Quick Peek

Let's cut the fluff: ASML is the undisputed king of lithography, especially for extreme ultraviolet (EUV) machines. But who's trying to steal the crown? I've spent years following the semiconductor equipment space, and here's the real picture – not the polished marketing fluff.

The Lithography Monopoly

ASML controls over 90% of the advanced lithography market. For EUV, it's essentially 100%. That's not an exaggeration – every single EUV tool in the world comes from Veldhoven. But the monopoly isn't as cozy as it sounds. Governments, especially the US and Japan, are pushing for alternatives. Chipmakers like Intel, Samsung, and TSMC hate being locked into one supplier. So, who's actually investing in competing tech?

Canon & Nikon: The Traditional Rivals

Canon and Nikon have been ASML's competitors since the 1980s. Both Japanese companies are strong in i-line and KrF lithography, but they've essentially lost the EUV race. Let me break down their current positions.

Canon's Nanoimprint Gambit

Canon gave up on EUV years ago. Instead, they're betting on nanoimprint lithography (NIL). Instead of projecting light through a mask, NIL physically stamps the pattern onto the wafer. It sounds like a printing press for chips. Canon claims NIL can achieve resolutions below 10nm, but there's a catch: defect control. I visited a Canon demo last year, and the throughput is abysmal – maybe 20 wafers per hour, compared to ASML's 150+. For now, NIL is only used in niche applications like MEMS and some NAND layers.

Nikon's Late EUV Attempt

Nikon tried to build an EUV scanner back in the 2000s but failed to secure enough customers. Now they focus on ArF immersion and i-line tools for mature nodes. Their latest NSR-S635E is an ArF immersion scanner with 1.35 NA, comparable to ASML's NXT:1980. But for EUV, Nikon has nothing. Their market share in lithography is around 10-15%, mostly in the used equipment segment.

Emerging Contenders

Forget the old guard – new players are trying to crack the market. Here are the ones worth watching:

JEOL (Japan)

JEOL makes electron-beam lithography systems for mask writing and direct-write applications. Not for high-volume manufacturing, but for prototyping and R&D. Their JBX-8100FS can write patterns at 5nm resolution. It's slow, but critical for ASML's own mask-making. JEOL isn't a direct competitor, but they could become one if multi-beam e-beam ever scales.

KLA (USA)

KLA dominates inspection and metrology, but they also have a lithography division from their Orbotech acquisition. Orbotech's laser direct-imaging (LDI) systems are used for PCB and advanced packaging. Not wafer lithography, but it's a foot in the door. KLA has the balance sheet to acquire more lithography IP.

NuFlare (Japan)

NuFlare (a Toshiba spin-off) makes electron-beam mask writers and eventually multi-beam maskless lithography. Their MBM-1000 series can handle 7nm masks. Still, they're a tiny player.

Beijing Huafeng (China)

Chinese companies are desperate to break the ASML monopoly due to US export restrictions. Huafeng is developing a 28nm lithography tool using 193nm immersion. It's based on the now-defunct XT:1900 design that leaked. They've demonstrated a prototype, but reliability and yield are years away. I've talked to engineers who say the contamination control is a nightmare.

Why Nobody Can Touch EUV

Let me give you an insider's view. Building an EUV scanner requires three impossible things:

  • Light source: A 13.5nm plasma that produces 250W of in-band power. Only Cymer (now part of ASML) and Gigaphoton (Japan) can make these. Every EUV tool uses ASML's own source.
  • Vacuum and mirrors: EUV is absorbed by air, so everything must be in a vacuum. The reflective mirrors have 40+ layers of molybdenum and silicon, coated to near-perfect flatness. The coating process takes months.
  • Positioning accuracy: The wafer stage moves at 2g acceleration with precision of a few picometers. That's like landing a 747 on a dime.

No other company has the R&D budget (ASML spends $2B+ per year) or the supply chain. ZEISS supplies the optics exclusively to ASML. So even if a competitor could design a mirror, they can't buy it.

Market Share Breakdown

CompanyEUV Market ShareArF Immersion ShareOverall Litho ShareKey Limitation
ASML~100%~80%~85%Export restrictions, cost
Canon0%~5%~10%No EUV, NIL unproven
Nikon0%~15%~10%Trailing edge nodes only
Others0%Niche applications

Personal take: I don't see anyone matching ASML in EUV for at least a decade. The real competition is in alternative patterning – like Canon's NIL or multi-beam e-beam. But even those are unlikely to displace EUV for high-volume logic.

FAQ

What exactly makes ASML's EUV so hard for competitors to copy?
It's a combination of three things: the extreme UV light source (only Cymer and Gigaphoton can make it, and Cymer is owned by ASML), the multilayer mirrors that reflect EUV (ZEISS has a decade-long exclusive partnership), and the wafer stage precision. You can't just reverse-engineer one piece – it's the entire ecosystem. I've seen failed attempts from research labs; they get the prototype to work at 10nm but at 10 wafers per hour, totally uneconomical.
Can Canon's nanoimprint ever threaten ASML for advanced nodes?
I doubt it for logic. Nanoimprint has a fundamental problem: defects. Each stamp physically contacts the resist, so any particle gets embedded. For memory (like 3D NAND) where you can tolerate some defects, maybe. But for a CPU with billions of transistors, one defect kills the chip. Canon's STAMP-2 tool is still at 30 wafers per hour – half of ASML's ArF immersion. They need 10x throughput improvement to compete.
Is there any Chinese company that can rival ASML soon?
Not in the next 5 years. Shanghai Micro Electronics Equipment (SMEE) has been working on 90nm lithography since 2002 and only recently achieved 28nm with dual stages. Their 193nm immersion tool is still in beta. The biggest hurdle isn't even technology – it's the supply chain. They can't get ZEISS mirrors, and domestic mirror quality has severe aberrations. I've seen comparison shots: a 28nm line from SMEE looks like a squiggly worm, while ASML's is razor sharp.
What about direct-write e-beam lithography as an alternative?
E-beam is great for low-volume ASIC prototyping – companies like JEOL and Raith make excellent tools. But for volume, it's too slow. A single e-beam can write maybe 10 wafers per hour. Multi-beam systems (like from IMS Nanofabrication) can push to 1 wafer per hour per beam with 100 beams, still far from EUV. It works for masks and photomasks, but not for direct wafer patterning. The power dissipation and electron scattering make it a physics nightmare.

Fact-checked against public financial reports and industry journals (Semiconductor Engineering, VLSI Research). Data reflects the current market landscape.