures can and cannot measure.
The second bridge: a gross-margin explanation is not a segment margin
ASML reported a third-quarter gross margin of 57.5 percent, up from 54.0 percent in the prior quarter. Management associated the stronger result with the quarter's execution and productivity-upgrade demand. That is the company's explanation of the aggregate result. Quarterly results and management commentary
A reader should not reverse-engineer a precise installed-base gross margin from those two percentages. Aggregate margin can move with product mix, recognition timing, cost absorption, and other changes. Without a disclosed decomposition, several combinations could produce the observed result.
This is especially relevant when comparing quarters with large system deliveries. A high-value mix shift can change the total even if an individual service activity's economics are unchanged. A compelling narrative about recurring services does not substitute for the missing allocation.
A careful earnings note distinguishes observed figures, management's explanation, and the analyst's interpretation. The interpretation might be that installed-fleet productivity is becoming a more visible source of commercial demand. It should remain an interpretation supported by the stated evidence, rather than a fabricated segment profit calculation.
At the fab: two proposals reach the same meeting
The fictional Meridian Fab needs more useful output for a supported process. Its manager receives two proposals. One adds another complete tool and the surrounding installation work. The other upgrades part of the installed fleet and coordinates maintenance differently. The manager asks both proposals to describe the same production objective.
The objective is not simply more rated wafers per hour. Meridian needs approved output for a particular process route, with the relevant quality conditions and downstream resources available. A faster tool that produces work the next stage cannot accept does not satisfy that objective by speed alone.
The proposals also have different schedules. A new tool can require space, utilities, installation, and process qualification. An upgrade can require downtime, parts, engineering work, and requalification of the changed configuration. The existing tool's presence does not make the second proposal free of interruption or technical acceptance work.
Meridian compares the alternatives over the same horizon and product mix. It identifies the work each proposal assumes other teams will perform. This avoids a common purchasing distortion in which the new-tool proposal includes every supporting expense while the upgrade proposal shows only the supplier's invoice.
Count passes before counting completed wafers
For an illustrative calculation, suppose one stage runs at three hundred wafer passes per hour for six hundred usable hours in a month. That produces 180,000 passes. A pass is one traversal of the specified operation, not automatically a newly completed wafer or a finished chip.
Suppose the simplified process route requires twenty passes of that type per wafer and all other assumptions remain fixed. The arithmetic corresponds to nine thousand wafer-route equivalents. The conversion is invented for this example. Real routes use different layers, equipment, and process requirements, so an analyst should not copy the ratio into a semiconductor forecast.
Now assume an upgrade raises the rate to three hundred thirty passes per hour. With the same six hundred usable hours, the stage could process 198,000 passes, equivalent to 9,900 under the same twenty-pass assumption. That is a ten-percent improvement at this stage before considering any other constraints.
The distinction between passes and completed output prevents a large counting error. It also makes the operational question sharper: does this stage constrain the desired product, and can the rest of the route support the extra work? An upgrade's value depends on those answers, not only on its rated processing rate.
Installation downtime changes the first month
Meridian's upgrade proposal requires twenty-four hours of the otherwise usable monthly time for installation and qualification under this fictional schedule. The first month therefore has 576 production hours at the higher rate, producing 190,080 passes. Compared with the old 180,000-pass baseline, the initial gain is 10,080 passes, or 5.6 percent.
The following month, if the planned six hundred usable hours are restored, the simple rate calculation yields the full ten-percent stage improvement. The two results are not contradictory. One includes the transition interruption; the other describes a steady operating month under the stated assumptions.
A project report that presents only the steady-state result can overstate the first quarter's benefit. A report that examines only the installation month can understate the longer-lived improvement. The investment needs a time profile that includes both.
The manager also tests a delayed qualification. If installation finishes on time but production approval takes longer, usable output follows the latter milestone. A physical modification is not the same as an accepted production configuration. The proposal should name the evidence and owner required for that transition.
Availability and speed are different levers
Another proposal leaves the rated processing rate unchanged but reduces unplanned interruption. This can increase useful output by adding usable hours. It is a different mechanism from making each hour faster, even though the final capacity benefit can look similar in a summary chart.
Suppose an invented operation runs at three hundred passes per hour. Recovering thirty otherwise lost hours adds nine thousand passes. A rate improvement applied only during existing operating hours produces a different calculation. The team should identify which mechanism it expects and what measurement would verify it.
ASML describes coordination of maintenance and upgrades with customers' production schedules. That supports examining the operating calendar, rather than treating technical support as an isolated repair activity. The exact improvement at a particular customer still requires local evidence. ASML's installed-fleet support approach
Meridian measures both availability and rate after the intervention. It also records reasons for downtime. If a later month improves because the fab ran a simpler product mix, the project should not assign the whole gain to a hardware upgrade. The causal interpretation needs the conditions under which the comparison was made.
The comparison must include work already scheduled
Meridian's first proposal includes a twenty-four-hour installation interruption. The manager now discovers that a maintenance stop of the same duration was already planned in the comparison month. That changes the counterfactual. Charging the upgrade for the whole interruption while assuming uninterrupted operation without it would make the upgrade appear less attractive for the wrong reason.
If the old configuration would have run for only 576 hours that month, its assumed three-hundred-pass rate yields 172,800 passes. The upgraded configuration at three hundred thirty passes for the same hours yields 190,080, a ten-percent increase. The earlier 5.6-percent result was correct for its explicitly different baseline of six hundred hours without the upgrade. The figures describe different alternatives, not a disagreement about multiplication.
The same discipline applies to spending. A planned replacement part or engineering visit may occur with either proposal. The incremental investment should identify which costs change because the upgrade is chosen and which would occur anyway. That does not make the common costs irrelevant to the fab's total budget; it prevents them from being counted as a difference between alternatives when they are not one.
Benefits can be double counted too. Suppose the proposal values additional output and separately values the revenue earned from selling that same additional output. Unless those terms represent genuinely different economic effects, adding them would count one benefit twice. The project worksheet should connect each benefit to its underlying operational change and explain how it enters cash flow.
A maintenance intervention that avoids a future interruption creates another comparison. Its benefit depends on the credible alternative failure and repair pattern, with uncertainty included. The manager should not assume that every hour of preventive work avoids an equal hour of production loss. The evidence for that relationship belongs in the proposal.
For an external investor, these examples explain why a customer's willingness to buy an upgrade can be rational even when a public data sheet does not reveal its return. The alternative may involve planned maintenance, deferred replacement, scarce space, or a specific production opportunity. The quarterly service-revenue figure captures the commercial result, while the customer's counterfactual remains a separate analysis. Recognizing that distinction supports a richer interpretation without inventing private economics.
Quality belongs in the acceptance condition
A faster process is useful only if the output remains suitable for its intended route. Meridian's acceptance plan therefore includes the relevant product-quality and process requirements, defined by its qualified engineering process. The article does not prescribe technical thresholds for a lithography tool.
The plan identifies which characteristics must be checked after the change, how the evidence will be reviewed, and which products are included in the approval. A configuration can be suitable for one product and require additional work for another. The upgrade should not silently expand its supported scope because the machine model is familiar.
This also affects the financial benefit calculation. If additional processing creates more rework or consumes another scarce resource, gross passes can rise without the expected increase in accepted output. The manager compares the completed route outcome with the original objective, while retaining intermediate measurements that explain the result.
An outside investor will rarely see this full acceptance package. They can still avoid equating a supplier's sales category with a universal throughput number. The commercial transaction and the customer's production acceptance are related events with different evidence.
The investment worksheet includes the transition
Meridian builds a cash-flow comparison over a defined horizon. It includes the purchase, supporting engineering work, installation interruption, ongoing expenses, and expected incremental benefits. Every uncertain input has a range or scenario. The supplier's quoted price is one input to the decision, not the entire decision.
Consider a purely fictional project costing two million currency units today and producing net cash benefits of seven hundred thousand at the end of each of five years. At an assumed ten-percent discount rate, the present value of those benefits is approximately 2.654 million, leaving a net present value of about 654,000.
If every benefit is delayed by one year while the initial payment remains today, the present value falls to approximately 2.412 million and the net value to about 412,000. The example holds all other assumptions constant. It illustrates how timing changes an investment even when the final technical capability is unchanged.
These figures are not estimates of an ASML upgrade's price or return. They show the questions a buyer and an investor should preserve: when does useful output begin, which costs precede it, how long does the benefit persist, and what changes if qualification or demand differs from plan?
Useful capacity also needs useful demand
The upgrade proposal assumes that the fab can sell or otherwise use the additional approved output. If demand is weaker, the physical benefit may remain real while the financial return falls. An equipment decision combines a technical opportunity with a commercial forecast.
Meridian tests its proposal under several demand paths. One uses the expected product mix; another reduces the volume that requires the upgraded stage; a third shifts demand toward a product that needs a different capability. The scenarios identify which assumption makes the project attractive rather than hide that assumption inside one average utilization rate.
The same reasoning applies to ASML's customers expressing interest in equipment for later years. Customer discussions can provide useful demand visibility, but interest, a firm order, shipment, acceptance, and revenue recognition are distinct events. The October release itself discusses future demand and capacity as expectations rather than completed results.
An earnings reader should retain those verbs. Replacing expects with has secured can materially change a claim. The industry's long planning cycle makes forward-looking coordination necessary; it does not make every expression of future interest equivalent to a realized sale.
A service relationship can be durable without being automatic
An installed machine creates continuing opportunities for support, parts, field options, and other services. That can make the installed fleet commercially important for many years. The existence of the fleet does not mean every future service euro arrives without cost, competition, customer decisions, or regulatory constraints.
ASML's support description emphasizes engineering staff, parts logistics, diagnostic work, software, and customer training. These activities require resources. An investor should not assume that a service category has negligible cost merely because it is associated with equipment already sold.
The nature of the work can also change with the age and mix of the fleet. A mature machine requiring maintenance differs from a recently installed platform receiving new options. The reported category combines economic relationships whose duration and timing may vary.
This makes installed-fleet analysis more interesting than a simplistic recurring-revenue label. The analyst can ask how the fleet grows, which upgrades customers find valuable, what work supports availability, and how reported revenue changes with the intervention mix. Those are concrete business questions even when a precise forecast remains uncertain.
Avoid calculating a machine price from the wrong numerator
Quarterly reporting also includes new and used lithography-system unit counts. Dividing total sales by those units produces an attractive average that mixes system sales with installed-base services. It does not establish the price of a representative new machine.
Subtracting Installed Base Management before dividing can remove one obvious mismatch, but it still does not reveal every product, configuration, recognition, and contract detail needed for a clean average selling price. A rough derived ratio should be labelled as such, with its limitations visible.
This is a useful discipline for semiconductor news generally. A denominator can be correct and still be attached to a numerator describing a broader economic scope. The calculation becomes misleading when its label claims a precision that the disclosed data does not support.
For the current quarter, the stronger analysis may be qualitative: more reported system units, more installed-base revenue, and a changed aggregate margin, with management's explanation and the relevant timing. A smaller set of well-labelled observations is more useful than an elaborate model built on an invented unit price.
What to watch when the next quarter arrives
The next review should keep the same boundaries. Did installed-base revenue change, and what did management identify as the relevant mix? Did the aggregate margin move? Were capacity expectations revised? Which reported milestones concern sales, production, or customer plans?
A disciplined reader retains the earlier figures and wording so a later change can be interpreted. Quietly replacing a forecast with its latest version removes information about execution and uncertainty. Keeping the original does not require treating every revision as a negative surprise; it makes the record understandable.
The fictional fab review follows the same principle. It compares the actual post-upgrade rate, usable hours, acceptance, and commercial use with the approved proposal. It identifies which differences came from the intervention and which came from product mix, demand, or the surrounding process.
That paired review connects the public supplier story to the customer's economic logic without pretending that either side's information is complete. It explains why installed machines deserve attention while preserving the gap between supplier revenue and customer output.
The quarter viewed through the machines already working
ASML's October results make the installed fleet part of the current capacity discussion. The news is not only a future factory or a new system shipment; it also concerns extracting more useful work from equipment customers already operate. That can affect both the supplier's quarter and the customer's production plan.
The careful interpretation has three pages: a financial reconciliation, a technical acceptance plan, and a timed investment case. None replaces the others. A revenue increase does not measure wafer capacity directly, a rated speed does not establish accepted production, and a technical improvement does not guarantee an attractive return under every demand path.
Reading those pages together produces a more useful understanding of the semiconductor cycle. Capacity can come from new equipment, better use of the installed fleet, and changes elsewhere in the route. The next question is which of those mechanisms the evidence actually supports, and when the customer can use the result.