ASML’s Results: How Much Chip Supply Can Lithography Upgrades Add?

The value of getting more from installed equipment

Adding chip supply does not always require a new fab or another machine. If equipment already in operation finishes more work in the same time, part of a factory can gain capacity. The size of that gain depends on what the work actually is. A lithography system transfers a circuit pattern onto a wafer. A wafer that passes through the system is not yet a finished chip. The equipment used, and the number of times it must process that wafer on the way to a finished product, determine where capacity is constrained. [4]

ASML Holding N.V.’s results announcement on October 14, 2026 directs attention to upgrades of installed lithography equipment. A fab needs additional processing that it can use for its products while retaining the required quality, rather than simply a larger number on a specification sheet. A faster machine does not provide the same benefit to operations that do not use it, or to operations with ample capacity already. The equipment eligible for an upgrade has to be considered alongside the places where work is getting held up. [1]

The distinction matters both to chip buyers and to investors evaluating equipment suppliers. More capacity at a particular lithography operation can give a customer room to revise production plans. An equipment supplier may gain upgrade and support work, while the customer’s need for additional machines depends on product design and production requirements. Before asking whether shortages are broadly resolved, it is more useful to identify the processing gained and whose constraint it eases. That is the starting point for connecting reported revenue with conditions inside a fab.

The quarter’s figures and the 2027–28 supply outlook

Installed Base Management sales were €3,289 million in the quarter ended September 27, 2026. This category combines service and field-option sales. ASML cited strong demand for productivity upgrades and said planned 2027–28 supply would still fall short of customers’ forecast demand. [1]

The category cannot be read as upgrade payments alone. Support that keeps a machine operating reliably provides a different customer benefit from a change that raises processing capacity. The contribution to production depends on the work performed, the equipment involved and the operations to which it applies. Revenue is measured in euros; process capacity is measured in processing events and time. Connecting them requires information about the capabilities added to particular systems. Inferring upgrade counts or additional output from the revenue total would omit differences between machines and the work sold.

Future supply plans also require separating the machines ASML supplies from the wafers customers manufacture using them. Higher equipment output does not directly measure higher output at a customer’s particular fab operation. Systems must match the type and timing of customer requirements and be integrated into the destination process. Upgrades can offer an interim way to add usable processing. Whether they close a capacity gap depends on the work they can handle. Reading the strength of demand separately from the means of supplying it makes the earnings assessment more specific.

For markets, the useful question extends beyond strong revenue to where the supply constraint moves. Better installed systems may allow some work to proceed before another machine arrives. Expansion can remain necessary for work involving finer patterns or a different equipment class. Treating an equipment supplier’s strength as proof that its customers’ procurement problems have disappeared would obscure the possibility that upgrades and further expansion are needed together.

Lithography repeatedly processes the same wafer

A wafer is the disc on which circuitry is built. Manufacturing applies light-sensitive material over a thin film, transfers patterns by lithography and processes the material. Work repeats to build layers. Further operations include inspection, dicing and packaging. Lithography handles one part of the route to finished products. [4]

The equipment metric wph means wafers per hour. Patterns are exposed at multiple locations on a wafer. One processing visit, encompassing loading, measurement, exposure and unloading, has a different unit from the number of individual locations exposed during that visit. Here, one wafer-processing visit means the system processes a wafer once. It is not a chip count or a count of the individual exposure fields. [5]

Measuring and exposing in parallel

Using a speed improvement requires coordinating machine operation with assessment of the process result. ASML explains that TWINSCAN measures one wafer while another is exposed. Overlapping measurement and exposure differs from merely shortening the time light is applied. Processing speed also depends on coordination of loading, measurement, exposure and unloading. Evaluating an upgrade means checking that coordination under the product’s operating conditions. [5]

When operations overlap, distinguish the time one wafer spends inside the system from the interval between successive wafers leaving it. Adding measurement and exposure times does not necessarily give the interval in continuous operation, because work on different wafers can overlap. Faster measurement cannot shorten that interval if the exposure side is not ready to receive the next job. Faster exposure can instead create a wait if measurement or transfer cannot keep up. Continuous-processing gains depend both on shortening a task and on handing work to the next task after it has been shortened.

This distinction gives a reader concrete questions to ask of a specification. Does it reduce the elapsed time for one wafer, increase the number leaving during sustained operation, or do both? The latter affects processing slots, but does not establish a matching reduction in the time needed for the whole manufacturing route. If waits between operations are long, time saved inside the machine may have little effect on shipment dates. This is a general way to read overlapping work, not an estimate of how many seconds PEP-E saves in any particular component.

If a product’s wafer revisits a particular equipment group, those repeat visits consume the group’s processing slots too. Additional hourly capacity does not necessarily allow an equal number of new wafers to enter production. The planning benefit differs depending on whether the gain supports repeat processing or creates room for new starts. This analysis need not count every operation in a fab; it can be confined to the work performed by the equipment group in question.

What the NXE:3800E upgrade changes

On April 15, 2026, ASML described PEP-E (Productivity Enhancement Package) for the NXE:3800E: 220 to 230 wph, with comparable overlay, or layer-alignment accuracy. This was an April announcement; October’s results did not introduce this feature. [3]

On the stated comparison basis, the difference is ten wafer-processing visits per hour. Dividing ten by 220 gives an increase of about 4.55%. That calculation describes the change from the stated baseline, rather than a measured increase in total fab output. Using the capability in production requires compatible equipment and operations, together with acceptable quality and operating time. This is why product-specific manufacturing conditions and usable time after an upgrade matter.

The value depends on how much work reaches that system. Additional processing slots are easier to use when wafers are waiting upstream and the following operations have room. Idle time on the eligible equipment may indicate a constraint elsewhere. Between a higher specified limit and using that limit are decisions about sequencing, product changes, inspection and maintenance. The customer needs capacity that is usable after those requirements have been accommodated.

Even with each machine’s assigned work and usable time held constant, a comparison for the upgraded machine differs from one for its entire equipment group. If other machines in the group are unchanged, the group-wide percentage gain is diluted according to the upgraded machine’s share of baseline processing. The number of compatible machines alone does not determine that share. A system operating for long periods contributes different processing from one assigned only limited work. Moving from an individual machine’s improvement to a group-wide result requires comparing the eligible machines’ baseline slots with the group’s total slots on the same basis.

Greater speed has to preserve precision

Lithography has to align a new layer with layers already formed. The accuracy of this alignment is overlay. ASML describes production metrology that measures parameters including overlay and returns data to the lithography system for process adjustments. A wafer processed quickly still has to meet the required position and pattern conditions before it provides useful work downstream. Speed and quality therefore belong in the same assessment. [6]

Quality here concerns whether the upgraded process can reliably make the same product to its required conditions. A good average result can still leave additional adjustment and checking work if variation increases during production. The acceptable range depends on the operation. A performance comparison should therefore identify which conditions have been held comparable alongside the speed figure. Simply omitting checks to obtain a faster result would not establish a useful improvement.

Two levels of assessment help explain this. The equipment assessment asks how throughput and precision change under defined conditions. The fab assessment asks what happens when that capability is integrated with actual products, materials and adjacent operations. A sound equipment improvement can still require time for process adjustment in a factory. A fab already running closely comparable work reliably may be better placed to translate it into production.

Overlay and yield are different measures. Yield is the share of fabricated chips that function, and factors beyond lithographic alignment affect it. A statement about maintaining alignment performance does not establish that total fab yield was unchanged or improved. Useful evidence includes quality results at the modified operation and the results of products that proceed through the rest of manufacturing. Those data can establish precisely where a speed improvement delivered value. [6]

Exposure requirements change the work demanded of equipment

A complex pattern can be divided into simpler patterns that are combined to form the required structure. ASML’s explanation of immersion lithography describes how this multi-patterning introduces additional processing. A greater patterning workload can absorb capacity released by faster equipment. The supply assessment therefore needs both the system’s processing capability and the work the product asks it to perform. [7]

Adding every exposure into one total is not enough. Operations that require different equipment or processing conditions cannot necessarily share the same slots freely. Fewer visits to one equipment group can coexist with more work at another. Separating the manufacturing route by lithography equipment class makes it easier to identify where capacity is released and where more capability is needed.

A design change provides another route for altering the workload. It may ask more work of existing equipment to form finer patterns, or use a different approach to consolidate processing. The comparison depends on the required pattern, available equipment, quality and adjacent processing. Counting fewer exposures alone would miss costs or qualification work transferred elsewhere. Evaluating a process change requires a broader comparison than evaluating an improvement in one machine’s speed.

Layers and processing visits

The correspondence between product layers and lithography visits also matters. Dividing a layer’s pattern across multiple patterns means that adding one layer and making one additional visit are different changes. Workload should be assessed through how each layer is formed on particular equipment. A layer count alone can miss differences between products with the same number of layers. Mapping which work a new manufacturing approach adds or removes identifies where speed improvements are needed. [7]

Combine equipment, processing speed and usable time

Consider the processing available from a particular lithography equipment group over a defined period. For each system, multiply its speed for the relevant work by the time actually available to perform it, then sum the results. This gives the group’s wafer-processing slots. Speeds must refer to comparable work, and available time excludes periods unavailable for processing, such as maintenance or changeovers. More machines, faster machines and more usable time can all affect this same pool.

Let p be the number of visits a product’s wafer requires to that group. In a simplified case where all slots can serve the same work, dividing the slots by p expresses the capacity as complete sets of required lithography work per wafer. Wafers may already be partly processed at the start of a period, so this is not a bound on wafers actually completed or newly started during that period. The count covers visits to the selected group, not all manufacturing steps or finished chips.

Distinguish work in progress from new starts

A period boundary also requires matching visits already completed with visits still needed. Many wafers that have nearly finished their work at the group can leave it after only a small amount of additional processing. A period dominated by newly arriving wafers can show fewer wafers finishing the full set of work despite the same total visit count. Calling the former a production success and the latter an upgrade failure would overlook the composition of work in progress. Speed assessment counts work performed; production planning also counts the work remaining for individual wafers.

Planning more starts also commits processing slots for later return visits. A wafer can fit into an unused slot today yet join a queue later if its return coincides with existing work. Conversely, giving priority to return visits from wafers already in production may reduce work in progress without increasing starts. Distinguishing whether the first change after an upgrade is in new starts or in waits for return visits helps identify how the additional capacity is being used. Starts, machine visits and completion of an operation should not be collapsed into one wafer count.

The relation sets out the conditions under which speed can add production capacity. With unchanged p, usable time and quality, higher speed adds capacity measured in complete sets of that work. More required visits absorb slots, while less available time erodes the benefit of greater speed. A fab making several products needs to match each product’s required visits with its planned volume. Collapsing everything into a single average wafer can conceal changes toward products that use the operation more intensively.

Matching operations with equipment also changes how total machine counts should be read. ASML’s technology explanation distinguishes lithography for layers requiring very fine features from that for larger features, with extreme ultraviolet (EUV) and deep ultraviolet (DUV) serving different roles. Adding all equipment into one capacity total can imply that work can be reassigned where it cannot. Defining the group narrowly identifies usable processing slots; it is not a way to understate available capacity. [8]

Adding systems and upgrading installed equipment

An upgrade decision starts with the work the customer wants to add. For an operation that can use the NXE:3800E speed improvement, the assessment identifies the products assigned to it and whether the modification preserves their required quality. Work requiring a different system or lithography approach needs suitable capability. Upgrading installed equipment and adding systems are not always interchangeable responses to the same capacity gap.

Installed equipment offers the advantage of an existing location and established material and workflow arrangements. A modification still has to be integrated and its equipment and product results checked. Adding another system requires decisions about its location, assigned work and integration with the existing flow, as well as its capacity. Either choice has to fit staffing, quality-assessment resources and the sequence in which wafers move through the fab.

The cost comparison extends beyond equipment prices. It includes processing unavailable during installation or modification, process qualification, added support and the time needed to stabilize operation. An option with a smaller purchase price can impose a substantial planning cost if it interrupts critical work. Where added capacity can be introduced gradually, the disruption may be spread out, provided the relevant machines can share the assigned work.

Equipment sales and installed-system work should therefore not be treated as a simple contest with one winner. Spending requirements depend on whether the customer wants a faster existing operation or additional capability elsewhere. If changes in materials or design add required operations, upgrades can coexist with demand for new capacity. Our discussion of Wolfspeed’s refinancing and shareholder claims examines the financing side of semiconductor production. Funding alone does not add usable lithography slots; the work that the investment is meant to serve remains essential to the connection.

Assessing the interruption required for an upgrade

An upgrade assessment includes the time unavailable while the modification is made. Stopping production for an upgrade forgoes processing that the equipment could otherwise have performed. Additional processing after restart takes time to recover that volume. An upgrade combined with scheduled maintenance differs from one that interrupts required production: the extra processing forgone differs. The relevant question is what work would have been performed during the interruption, as well as how long the interruption lasts.

The published speed comparison permits a narrow calculation of processing forgone and the gain after restart. Suppose D hours of productive time at the old conditions are lost, and the same work then runs at 230 rather than 220 wph. Forgone processing is 220 × D, and the additional capacity is ten per hour, so recovering it using only the increment takes 22 × D usable processing hours. Both D and 22 × D use hours available for comparable processing; conversion to elapsed calendar time requires the operating schedule. D is an input, not a reported interruption duration. The relation assumes unchanged operating conditions and that all added slots can perform the same work.

This is a recovery time for processing volume, not a financial payback period. Financial payback also depends on upgrade and operating costs, saleable product prices and whether the additional volume can be sold. Even the physical recovery takes longer if work is unavailable immediately after restart or adjustment increases nonproductive time. If the modification occurs during time already unavailable for processing, the extra volume forgone is smaller. The simple relation cannot replace an assessment of the interruption or its economics.

Which products need the forgone processing also matters. Recovering an equal volume later may still disrupt the order in which products have to be supplied, changing customer commitments and work inside the fab. Routing work to another compatible machine during an interruption can sometimes soften the effect. That option is unavailable where materials or process requirements are incompatible. The upgrade assessment therefore needs the arrangement of work during the interruption alongside the speed comparison.

Work can accumulate beyond a faster operation

After lithography, a wafer moves to further work. If the next processing or checking operation cannot handle it, the extra volume produced by faster lithography can wait between operations. The improved system may complete more work without increasing the factory’s finished-output rate. Where the upgraded operation differs from the operation constraining the fab as a whole, the use of additional capacity needs to be reconsidered.

More waiting also has economic consequences. Materials that have already received work and absorbed costs spend longer unable to proceed. A longer interval before they become saleable products changes work-in-process management, scheduling and material preparation. The assessment should follow the volume accepted by subsequent operations and waiting times alongside the volume through lithography. Improvement across those measures would strengthen the case that the additional processing has reached supply.

A lack of immediate growth in finished output does not by itself establish that the upgrade had no value. Spare capacity at the operation may make maintenance easier to schedule or allow more flexibility in product sequencing. Capacity used to accelerate production and capacity retained to stabilize operation are two uses of the same pool. The second benefit can be hard to see in a short-period processing count. Results should be assessed against the purpose of the modification.

A before-and-after comparison showing more output after an announcement is insufficient to establish this connection. Material availability, additional equipment or product mix may have changed at the same time. Comparing the same work on the eligible system, the volume accepted by adjacent operations and comparable quality conditions makes the upgrade’s contribution easier to identify. Our guide to lead–lag relationships and causation supports this approach. Rather than assigning all growth to one announcement, the analysis follows the operation that changed.

The value depends on the products a fab makes

Priorities matter even when production increases within one product. Expanding an established product offers a clearer pre-upgrade comparison. Introducing a new product can combine equipment and manufacturing-condition changes. A successful ramp does not make every additional wafer attributable to the upgrade. Conversely, a demanding introduction can be followed by more effective use of the capability once operating conditions stabilize.

AI demand should be connected through these operations too. Even with broader AI use, changes in chip type, design or production location alter the lithography work required. Our article on slower AI development and chip and memory demand examines the composition of demand, including training and inference. For this upgrade discussion, the next question is which manufacturing operations the required products load. An AI end-use label does not establish the same demand growth for every lithography system.

SG Group View: Focus on usable lithography work

SG Group sees installed-equipment upgrades as a route for adding slots at constrained lithography operations. Assessing their value requires three connections: the match between equipment and work, the relation between processing speed and required visits, and processing forgone to make the change. Together these distinguish a fab that can use the capability well from one unable to use the same specification improvement fully. An unchanged equipment count does not establish unchanged processing capacity.

The frequently overstated case applies the specification’s improvement rate directly to product supply. That rate is unsuitable when required visits rise or eligible systems handle only part of the relevant work. The easily understated case is additional processing where compatible installed equipment has work waiting and constrains manufacturing. Even a modest speed gain can then release work to other equipment. Both assessments depend on information about the selected operation.

Specific observations would weaken this view. If eligible equipment has spare capacity and the principal constraint lies elsewhere, upgrades become less central to the supply assessment. Failure to maintain product quality reliably, or substantially more interruption and adjustment time, would also erode the expected benefit. More processing of the same work to comparable quality, accepted by downstream operations, would strengthen the interpretation that upgrades support supply. Revenue and specifications alone should not fix the conclusion.

Chip buyers and equipment suppliers receive different benefits

If the eligible operation gains capacity and more work proceeds through subsequent manufacturing, a chipmaker can potentially use its existing investment across more products. More output spreads broadly unchanged fixed costs across more units. Upgrade payments and additional operating costs also matter, however, and the reduction in total cost per product depends on the fab. The speed improvement cannot simply be applied as the same percentage reduction in costs.

Equipment suppliers can gain work adding functionality to installed machines and supporting their operation. Even an upgrade with substantial customer value does not determine the supplier’s profit increase: prices, work performed, parts and support requirements intervene. A customer able to address an immediate gap with installed systems may change the sequence of equipment purchases. Investment in capabilities for different operations can remain necessary. Work for both new systems and upgrades does not imply an unchanged revenue or profit mix.

For businesses that use chips, the transmission depends on whether supply of the required products increases. An improvement serving a different product will not ease their procurement constraint. Even added supply of the right product affects prices and lead times through other demand, inventories and purchasing terms. Following which products reach which buyers and when offers a more useful business assessment than jumping from spare fab capacity to lower finished-product prices.

For equity markets, the question is which firms can turn the added processing into earnings. Owners of eligible equipment, suppliers of materials and support, and users of finished products receive different benefits and incur different upfront costs. A strong announcement still requires a separate assessment of expectations already reflected in prices. Our explanation of reported results, company guidance and market expectations provides a foundation for that comparison. The equipment-performance calculation alone cannot forecast a stock’s movement or future profit margin.

How the change reaches work and procurement decisions

Production planners then need to establish where the additional work can go. More processing affects material preparation, wafer-movement sequences and inspection schedules. If the next operation is already busy, adjusting it first may make the speed improvement more usable. This does not establish an across-the-board staffing increase or decrease. The relevant assessment identifies added tasks, reduced waits and checking work still required, and whether planners, maintenance staff and quality teams can support the same change.

Chip buyers should ask how a supplier’s intended products connect with the capacity used to make them. An upgrade program alone does not establish easier procurement of their required chips. Relevant answers concern the products served, volume-production conditions and the operations connecting the improvement with shipments. Our article on Apple’s AI investment and use of external technology examines decisions on the product-provider side. End-use plans and manufacturing arrangements need not change together. The product requirements linking them make equipment news useful for procurement decisions.

When an upgrade has more or less effect

The gain is easiest to use when compatible systems have work waiting, the required visits are unchanged and subsequent operations have room. The added slots can serve work already needed. The relevant measures are processing at the upgraded system and results at the operation accepting it. Improvement in both, with stable quality, makes the contribution to fab output easier to identify.

With work waiting but a product change adding lithography requirements, part of the speed gain absorbs the increased workload. Maintaining existing volume under the new requirements can have value without more wafer starts. Low apparent output growth can conceal a larger constraint that would have existed without the change. This comparison needs the changed product requirements, not only before-and-after operating results.

Where work reaches the eligible machine but the next operation cannot accept it, the waiting point may move. More processing without a shorter time to completion is a reason to revisit the fab’s overall constraint. Matching the work released to downstream capacity may be easier to manage than immediately maximizing volume. The assessment should consider using spare capacity to stabilize the manufacturing flow, rather than treating full use of the increment as an objective in itself.

Little work assigned to the eligible system makes an immediate volume benefit harder to establish. Capacity retained for future use can have value, but that assessment needs the intended products and timing. Weak demand limits the immediate conversion of capability into revenue. An upgrade for anticipated work and one addressing a current shortage need different outcome measures. The four cases in the diagram describe conditions, not probabilities.

Two objections: modest gains and future equipment demand

The first objection is that a modest speed improvement for one model cannot change the supply assessment. Testing it means examining the product operations served by eligible systems, rather than all machines owned by a company. A small group can matter where few alternatives perform its work and that work is waiting. Numerous upgrades can have little immediate supply effect when eligible systems have spare capacity. Matching equipment with the shortage of compatible work makes the objection testable.

Required work can change after an upgrade

The second objection is that faster installed systems eliminate demand for new equipment. For unchanged products and output, they can cover more of the required processing. Greater production or changed manufacturing requirements alter the workload and capability needed, however. Spare capacity in the upgraded group can coexist with a need to expand another group. Changes to the timing of equipment purchases and elimination of future equipment demand are different outcomes.

The next evidence: applicable systems and changes in operations

The next company materials should be read for what installed-system work improves, as well as its revenue. Equipment type, applicable scope, post-change performance and operating conditions help connect the specification comparison with production operations. The products and lithography work served by modified machines matter too. The same package can have a different company-wide contribution when applied to different work.

Customer disclosures should connect wafer starts, processing at the eligible group, required exposures and acceptance by subsequent operations. More starts with little change in the group’s workload could reflect product mix or a different processing approach. More work at the upgraded system without more volume proceeding toward completion is evidence to examine queues or another constraint. Neither pattern can be explained by the single fact that a machine has become faster.

A comparison record should separate specifications from operating results and retain the period, model, product and quality conditions. The specification describes a capability; operating results show how much of it was used under the actual conditions. A difference should prompt examination of unavailable work, maintenance, adjustments and product changes, rather than an immediate success-or-failure label. Quarterly segment revenue does not reveal those operation-level relationships.

Extending an improvement at one lithography operation to the relative strength of countries or industries requires evidence about other products and fabs. Our analysis of US–China AI competition(Full article paid) addresses that wider context.

Final assessment: Trace supply changes through manufacturing

ASML’s results provide a basis for examining efforts to get more required lithography work from customers’ installed equipment. The effect appears in product operations served by compatible systems. Preserved usable time and precision, with a match to the required exposure workload, allow greater speed to supplement scarce processing slots. Considering upgrades alongside new equipment supply identifies which constraints customers are trying to address and by what means.

The supply assessment follows results from the equipment to subsequent operations and eventually to products. Where that flow changes after a better specification determines the interpretation. Firms completing more necessary work gain business options, while equipment and material suppliers receive different work. Connecting the demand signaled by the results with those operations and outcomes makes the improvement’s significance for global semiconductor supply more concrete.