Inside ASML's Lithography Dominance and Why It Justifies 51x Earnings

Understanding ASML's moat requires understanding the physics. Semiconductor lithography uses light to pattern circuits onto silicon wafers. Smaller circuits require shorter wavelengths of light. For decades, the industry used deep ultraviolet (DUV) light at 193 nanometres. EUV uses light at 13.5 nanometres, a wavelength so short that it is absorbed by virtually all matter, including air and glass lenses.

Building a machine that can harness 13.5nm light required ASML to solve problems that most physicists considered commercially impossible. The light source uses a laser that fires 50,000 pulses per second at tiny tin droplets, creating a plasma that emits EUV radiation. The optics use multilayer mirrors rather than lenses (because EUV light passes through nothing). The entire optical path operates in a near-perfect vacuum. Each EUV machine contains over 100,000 components, supplied by a network of specialist vendors that ASML has cultivated over three decades.

The development programme cost more than EUR 6 billion and took over 20 years. Nikon, the only potential competitor, abandoned its EUV effort in the mid-2010s after concluding the technology was too complex and the investment too large. Canon never seriously attempted it. ASML's position is not merely dominant; it is singular. There is no second source, and the barriers to entry are not just financial but physical and institutional.

The market focuses on new EUV system orders, which are lumpy and headline-grabbing. The more important revenue stream is the installed base. ASML has approximately 200 EUV machines in the field globally, plus thousands of DUV systems. Each EUV machine requires annual service contracts worth EUR 30 to 50 million, plus periodic upgrades that can cost EUR 20 to 30 million per machine.

The maths becomes compelling at scale. Two hundred EUV machines at an average service revenue of EUR 40 million generates EUR 8 billion in recurring revenue annually, before any new system sales. As the installed base grows toward 300+ machines by 2028, this service revenue stream alone could approach EUR 12 to 15 billion, providing a floor under total revenue regardless of order volatility.

This dynamic mirrors what happened in the jet engine industry. Rolls-Royce and GE generate the majority of their profits not from selling new engines but from servicing the installed fleet over decades. ASML's EUV installed base is entering its early adolescence; the oldest machines are only six to seven years old and have 20+ years of economic life remaining. The service revenue runway extends into the 2040s.

The comparison to enterprise software is also instructive. Microsoft's Azure generates recurring revenue from an installed base of enterprise customers with high switching costs. ASML's installed base has even higher switching costs because the customer literally cannot manufacture chips without the machines. Churn is zero. Retention is 100%. No software company can claim that.

ASML has begun shipping its next-generation lithography platform: High-NA EUV (numerical aperture of 0.55, up from 0.33 in current EUV). Each High-NA system costs approximately EUR 350 million, roughly double the price of a standard EUV machine. Intel was the first customer, taking delivery of the initial systems for its advanced process development.

High-NA enables chipmakers to continue shrinking transistor dimensions beyond the limits of current EUV technology. Without it, Moore's Law stalls. With it, TSMC and Intel can produce chips at the 1.4 nanometre node and below, enabling the next generation of AI accelerators, mobile processors, and data centre chips.

The revenue impact is substantial. If ASML ships 20 to 30 High-NA systems per year at EUR 350 million each, that adds EUR 7 to 10.5 billion in annual system revenue on top of the existing EUV and DUV business. Management has guided for revenue to reach EUR 44 to 60 billion by 2030, implying a CAGR of 10-18% from 2025 levels. Even the low end of that range suggests significant earnings growth.

The bear case on High-NA is that adoption will be slower than ASML projects because chipmakers can extend current EUV through design tricks and multi-patterning. That is a valid near-term concern, but history suggests that the semiconductor industry always adopts the most advanced lithography within 2-3 years of commercial availability. The economics demand it: more advanced lithography produces more chips per wafer, reducing the cost per transistor. The ROI calculation almost always favours upgrading.

The most significant risk to ASML is not competition (there is none) but geopolitics. The US, Netherlands, and Japan have implemented export controls restricting the sale of advanced lithography equipment to China. ASML can no longer ship EUV systems to Chinese customers, and restrictions on advanced DUV systems have tightened progressively.

China represented approximately 26-29% of ASML's revenue in 2024, primarily from DUV systems. The export controls have reduced this to an estimated 15-20% in 2025 as DUV restrictions tighten. The revenue impact is real but manageable because non-China demand for EUV systems exceeds ASML's production capacity. Every EUV machine that cannot be sold to China is sold to TSMC, Samsung, or Intel instead.

The longer-term risk is that China develops domestic lithography capability. Significant resources are being invested in this effort, but the technology gap remains enormous. ASML's EUV technology required 20 years and EUR 6 billion to develop, with deep collaboration across hundreds of specialist suppliers. Replicating that ecosystem from scratch would take a decade or more, even with unlimited funding. The physics has not changed, and the institutional knowledge resides almost entirely within ASML's supply chain.

Historically, technology export controls have delayed but not prevented capability development. Japan's semiconductor equipment industry overcame US restrictions in the 1980s, though it took 15-20 years. China may follow a similar path, but the timeline provides ASML with a multi-decade window of unchallenged dominance.

The artificial intelligence infrastructure build-out is the single largest demand driver for advanced semiconductors in history. Every AI training cluster requires thousands of Nvidia, AMD, or custom-designed chips manufactured on TSMC's most advanced processes, which in turn require ASML's EUV lithography.

ASML does not sell directly to AI companies, but the chain of demand is unambiguous: more AI spending means more chip demand means more EUV machine orders. TSMC's capital expenditure guidance of approximately $38 to $42 billion for 2026 is heavily weighted toward advanced process capacity that requires EUV. Samsung and Intel are making similar, if smaller, investments.

The AI demand is structural rather than cyclical. Training frontier models requires exponentially more compute each generation. Inference workloads scale with deployment. Edge AI adds a new demand vector for advanced chips. Each of these trends drives more EUV lithography demand over the next decade.

The stock's beta of 1.38 reflects this connection to the broader semiconductor and AI cycle. When sentiment on AI spending turns negative, ASML sells off harder than the market. When it recovers, ASML rallies harder. The 52-week range of $609 to $1,547 captures both extremes. At $1,478, the stock sits near the upper end of its range, which makes valuation the critical question.