One Machine to Rule Them All
ASML Holding N.V., based in Veldhoven, Netherlands, is the sole supplier of extreme ultraviolet lithography machines in the world. Not one of the leading suppliers — the only one. No other company on the planet can build an EUV scanner that works at production volume. That's not because nobody tried. It's because the engineering challenges were so extreme that it took three decades of R&D and tens of billions of dollars to get here.
Their latest production machine, the TWINSCAN NXE:3800E, costs roughly $200 million per unit. The next generation — High-NA EUV (TWINSCAN EXE:5000 series) — runs closer to $380 million. TSMC, Samsung, and Intel are the primary customers, and they're all competing for allocation.
How EUV Actually Works
The core challenge with EUV is that 13.5nm light gets absorbed by essentially everything — including air. The entire optical path must operate in a near-perfect vacuum.
The Light Source
ASML's EUV source starts by firing a high-power CO₂ laser at tiny droplets of molten tin, each about 25 micrometers in diameter. The tin droplets fall at a rate of 50,000 per second. The laser hits each droplet twice — a pre-pulse flattens it into a pancake shape, then the main pulse vaporizes it into a plasma that emits EUV radiation. The conversion efficiency from laser power to usable EUV light is only about 5-6%. That's terrible by most standards, but it took years to get even that high.
For a related perspective, see Sensor Chip Design: CMOS Image Sensors, LiDAR Receivers, and.
The source generates about 250-300 watts of EUV power at the intermediate focus, which translates to throughput around 160 wafers per hour. For comparison, a mature 193nm immersion scanner does over 300 wafers per hour — EUV is still playing catch-up on raw throughput.
The Optics
Since EUV light can't pass through glass lenses, the entire optical system uses multilayer mirrors — alternating layers of molybdenum and silicon, each pair about 7nm thick, stacked 40-50 times. These mirrors reflect roughly 70% of the incoming EUV light each. With 11 mirrors in the optical train (6 in the illuminator, 4 in the projection optics plus the mask), only about 2% of the generated light actually reaches the wafer. Every mirror must be polished to sub-angstrom surface roughness. Carl Zeiss SMT in Oberkochen, Germany makes these optics — they're the only company that can.
Why Nobody Else Can Build One
It's tempting to think some well-funded competitor could just reverse-engineer an EUV machine. That fundamentally misunderstands the supply chain.
For a related perspective, see Chip Ô Tô Tự Lái 2026: Tesla vs Qualcomm vs Mobileye.
An EUV scanner has over 100,000 components sourced from thousands of suppliers across dozens of countries. The mirrors come from Zeiss (Germany). The laser from Trumpf (also Germany). The vacuum chambers, the wafer stages accurate to sub-nanometer positioning, the pellicle membranes that protect the mask, the contamination control systems — each of these represents decades of specialized know-how. ASML doesn't just assemble a machine; it orchestrates an ecosystem that can't be replicated.
Nikon and Canon, who compete with ASML in 193nm immersion lithography, both attempted EUV programs. Both abandoned them. The financial and technical barriers were simply too high. Canon reportedly spent over a decade on their effort before exiting.
The Investment Scale
ASML has invested over €10 billion in EUV R&D since the early 1990s. The initial research consortium, funded partly by Intel, included national labs like Lawrence Livermore and Sandia in the US, plus research institutions in Europe and Asia. The company now employs over 40,000 people globally, with thousands of engineers focused purely on EUV technology.
See also: Quantum Computing Chips in 2026: Superconducting, Trapped Io.
Geopolitical Weight
ASML's monopoly makes the company a geopolitical lever. The Netherlands, under pressure from the United States, has restricted exports of EUV machines to China since 2019. In 2023, restrictions expanded to include older deep-UV immersion lithography tools as well. This effectively caps China's domestic chipmaking at roughly 7nm-equivalent technology — possibly worse, since multi-patterning with DUV is extremely expensive and yields are lower.
China's largest foundry, SMIC, managed to produce a 7nm-class chip (the Kirin 9000S) using DUV multi-patterning, but the economics don't scale well. Without access to EUV, reaching 5nm or below in volume production is considered impractical by most industry analysts.
What Comes After EUV
High-NA EUV (numerical aperture of 0.55, up from 0.33) is the immediate next step, enabling finer patterning for 2nm and below. Beyond that, ASML is researching hyper-NA systems with even higher resolution. There's also been renewed interest in alternative patterning techniques — directed self-assembly (DSA) and nanoimprint lithography — but honestly, none of these are close to displacing EUV for leading-edge logic production. ASML's position looks secure for at least another decade.