Why ASML matters
ASML is the sole maker of extreme ultraviolet (EUV) lithography machines, the tools without which no advanced logic or memory chip on Earth can be manufactured. Every AI accelerator, every leading-edge smartphone processor, and every advanced automotive chip exists because a handful of ASML machines patterned the wafer it was cut from. That gives one Dutch company based in Veldhoven, a town of roughly 45,000 people, a degree of leverage over the global technology economy that is wildly out of proportion to its size - a chokepoint that TSMC, Samsung, and Intel all depend on and none can replicate.
The scale of that dependence shows up in ASML's 2026 numbers: full-year sales guidance raised twice to 43-45 billion euros, a backlog of 38.8 billion euros at the end of 2025, and a market capitalization of roughly $601B that makes it one of the most valuable companies in Europe. The AI buildup reshaping the technology industry runs, ultimately, through ASML's order book.
From a shed in Eindhoven to a chokepoint
ASML began on April 1, 1984, as a 50/50 joint venture between Philips and ASM International, tasked with commercializing a wafer stepper design Philips had developed internally. The new company built its first machine, the PAS 2000, in a shed next to a Philips office in Eindhoven. It nearly did not survive the industry's 1980s downturn: financial pressure forced ASM International to sell its half of the venture back to Philips in 1988, leaving the company's future genuinely in doubt for several years.
The PAS 5500 platform, introduced in 1991, turned the struggling venture profitable and began its climb against the then-dominant Japanese lithography makers Nikon and Canon. ASML went public on the Amsterdam and New York exchanges in 1995, giving it the capital to keep pace with an industry where each new lithography generation costs more to develop than the last. That capital intensity, more than any single technical breakthrough, is what eventually thinned the leading-edge lithography field down to three real players, and then to one.
The machine nobody else can build
EUV lithography fires a laser at molten tin droplets 50,000 times a second to generate a plasma that emits 13.5-nanometer light, then reflects that light off mirrors polished to atomic-scale precision, since ordinary glass absorbs EUV wavelengths outright. ASML spent more than two decades and billions of dollars turning this into a production tool, and its three biggest customers effectively bought into that development in 2012, with Intel alone contributing $4.1B for a 15% stake, because none of them could afford for EUV to fail.
The optics inside every EUV machine are made exclusively by Carl Zeiss SMT, a German firm in which ASML bought a 24.9% stake for about 1 billion euros in 2016 specifically to lock in the relationship. Zeiss cannot sell EUV-grade optics to anyone else, and no other optics maker in the world can match its precision, so the two companies function as a single production system that neither Nikon, Canon, nor any Chinese equipment maker has been able to replicate at the leading edge. The 2025 handoff of the first High-NA systems, at roughly $380M each, extends the same dynamic into the sub-2nm era: a production tool only ASML makes, using optics only Zeiss can grind.
A backlog business with a recurring tail
ASML does not sell into a spot market. Its EUV and High-NA systems are ordered years ahead of multi-year fab construction plans, which is why a 38.8-billion-euro backlog and a record 13.2-billion-euro quarter of Q4 2025 bookings matter more to the investment case than any single quarter's shipped-unit count. Roughly a quarter of revenue now comes from Installed Base Management - service contracts, spare parts, software, and performance upgrades sold against a fleet of thousands of machines already in the field - a recurring stream that keeps growing even when new-system orders slow.
That structure gives ASML unusual pricing power for a capital-equipment maker: it has pushed through price increases on mature DUV tools even as its largest customer, TSMC, pushes back, and it can afford to walk away from a fab that will not commit to a multi-year order because the queue of fabs that will is years long. The September 2025 decision to put 1.3 billion euros into French AI lab Mistral AI, becoming its largest shareholder, is a rare use of that balance-sheet strength for something outside the lithography business itself, and signals that ASML wants a stake in the AI stack its machines are built to serve, not just to supply it.
Two numbers investors watch: backlog and China
ASML's 2025 revenue was 32.7 billion euros; in 2026 it raised full-year guidance twice, most recently to 43-45 billion euros after a 9.3-billion-euro second quarter that beat expectations, with CEO Christophe Fouquet saying customers are pulling capacity plans forward because demand for chips is outpacing supply. The company plans to ship 65 low-NA EUV systems in 2026, climbing roughly 30% to 80-85 systems in 2027, alongside a similar percentage increase in DUV immersion capacity - an unusually aggressive ramp for a company whose tools take months to build.
The other number is China, which supplied 33% of ASML's revenue in 2025 but has fallen to roughly 20% of sales in 2026 as expanding US and Dutch export controls cut off EUV shipments and increasingly constrain DUV tools too. A bipartisan bill in the US Congress, the MATCH Act, would go further and ban servicing of DUV machines already installed in China. On July 27, 2026, a report that a Chinese state-linked firm had begun mass-producing a domestic immersion DUV tool briefly knocked ASML shares down as much as 8%, even though the tool was assessed as roughly 15 to 18 years behind ASML's current platform - a reminder of how sensitive the stock is to any sign that the China chokepoint could erode.
What to watch next
The near-term test is whether High-NA EUV, ASML's most expensive tool at around $380M per machine, finds a second real customer beyond Intel. Intel's July 2026 shipment of Panther Lake chips patterned with High-NA marked the first time the tool reached high-volume production, a milestone a full node ahead of Intel's own 14A timeline, but Samsung is reportedly holding its High-NA systems back from mass production to contain costs, and TSMC has ruled out High-NA adoption entirely through its next four process nodes, planned into 2029. Whether Samsung, SK hynix, or a memory maker chasing HBM capacity becomes the second real High-NA customer will decide whether the tool is a broad platform or one built mostly for Intel.
Longer term, the questions are political rather than technical: whether the MATCH Act or further Dutch export restrictions push China's contribution toward zero, and whether AI-driven capex proves durable enough to justify the 30% annual capacity increases ASML has committed to for 2026 and 2027. ASML reports Q3 2026 results on October 14, 2026, with guidance of 11-12 billion euros in net sales; a beat would confirm the AI buildup is still accelerating, while a miss would be the first real sign that customers' capacity plans have gotten ahead of actual chip demand.