The MoatWide moat

ASML is, to my eye, one of the widest and most extraordinary moats in the whole of business — a company that has achieved something almost unheard of: a genuine, durable monopoly on a technology the entire modern world depends on. Every advanced computer chip — the processors that run artificial intelligence, smartphones, data centers, and the machinery of the digital economy — is printed using a process called photolithography, and the most advanced chips can only be made using extreme-ultraviolet (EUV) lithography. ASML is the only company on earth that can build an EUV machine¹. Not the best; the only. If you want to manufacture a cutting-edge chip anywhere in the world, you must buy your lithography machine from this one Dutch company, and there is no alternative at any price. That is not a strong competitive position. That is a chokepoint on the future of computing.

The scale of the moat is easiest to grasp by contemplating what it would take to compete. An EUV machine is arguably the most complex device humanity has ever mass-produced — the size of a bus, made of hundreds of thousands of parts, firing molten tin droplets fifty thousand times a second and vaporizing them with a laser to produce light at a wavelength so short it must travel through a vacuum and bounce off the flattest mirrors ever manufactured. It took ASML more than two decades and tens of billions of euros, in partnership with a web of specialized suppliers, to make it work. The knowledge is spread across the company, its physics, its supply chain, and the muscle memory of thousands of engineers, and it cannot be bought, copied, or hurried. A rival starting today, with unlimited money, would be a decade or more behind and might never arrive. This is why no one has: ASML's EUV monopoly is not a temporary lead but a structural fortress.

Layered on the monopoly is a technology-and-supply-chain moat that deepens it. ASML does not make everything itself; it is the systems integrator at the center of an ecosystem — Zeiss makes the impossibly precise optics, Cymer (which ASML owns) makes the light source, and thousands of suppliers make the rest. ASML orchestrates the whole, and that orchestration, refined over decades, is itself irreplaceable. The company spends billions a year on research to stay ahead, and each generation of machine — now the next-generation High-NA EUV, at over €350 million apiece — extends the monopoly forward to the following node of Moore's Law. The lead does not erode with time; it compounds.

And beneath the machines sits a beautiful second business: the installed base. ASML's machines, once sold, run for years and require a constant stream of service, spare parts, software, and performance upgrades — a recurring, high-margin annuity that grows with every system shipped. This 'Installed Base Management' business generated €8.2 billion in 2025 and is growing at over 25% a year; it is the razor-and-blades model applied to the most expensive razors on earth, and it turns a lumpy equipment maker into something with a large, sticky, recurring core.

The economics that flow from all this are what you would expect of a monopoly on an essential technology: gross margins above 50%, a return on capital that runs into the forties, prodigious free cash flow (€11 billion in 2025), and an order backlog stretching years into the future. Demand, driven by the artificial-intelligence build-out, is accelerating: in 2026 the company twice raised its guidance, now expecting €43–45 billion in sales, and it targets far more by 2030. Its handful of customers — TSMC, Samsung, Intel — are utterly dependent on it, and it on them, a co-dependence that locks the relationship in both directions.

The honest cautions are real and worth stating plainly. ASML sits at the center of the US–China technology war: it is forbidden to sell its most advanced machines to China, its sales there, 36% of the total in 2024 and 29% in 2025, fell to 16% in the first half of 2026 under export controls², and China is straining to build domestic alternatives. The semiconductor industry is cyclical, its customers are few, and the whole thesis now leans heavily on the AI boom continuing. And the stock, having roughly re-rated in that boom to over fifty times earnings, is priced for a great deal of future success. But none of these threats touches the core fact: ASML makes the one machine the future of computing cannot be built without, and no one else can. That is the definition of a wide moat, and ASML holds the widest version of it technology has to offer. The number that would signal trouble is net bookings across a cycle: the monopoly guarantees ASML the order whenever the industry builds, so bookings falling out of step with fab construction would mean one of only two things — a rival EUV tool at last (none exists, none is close), or customers learning to make do with fewer of ASML's machines per wafer. Either would show up in the bookings line before anywhere else.