A chipmaker can have strong demand for advanced processors and still struggle to manufacture them profitably. Printing billions of tiny features accurately, across many production steps, is one reason. EUV lithography sits at the center of that challenge. For investors, understanding it helps explain why a small group of equipment suppliers matters so much—and why rising demand for AI chips does not translate automatically into higher earnings for every semiconductor company.
What EUV Lithography Does
EUV stands for extreme ultraviolet. EUV lithography uses light with a wavelength of about 13.5 nanometers to transfer circuit patterns onto silicon wafers. Those patterns guide later manufacturing steps that build transistors and the connections between them.
Lithography is not the same as making a complete chip. A wafer repeatedly moves through deposition, exposure, development, etching, cleaning and inspection. Lithography determines where selected processes act. Other equipment creates or removes the actual materials.
The wavelength matters because shorter-wavelength light can help reproduce smaller features. Conventional advanced deep ultraviolet, or DUV, lithography uses 193-nanometer light. Manufacturers can extend DUV through immersion optics and multiple patterning, which divides a complex pattern across several exposures and processing steps. But extra steps add cost, time and opportunities for error.
EUV can replace some of those complicated sequences with fewer patterning steps. It does not eliminate DUV: modern fabrication plants use both, choosing the approach that makes economic and technical sense for each layer.
ASML (ASML) is the only commercial supplier of EUV lithography systems used in high-volume semiconductor manufacturing. Its position makes EUV unusually concentrated from an investment perspective. Several chipmakers use the technology, but the exposure through the machine supplier centers on one listed company.
Three distinctions keep the subject clear:
- Light wavelength is not transistor size. A 13.5-nanometer wavelength does not mean every printed feature measures 13.5 nanometers.
- A process-node name is not a literal measurement. Labels such as 3 nanometers describe a manufacturing generation rather than one universal physical dimension.
- Not every chip needs EUV. Many automotive, industrial, power-management and other semiconductors remain economical on older processes.
The investment question is therefore not simply whether chips become more advanced. It is how much additional EUV equipment customers need, when they need it, and whether that spending produces acceptable manufacturing costs.
How the Machine Turns Light Into a Pattern
An EUV scanner combines a light source, precision optics, a patterned mask and rapidly moving wafer stages. Each part has to function reliably enough for factory production, not just a laboratory demonstration.
From Tin Droplets to Developed Photoresist
The process begins with tiny droplets of tin. Laser pulses strike the droplets, creating a hot plasma that emits EUV radiation. A collector mirror gathers part of that radiation and directs it toward the scanner’s illumination system.
Ordinary lenses cannot handle EUV as they do visible light. Most materials absorb this radiation strongly, including air. The optical path therefore operates in a vacuum and uses specialized multilayer mirrors rather than conventional transmissive lenses.
The light reaches a reflective mask, also called a reticle, carrying the pattern for a particular chip layer. Projection mirrors reduce that pattern and direct it onto a wafer coated with photoresist, a light-sensitive material. The scanner coordinates the motion of the mask and wafer while exposing successive areas.
After exposure and development, selected areas of resist remain. Subsequent steps use that temporary pattern to guide etching or other processing. The resist is eventually removed, and the wafer continues through the manufacturing sequence.
| Component | What it does | Why investors should care |
|---|---|---|
| Tin-plasma source | Generates EUV radiation | Source power and reliability affect output and maintenance |
| Multilayer mirrors | Collect and direct EUV light | Optical losses make efficiency and cleanliness critical |
| Reflective mask | Carries a layer’s pattern | Mask defects can create repeated manufacturing defects |
| Photoresist | Converts exposure into a usable pattern | Sensitivity and random defects affect yield and throughput |
| Precision stages | Position the wafer and mask | Alignment errors can reduce the number of working chips |
These requirements explain why EUV is difficult to replicate. A competing system would need more than a suitable light source. It would need an integrated supply chain, precision manufacturing, software, installation expertise and years of operating experience inside customer factories.
They also explain why equipment uptime matters. A technically capable machine that frequently stops production can be economically inferior to one with slightly less ambitious specifications but steadier output.
Why EUV Changes Semiconductor Economics
The appeal of EUV is not that it makes every manufacturing step cheaper. EUV scanners are expensive, complex systems with substantial requirements for power, maintenance and factory infrastructure. Their value comes from improving the economics of selected advanced chip layers.
Consider a layer that otherwise requires several DUV patterning cycles. Each additional cycle can require materials, processing equipment, inspections and careful alignment. Replacing part of that sequence with EUV can shorten the process and reduce some sources of variation.
The relevant measure is cost per good chip, not simply equipment purchase price. A more expensive scanner can be worthwhile if it reduces enough other processing costs or increases the number of functioning chips produced from each wafer. Conversely, disappointing reliability or yield can weaken the case.
Taiwan Semiconductor Manufacturing Company (TSM), commonly called TSMC, uses EUV in advanced manufacturing processes sold to chip designers. Its customers pay for manufacturing capacity and technology; TSMC must turn its equipment spending into competitive pricing, strong yields and sufficient factory utilization. Our explanation of how TSMC makes money connects those manufacturing variables to revenue and margins.
For equipment investors, that creates an important distinction between customer demand and supplier earnings. Strong demand for advanced chips may encourage new factory investment, but orders can arrive well before production revenue. Customers can also delay equipment deliveries when existing capacity is sufficient.
EUV also does not solve every constraint in AI computing. Memory bandwidth, advanced packaging, power availability and chip design remain important. The separate role of memory is covered in our guide to why AI chips need high-bandwidth memory. A company can benefit from one bottleneck while remaining exposed to weakness elsewhere in the production chain.
The Main Players and Their Revenue Models
ASML provides the most direct listed exposure to EUV equipment. It earns revenue from new systems and from servicing, maintaining and upgrading its installed base. It also sells DUV systems and other products, so its total revenue is not a pure measure of EUV demand.
System sales are large and uneven. Service activity can provide a more recurring source of revenue, but it still depends on customer operations, contracts and the needs of the installed equipment. Investors should separate equipment revenue, service revenue, orders and backlog rather than treating them as interchangeable.
ASML depends on specialist suppliers. Carl Zeiss SMT supplies critical optics, while TRUMPF supplies important laser technology. Neither offers a straightforward, independently listed equity route into those specific businesses. This limits the number of direct public-market alternatives to ASML.
The principal manufacturing customers include TSMC, Samsung Electronics (KRX: 005930) and Intel (INTC). Their economics differ:
- TSMC manufactures chips for outside designers. EUV supports its ability to offer advanced processes, but equipment spending must be recovered through customer demand, pricing and factory utilization.
- Samsung Electronics combines memory, foundry operations and other businesses. Its stock reflects far more than EUV adoption, including memory pricing and consumer electronics performance.
- Intel combines processor products with manufacturing and foundry ambitions. Its results depend on product competitiveness, execution, factory economics and financing requirements—not merely equipment access.
Nvidia (NVDA) is a downstream beneficiary rather than an EUV equipment supplier. It designs processors and relies on manufacturing partners. EUV can support its product roadmap, but its earnings are more directly tied to chip demand, product pricing, software adoption and competitive positioning.
This distinction prevents a common analytical mistake: assuming every company associated with advanced chips provides the same exposure. The equipment supplier receives factory spending. The manufacturer must earn a return on that spending. The chip designer needs successful products and dependable manufacturing access. Each can perform differently during the same semiconductor cycle.
How Investors Can Assess Public-Market Exposure
There are three broad routes: the equipment supplier, manufacturing customers, or a semiconductor fund. None is a pure claim on the physical usefulness of EUV. Investors own businesses or portfolios whose results also depend on valuation, management decisions and industry conditions.
Direct ASML exposure places more emphasis on lithography spending, product execution and the installed service base. Manufacturing exposure adds greater sensitivity to yields, utilization, product mix and competition for customers. A semiconductor ETF spreads holdings across companies but may be dominated by chip designers rather than equipment makers.
Start an ETF review with its holdings, not its name. Check ASML’s weight, the combined allocation to equipment companies, geographic exposure and overlap with funds already held. Two semiconductor funds can have materially different exposure to the same manufacturing trend.
Morningstar (MORN) provides company and fund research that can support this comparison. Available coverage and portfolio features vary by product and subscription, so confirm that the relevant securities and exchanges are included.
For an individual company, build a short monitoring sheet:
- Revenue mix: How much comes from EUV systems, DUV systems, services or unrelated operations?
- Orders and backlog: Are customers adding commitments, changing delivery schedules or deferring projects?
- Profitability and cash flow: Are reported profits converting into cash, and how do customer advances affect that comparison?
- Customer spending: Do major manufacturers have enough demand and financial capacity to support their factory plans?
- Valuation: What growth and margins appear necessary to justify the share price?
Avoid comparing valuation ratios without checking business models. A manufacturer carrying heavy factory investment and an equipment supplier collecting customer advances can report very different cash-flow patterns.
Finally, consider portfolio overlap. A broad equity fund, semiconductor ETF and individual chip holdings may all depend on the same capital-spending cycle. Our guide to position sizing for high-risk technology stocks explains how to assess that shared downside rather than counting tickers as independent exposures.
What Could Weaken the Investment Case
EUV’s technical importance does not remove investment risk. A company can retain a strong competitive position while its shares disappoint because earnings fall short of expectations or the initial valuation assumed too much growth.
Semiconductor spending is cyclical. Customers can expand aggressively during shortages and later postpone projects as demand slows. Equipment orders may weaken before the change becomes obvious in reported sales. A large backlog provides information, but not certainty about delivery timing or future profitability.
Export controls affect the available market. Restrictions on advanced equipment can limit sales to particular countries and can change over time. ASML’s EUV systems have been restricted from shipment to China. Investors should distinguish restrictions on EUV from rules affecting particular DUV systems, upgrades or services rather than assuming all products face identical treatment.
High-NA EUV brings opportunity and execution risk. Numerical aperture describes an optical system’s ability to collect light and resolve detail. High-NA EUV raises this measure from 0.33 to 0.55, enabling finer patterning. However, adoption requires customers to justify equipment, process-development and infrastructure costs. Its smaller exposure field can also complicate manufacturing large chips, depending on the design and production approach.
Higher technical capability therefore does not guarantee immediate adoption across every advanced layer. Manufacturers will compare it with existing EUV approaches, including additional patterning where necessary. The timing of customer acceptance matters for both supplier revenue and customer capital efficiency.
Concentrated supply chains create operational exposure. Specialized optics, light-source components and precision assemblies are difficult to substitute quickly. Disruption at a critical supplier can affect deliveries even when end demand remains healthy.
The practical takeaway is to separate three judgments. First, EUV is an important enabling technology for advanced semiconductor production. Second, different companies capture that importance through different revenue models. Third, the price paid for those earnings determines how demanding the investment case becomes.
A useful review therefore follows the chain from chip demand to factory utilization, customer spending, equipment deliveries and cash generation. That approach gives investors a clearer basis for evaluating EUV exposure than relying on smaller node names or AI demand alone.