Summary
- ASML's economic power comes from making chipmakers pay for a scarce manufacturing route: without its EUV, DUV, metrology, inspection, software and service stack, the most advanced customers either accept more process steps, lower yield, slower capacity ramps or delayed product roadmaps.
- The same scarcity raises ASML's own risk. In 2025 it generated EUR 32.7 billion of net sales, 52.8% gross margin and EUR 9.6 billion of net income, but it also spent EUR 4.7 billion on R&D, relied on about 5,100 suppliers, served a small group of dominant chipmakers and entered 2026 with export-control outcomes still wide enough for management to include them inside guidance.
- The judgment is constructive but conditional: ASML can preserve superior economics if installed-base revenue, productivity upgrades and High NA adoption turn customer dependence into durable cash flow; the thesis weakens if supplier capacity, China restrictions, customer delays or cheaper extension paths make capital intensity grow faster than the process advantage.
Scarcity is the product, but uptime is the toll road
ASML is paid because the customer is not simply buying a machine. The customer is buying access to a manufacturing possibility. A leading foundry, logic maker or memory producer can delay a node, stretch older tools, add multi-patterning steps or tune design rules, but it cannot casually replace the lithography route that prints the most critical chip layers at scale. ASML's strongest economic position begins there: the buyer's alternative is rarely another like-for-like supplier with the same installed knowledge, service reach and process roadmap. The realistic alternative is a more expensive or slower manufacturing plan.
That matters because the customer who pays for an ASML tool is not trying to own equipment for its own sake. The customer is trying to sell wafer starts, advanced chips, memory capacity, foundry slots or strategic autonomy. If a new lithography system improves the number of good chips per wafer, removes process steps, shortens cycle time or lets a customer start a node before rivals, the price of the tool is judged against the customer's downstream economics. ASML captures part of that value because it sits at the bottleneck. The customer benefits if the tool improves yield and time to market.
The downside sits with both sides: the buyer carries fab capital and underutilization risk; ASML carries R&D, supplier, field-service and regulatory risk before the customer revenue is fully recognized.
The distinction between revenue growth and value creation is therefore important. ASML can grow sales because chipmakers expand capacity during an AI or memory upturn. That does not automatically mean value creation. Value is created only if the extra systems, field options and software produce returns above the capital needed to design, build, install and support them. The 2025 numbers show the current strength: EUR 32.7 billion of net sales, EUR 17.3 billion of gross profit, EUR 11.3 billion of income from operations and EUR 9.6 billion of net income. The 52.8% gross margin says the company is not just pushing volume.
It is still monetizing scarcity.
The risk is that scarcity becomes more expensive to defend. High NA EUV, better DUV extension, metrology, computational lithography and uptime service all demand years of engineering and a supplier base that can meet tolerances few industries require. If the next tool generation requires disproportionately more money, factory space, supplier capital, service labor and inventory to deliver each marginal unit of customer value, ASML's gross margin can remain optically high while returns on incremental effort become harder to earn. The company must make technological scarcity outrun capital intensity, not merely coexist with it.
ASML Netherlands B.V. is infrastructure evidence, not a carrier
The named entity in this research is ASML Netherlands B.V., recorded publicly by the RIPE NCC member directory with an address at De Run 6501, 5504 DR Veldhoven, Netherlands, and an areas-serviced row for NL. That evidence should be read narrowly. RIPE NCC membership indicates participation in the Internet number-resource governance environment. It does not prove that ASML Netherlands B.V. sells broadband, transit, hosting, registry services or managed network connectivity.
The RIPE NCC itself describes its role as allocating and registering Internet number resources to members in its service region, including IPv4, IPv6 and Autonomous System Numbers. The directory signal is therefore resource and operational-context evidence, not a telecom revenue claim.
That narrow reading still matters. ASML is a manufacturing technology company whose products sit inside globally distributed chip fabs, but its own operations also rely on secure communications, customer portals, support centers, software delivery, data exchange and local governance of digital resources. A lithography system in a customer fab is not an isolated metal box. It is a controlled, serviced, monitored and upgraded production asset. ASML's public support materials describe CustomerNet as a secure customer portal containing training information, electronic documentation and software-tool downloads.
Its software pages describe embedded software, scanner metrology software, application software, computational lithography and machine-learning methods. That combination creates a digital-service surface around a manufacturing asset.
The correct economic interpretation is therefore not "ASML is a telecom operator." It is that network-resource evidence belongs to the cost and control surface of a company whose value depends on precision equipment operating inside critical infrastructure. Secure portals, support logistics, software updates, customer data, diagnostics and regional support presence affect uptime economics. If a machine can cost customers thousands of euros per minute when down, communication reliability and data handling become part of the service promise.
The public RIPE member evidence supports that ASML has a formal number-resource governance footprint in the Netherlands; it does not expand the product boundary beyond semiconductor equipment, software and services.
This boundary is useful because it prevents two opposite mistakes. The first mistake is to treat every RIPE member as an ISP. The second is to ignore the digital operations behind physical equipment. ASML's business is capital equipment, but the economics increasingly depend on software, data, support portals and field options that make the equipment more productive over time.
For readers interested in telecom economics, ASML is relevant because it shows how network-resource governance, data locality and secure service operations can become supporting infrastructure for industrial scarcity, even when connectivity is not the thing sold to the customer.
The buyer pays because process options are narrower than budgets
The central reason ASML can charge attractive prices is that advanced chipmakers face a constrained process map. EUV systems print foundation layers in advanced 7 nm, 5 nm, 3 nm and newer nodes that DUV cannot economically reproduce at the same complexity. ASML says its NXE EUV systems use a 13.5 nm light source and print features with 13 nm resolution that is unreachable with DUV. Its EXE platform, the High NA generation, is positioned for advanced logic and memory, including sub-2 nm logic and leading-edge DRAM. The buyer is paying for fewer process compromises.
That does not make every layer or every chip dependent on the latest ASML tool. DUV remains important. ASML describes immersion DUV machines as industry workhorses and says dry systems can still print many layers in a chip more cheaply than leading-edge immersion or EUV tools. This is central to the economics. A customer rationally mixes tools: expensive EUV for the layers where it is needed, DUV for less critical layers, refurbished or mature tools for applications such as sensors, MEMS, photonics and radio-frequency chips. The value of ASML's portfolio is not only that EUV is scarce.
It is that ASML offers a family of tools and upgrades across the node ladder.
That portfolio gives customers choices inside ASML's ecosystem rather than outside it. If a fab can upgrade an NXT immersion system to a newer package, use computational lithography to correct distortions, add metrology, improve overlay or refurbish mature tools for another application, ASML can capture value even when the customer is not buying the most expensive new system. This is the practical route by which scarcity becomes more durable. The customer has alternatives to a new EUV order, but many of those alternatives still pass through ASML's installed-base services, software, field options and mature-tool support.
The economic bargain is therefore less absolute than the phrase "near-irreplaceable" suggests, but more defensible than ordinary equipment competition. ASML does not need every chip layer to require EUV. It needs the layers that define advanced performance to require its technology, and it needs the surrounding fleet to remain economically tied to ASML support. The buyer pays because the opportunity cost of not paying can be a delayed node, lower yield, more process steps or weaker customer commitments. The buyer benefits if the tool lowers total cost per good die.
ASML benefits if it can price against that benefit while keeping its own delivery costs under control.
Systems revenue still carries the cycle
ASML's 2025 results show how much the model still depends on system shipments. The company reported 535 system sales units in 2025: 48 EUV systems, 279 DUV systems and 208 metrology and inspection systems. The headline revenue base remains the sale of tools into customer capacity plans. In Q1 2026, total net sales were EUR 8.8 billion, with 67 new lithography systems and 12 used lithography systems sold. Management increased 2026 net-sales guidance to EUR 36 billion to EUR 40 billion and pointed to stronger AI-related infrastructure demand, customer capacity expansion and a combination of new systems and installed-base upgrades.
Those numbers make the upside clear. When customers accelerate capacity, ASML can sell high-value systems, recognize field-option revenue and improve absorption across an organization built for scale. Gross margin of 53.0% in Q1 2026, after 52.8% in 2025, suggests the company still has pricing and mix support. Q4 2025 was particularly strong, with EUR 9.7 billion of net sales and revenue recognized for two High NA systems. If High NA moves from early adoption to broader manufacturing use without destroying throughput economics, ASML can extend the value per advanced customer for another generation.
The same system dependence creates cyclicality. A chipmaker can believe ASML is strategically essential and still defer orders if memory prices weaken, fab utilization drops, subsidies shift, or customers delay advanced products. ASML's order book gives visibility but not immunity. Backlog is a promise of future sales only if the customer still wants the tool, export licenses remain available, suppliers deliver modules, installation slots are ready and ASML can convert orders into revenue. The company defines backlog as accumulated system order value not yet recorded as sales. That is useful but conditional visibility.
This is why the capital-intensity question matters. To support high future demand, ASML must invest ahead of revenue. Cleanrooms, engineering headcount, supplier commitments and field capacity cannot be switched on only after the customer is certain. If the industry cycle turns, the company can be left with costs designed for a larger sales base. Management's share buybacks and dividend policy show confidence in cash generation, but cash returned to shareholders also competes with the need to fund the next tool generation, supplier expansion and service capacity.
The strongest version of the ASML case is not that systems sales will be smooth. They will not. It is that the company can earn enough on each upcycle, and enough recurring or semi-recurring installed-base revenue between system waves, to justify maintaining the scarce capability. That requires disciplined resource allocation. Strategy without resource allocation would be marketing; ASML's real strategy is the decision to keep spending heavily on lithography leadership while customers still value the result more than the cost.
Installed-base work turns scarcity into repeat cash flow
The installed base is where ASML's economics become less dependent on a single shipment date. In Q1 2026, installed base management sales were EUR 2.488 billion, up from EUR 2.134 billion in Q4 2025. ASML defines this category as net service and field option sales. That definition matters because it is not a vague maintenance label. It includes the work that keeps expensive tools productive and can improve their performance after installation.
ASML's public customer-support page is unusually explicit about customer downtime. It says thousands of systems sit in customer fabs worldwide, customers want them running all day and all year, and a machine that is down can cost thousands of euros per minute. That is the toll-road part of the business. The customer has already committed to the tool, process flow and fab plan. Once that happens, ASML's service organization can sell uptime, diagnostics, parts, training, relocation, refurbishment, upgrades and performance improvement. With around 10,000 customer-support employees, ASML turns installed complexity into ongoing work.
The next-day parts claim is economically important. ASML says it can fulfill more than 99% of all parts requests within 24 hours by storing spare parts close to customer systems. That inventory and logistics system is costly, but it is also a competitive barrier. A rival cannot match ASML by building a clever exposure tool alone. It would need global service reach, local spares, customer trust, process knowledge and data feedback from systems already operating in high-volume fabs. The installed base therefore protects the original system sale and creates incremental revenue.
This is also where data sovereignty and locality enter the economics. Customers in Taiwan, South Korea, the United States, China, Japan and Europe are not handing over ordinary consumer data. They are operating sensitive manufacturing assets tied to national industrial policy and customer product roadmaps. ASML's support model depends on information moving across portals, software tools, service teams and local support centers, but the customer and regulator may care where data sits, who can see it, how updates are delivered and which engineers are allowed to work on controlled technology.
Public disclosure does not provide enough detail to quantify cloud-service dependency, but ASML's own pages show software, machine learning, performance monitoring and customer portals as part of the operating model.
The installed-base opportunity is therefore valuable but not frictionless. Each additional service euro depends on trust, security, local compliance and the ability to improve tool performance without interrupting fab output. ASML benefits when customers choose to extend older tools, upgrade NXT systems, refurbish mature systems or improve process control rather than scrap equipment. The customer benefits if these options reduce cost per good die. The risk is that service complexity grows faster than service margin, especially if export controls, regional data rules or customer-specific restrictions fragment the support model.
Research spending is the price of staying alone
ASML's scarcity is not free. In 2025, the company spent EUR 4.7 billion on research and development. In Q1 2026 alone, R&D costs were EUR 1.2 billion. These numbers are not optional overhead. They are the cost of staying at the front of lithography while competitors, customers and governments search for alternatives. EUV itself took more than two decades to industrialize, and ASML says it invested more than EUR 6 billion in EUV R&D over 17 years. That history is a reminder: the economic moat was built through long, risky spending before the revenue prize became obvious.
The current challenge is that each additional generation must solve harder physics and manufacturing problems. ASML's EUV systems use tin droplets hit by CO2 laser pulses up to 50,000 times per second to generate EUV light. The light path must be in high vacuum because EUV is absorbed by air. Mirrors, stages, wafer handlers and metrology must operate at extraordinary precision. ASML says the wafer stage positions the wafer to within a quarter of a nanometer for each exposure while checking and adjusting 20,000 times per second. This is not ordinary equipment improvement.
It is precision engineering at the edge of what customers can manufacture repeatedly.
Software makes the R&D burden broader. ASML says it has one of the world's largest software communities and that customers could not manufacture at dimensions of 10 nm or smaller without ASML software. Embedded code controls the machines, scanner metrology software compensates for sub-nanometer inaccuracies, application software supports calibration and diagnostics, and computational lithography predicts and corrects patterning distortions. The company also describes machine learning and high-volume data as tools for improving process control and product performance.
The research load is therefore mechanical, optical, chemical, computational and data-driven at once.
This breadth supports pricing power because a rival must replicate more than one invention. It also raises the failure cost. If a new platform arrives late, underperforms or requires too much customer process change, ASML cannot redirect its R&D spend into a completely different market with equal returns. The company's chosen strategy is to deepen a bottleneck, not diversify away from it. That can be excellent capital allocation if the bottleneck remains essential. It can also be dangerous if the next generation requires too much spending for too little customer value.
The key question is not whether ASML spends a lot. It must. The question is whether each new euro of R&D preserves a price umbrella or merely defends the current revenue base. High NA adoption will be the test. If it reduces process steps, defects, cost and cycle time in high-volume manufacturing, ASML can justify the investment. If customers slow adoption, rely longer on current EUV plus DUV extension, or delay fabs, R&D remains necessary but less obviously value-accretive.
Supplier concentration is the hidden balance-sheet item
ASML reported about 5,100 suppliers in 2025, including roughly 1,600 in the Netherlands, 700 elsewhere in EMEA, 1,350 in North America and 1,450 in Asia. That breadth can look diversified, but the economics are more concentrated than the count suggests. A lithography system is not a commodity assembly where any missing part can be bought from a substitute catalog. Some modules, optics, stages, light-source components, sensors, precision frames, software inputs and cleanroom processes are specialized enough that supplier capacity becomes part of ASML's own capacity.
The supplier base matters for two reasons. First, ASML's revenue depends on converting customer orders into finished, shipped and installed systems. If a bottleneck supplier cannot scale, ASML cannot always solve the problem with price alone. It may need joint engineering, prepayments, long-term commitments, minority investments, tooling support or redesign. Second, ASML's own capital at risk extends beyond its factories. When suppliers expand for ASML demand, ASML is implicitly relying on their balance sheets and execution. If the customer cycle changes, pressure can move backward into the supplier chain.
The company has a strong reason to manage suppliers closely. ASML's tools are valuable partly because they combine rare technologies into a working system. A customer does not pay for a mirror, a laser source, a wafer stage or a software model in isolation. The customer pays for a machine that exposes wafers reliably inside a production process. That integration gives ASML value capture. It also means one weak supplier can constrain the value of all the others. In a complex system, the bottleneck often decides the shipment.
Supplier dependency is also geopolitical. ASML's supplier footprint spans the United States, Europe and Asia. That gives access to specialized industrial regions but exposes the company to export controls, sanctions, local content demands, logistics risk and industrial-policy bargaining. Customers want capacity, governments want control and suppliers want enough confidence to invest. ASML stands in the middle. Its scarcity gives it leverage, but it also makes the company a target for policy pressure.
The resource-allocation question is whether ASML can expand without turning supplier support into a permanent drain. Spending to raise supplier capacity can be value-creating if it unlocks high-margin system revenue and installed-base cash flow. It is value-destructive if it funds capacity that customers later delay or regulators block. The supplier base is therefore a hidden balance-sheet item: not always visible as ASML's own plant and equipment, but economically tied to ASML's ability to satisfy demand.
Customers are few, global and state-sensitive
ASML's customers are not numerous in the way software customers or consumer-device buyers are numerous. The buyers capable of absorbing the most advanced lithography systems are large foundries, integrated device manufacturers and memory producers with multibillion-euro fab plans. This creates concentration risk even without a single public customer-percentage figure in the article. A small number of customers can drive order timing, product priorities, tool qualification and negotiating pressure. When those customers accelerate capacity for AI-related demand, ASML benefits. When they delay fabs, ASML feels it.
The 2025 geographic sales split also reveals dependence. ASML's annual-report highlights show EUR 28.1 billion of net sales in Asia, EUR 4.1 billion in the United States and EUR 0.5 billion in EMEA. That does not mean Asia is a single market; it includes different customer and policy environments. But it does show where the manufacturing center of gravity sits. ASML is headquartered in the Netherlands, but its revenue is tied to customers and fabs across Taiwan, South Korea, China, Japan, the United States and other regions.
Customer concentration changes the bargaining structure. ASML has technological leverage, but the buyer often has strategic leverage. A leading customer may influence roadmap priorities because its adoption decision validates a platform. A customer can also choose to extend existing tools, slow a fab ramp, mix DUV and EUV differently, or demand field-option performance before committing to broader deployment. ASML's sales are therefore co-produced with customer capital plans. The company cannot simply create demand by building more tools.
The state-sensitive nature of the customer base raises the stakes. Advanced chipmaking is now industrial policy, not just private investment. Governments fund fabs, restrict exports, scrutinize data flows and treat leading-edge manufacturing as strategic infrastructure. That can help ASML when subsidies support new fabs and customers sign long-term capacity agreements. It can hurt when export licenses, sanctions, technology-transfer rules or security concerns narrow the addressable market. ASML's Q1 2026 statement explicitly said the bandwidth in 2026 guidance accommodates potential outcomes of ongoing export-control discussions.
The customer question is therefore not only "how many systems will chipmakers buy?" It is "which customers are allowed, funded and willing to buy, and how much of their downstream demand is durable?" AI-related demand can justify accelerated capacity if cloud builders, device makers and enterprise customers keep buying advanced chips. If AI capacity spending slows, memory prices weaken or foundry customers delay node migration, ASML's scarcity still exists but near-term demand can soften. Scarcity is a strong negotiating position, not a guarantee of constant utilization.
Working capital matters before revenue arrives
ASML's business model requires commitments before revenue is fully realized. The company must fund research, supplier capacity, inventory, customer support and factory readiness in anticipation of customer demand. Q1 2026 shows how quickly cash can move even in a strong business. Cash, cash equivalents and short-term investments fell from EUR 13.322 billion at the end of Q4 2025 to EUR 8.376 billion at the end of Q1 2026. Some of that movement reflects ordinary timing, shareholder returns and operating investment, but it reinforces a simple point: the company is cash generative, yet it is not capital-light.
The timing burden is connected to tool complexity. A system sale is not completed the moment a purchase order exists. Modules must be produced, integrated, tested, shipped, installed, accepted and supported. A High NA tool adds another layer of qualification and customer learning. ASML's order book gives revenue visibility, but the working-capital load sits between the order and the cash. If customers accelerate demand, ASML may need to spend more ahead of revenue. If they slow demand, ASML may hold inventory, supplier commitments or support capacity for longer than planned.
Gross margin can hide this timing pressure. A 52% to 53% gross margin is excellent for complex equipment, but it does not tell the whole story. The economic return depends on how much R&D, supplier funding, inventory, receivables, field labor and customer-specific engineering were required to achieve that margin. Installed-base sales can improve the mix because service and field options monetize tools already in fabs, but they also require spare-parts positioning and expert labor close to customers.
ASML's ability to fulfill more than 99% of parts requests within 24 hours is valuable because downtime is expensive; it is costly for the same reason.
Shareholder returns make the trade-off visible. ASML returned EUR 8.5 billion to shareholders in 2025, including EUR 2.6 billion of dividends and EUR 5.9 billion of share buybacks. It also announced a new buyback program of up to EUR 12 billion through 2028. Returning cash can be rational when the business earns more than it needs to reinvest. But in a company facing High NA scaling, supplier expansion, export uncertainty and rising support obligations, capital return must not substitute for necessary reinvestment. The market rewards scarcity today; customers pay for capability tomorrow.
The working-capital test is practical. If ASML can convert demand into shipments without excessive inventory growth, support installed tools without bloated spares, and maintain high gross margin while funding R&D, the model remains highly attractive. If capacity commitments rise faster than accepted systems and field revenue, the model becomes more fragile. The company can afford heavy investment, but only if that investment keeps producing customer productivity gains that the customer cannot get more cheaply elsewhere.
Substitutes restrain price without removing dependence
ASML's moat is strong, but customers and governments keep looking for ways around it. The most realistic substitutes are not always direct competitors. They include extending older DUV tools, adding multi-patterning, changing chip architecture, using chiplets, relaxing design targets, delaying a fab, refurbishing mature tools, improving process-control software or accepting lower productivity. These substitutes can be inferior and still economically relevant. They discipline ASML's pricing because the customer compares a new tool against the best imperfect workaround.
ASML itself sells some of those workarounds. The DUV page says NXT systems can be upgraded to the latest technology through system node enhancement packages. The refurbished-systems page says almost every ASML lithography system sold is still in use at a customer fab and that mature PAS 5500 systems can be adapted for MEMS, through-silicon vias, radio-frequency chips, thin-film heads and LEDs. That is good for ASML because it keeps customers in its ecosystem. It is also a reminder that not every problem requires the latest system.
Unofficial and secondary market signals support the same caution. Reports about Chinese manufacturers upgrading older ASML DUV tools, industry discussion of fragmented Chinese alternatives and market commentary about analyst expectations for ASML's earnings are signals, not audited company facts. They should not be treated as proof of a new competitor capable of replacing ASML at the leading edge. They do show that customers and states will spend effort to stretch older tools and reduce dependence when access to newer tools is restricted. A monopoly-like position invites workarounds.
The direct competitive threat remains difficult. Recreating ASML's combination of EUV physics, precision mechatronics, optics, light-source knowledge, software, metrology, supplier depth and installed support would require enormous time and capital. Even if a rival develops a partial alternative, customers must qualify it in high-volume manufacturing. In advanced fabs, a cheap tool that damages yield is not cheap. The customer measures total cost per good die, not purchase price alone.
Still, the substitute question is what changes the ceiling on ASML's value capture. If customers can extend older tools for longer, ASML may still earn service and upgrade revenue but delay some new-system sales. If chiplets and advanced packaging reduce pressure on monolithic scaling, the mix of lithography demand can shift. If governments subsidize domestic alternatives for strategic reasons, those projects may persist even when commercial returns are weak. ASML does not need substitutes to be perfect to feel their effect. It only needs them to be credible enough to change customer timing or negotiating leverage.
Export controls turn orders into conditional revenue
Export controls are now part of ASML's business model, not an external footnote. The Netherlands, the United States and other allied governments have tightened restrictions on advanced semiconductor manufacturing equipment sold into China and other sensitive settings. ASML has repeatedly disclosed that guidance includes uncertainty around export-control discussions. In 2025, China-related demand and global AI capacity helped support results, but the future addressable market is partly a policy decision.
This changes the economics in three ways. First, it can remove revenue that would otherwise be technically and commercially available. A customer may want a tool, ASML may be able to build it and the price may be attractive, but the license may not be granted. Second, it can alter product mix. Restrictions on the most advanced tools may shift customer behavior toward older DUV, refurbished systems, service, parts or unauthorized attempts to extend equipment. Third, it can raise compliance and operating costs. ASML must monitor rules, classify products, manage service boundaries, screen customers and protect controlled knowledge.
The upside is that restrictions can reinforce scarcity among allowed customers. If advanced tools are restricted to a smaller set of trusted customers and regions, those customers may value access even more. Government subsidies for fabs in the United States, Europe, Japan or South Korea can support demand that might not have been built as quickly on purely private returns. The downside is that market access becomes less predictable. A company can have the best tool and still be unable to sell or support it in a specific setting.
Data locality and support control also matter here. Export rules do not apply only at the moment of shipment. They can affect software, service, spare parts, upgrades, employee access, remote diagnostics and knowledge transfer. ASML's service model depends on long-lived customer relationships. If regulations restrict what can be serviced, upgraded or remotely supported, the installed-base opportunity changes. Customers may also demand more local control over data and support processes as fabs become national assets.
The best management response is not to pretend the risk can be diversified away. It cannot. ASML's core technology is strategically sensitive because it is useful. The company must instead price, plan and communicate as if some orders are conditional until licenses, customer readiness and policy outcomes are clear. That makes the order book less simple than in a normal industrial cycle. It also raises the return threshold for every investment aimed at a restricted market. Scarcity remains valuable, but policy decides who can access it.
Judgment: value creation still has to outrun capital at risk
ASML's position is better than almost any ordinary equipment maker's. It has a technological bottleneck, a global installed base, high gross margins, deep customer support, major software capability and a supplier network that is hard to replicate. Its 2025 and Q1 2026 numbers show real value capture: high revenue, strong margins, significant net income and rising guidance. The company is not living on narrative alone. Customers are paying.
The question is whether they will keep paying enough to cover the rising capital at risk. On that point, the answer is constructive but not unconditional. ASML is most attractive when customers need capacity urgently, High NA proves economically useful, installed-base upgrades expand, service revenue compounds and export restrictions do not remove too much addressable demand. In that scenario, ASML's research and supplier spending are not a burden; they are the cost of owning the bottleneck.
The thesis weakens if the opposite facts arrive. The most important new facts would be a sustained slowdown in customer fab investment, weaker AI or memory demand, High NA qualification delays, gross margin pressure from supplier costs, service margins falling as support complexity rises, a sharper-than-guided China decline, broader restrictions on DUV service or parts, or evidence that customers can extend older tools longer than ASML expects without sacrificing too much yield. A real rival would matter, but a perfect rival is not required. Delayed adoption can damage returns before direct replacement appears.
The capital-allocation standard should be strict. ASML can spend heavily because it has earned the right to defend a rare position. But spending is justified only if it keeps customers economically dependent in a way that produces cash, not just admiration for engineering. The installed base helps because it turns each tool into a long support relationship. Software helps because it makes process control and productivity less separable from ASML. Supplier coordination helps if it unlocks scarce shipments. Shareholder returns are acceptable if they do not starve the next generation.
The conclusion is that ASML can preserve the economics of near-irreplaceable lithography, but only by remaining ruthless about the link between technology and customer payback. Scarcity alone is not enough. The company must keep proving that its tools reduce the customer's total cost per good die, shorten manufacturing cycles, protect uptime and enable nodes that would otherwise be delayed. If it does, capital intensity becomes the price of a superior franchise. If it does not, the same capital intensity becomes the mechanism by which an exceptional business slowly gives back part of its advantage.