ASML Won the Lithography War by Making Its Rivals Co-Own the Only Road Forward

In the summer of 2012, ASML did something a lithography company is not supposed to do: it sold pieces of itself to its own customers. Over seven weeks, Intel, TSMC, and Samsung — three companies that buy ASML's machines and could, in principle, fund someone else's — wired in money for equity. Intel went first on July 9, agreeing to roughly 10 percent of ASML's shares with a right to buy more.¹ TSMC followed on August 5 for a 5 percent stake.² Samsung closed it out on August 27 for 3 percent.³ Together they bought 23 percent of ASML for EUR 3.85 billion.³ On paper it was a funding round. In fact it was the moment the lithography standards war ended — though almost nobody noticed it was being fought.

The official story is that ASML won EUV with better physics. It out-engineered Nikon, out-engineered Canon, and the market rewarded the superior tool. That story is half true and entirely beside the point. ASML did build the better machine. But superior technology is not what made the win permanent. What made it permanent was a capital structure nobody could route around.

A standards war needs two armies. ASML quietly disarmed one.

Extreme ultraviolet lithography is not a product you buy off a shelf. It is a decade-long, multi-billion-dollar bet on a light source so difficult that ASML had been studying the shift since 1997 and developing the technology in earnest from around 2000.⁷⁸ A bet that size only pays off if the world's chipmakers commit to it as the standard — if Intel, TSMC, and Samsung agree to build their next process nodes around your machine and no one else's. And here is the strategic trap: those same chipmakers are the only entities on earth with the cash and the motive to fund a competing supplier. They want a second source. A monopoly supplier of the single most critical tool in their factory is their nightmare. So the natural move for them was to keep Nikon or Canon alive as a hedge.

ASML's 2012 program made that hedge irrational. The moment Intel, TSMC, and Samsung held equity in ASML — and had committed EUR 1.38 billion in R&D funding on top of it³ — every dollar they might have spent propping up a rival became a dollar spent devaluing their own stake. They were no longer customers shopping for a second source. They were co-owners with skin in ASML's roadmap. The incentive to back an alternative standard didn't lose the argument. It simply stopped existing.

Then, two months later, ASML closed the other door. On October 17, 2012, it announced it would acquire Cymer — the maker of the laser-produced plasma light source that EUV runs on — in a cash-and-stock deal valued at EUR 1.95 billion.⁴ This is widely misremembered as ASML buying its way into EUV. It wasn't. Cymer was already ASML's key supplier for the LPP architecture ASML had chosen precisely because it promised the uptime and power scaling EUV demanded.⁸ Buying Cymer was vertical integration of a relationship that already existed — ASML pulling the single hardest, most failure-prone component of the EUV chain inside its own walls so no rival could rent it. With its customers as co-owners and its light source as a subsidiary, ASML had assembled the only complete EUV supply chain in the world.

Where was the other army the whole time?

Here the popular telling overstates the drama. There was no head-to-head EUV showdown in which chipmakers voted for ASML over a worthy Japanese rival. ASML had already passed Nikon and Canon a decade earlier on the strength of ordinary deep-ultraviolet machines: in 2001 it held about 28.8 percent of the lithography market behind Nikon's 39.4 percent and Canon's 31.2 percent, and by 2002 it had vaulted to 54.3 percent, becoming the world's largest supplier.⁵⁷ By the time EUV mattered, Nikon and Canon were not gathering forces for a counterattack. They were retreating. The contest was settled less by a battle than by an absence — the rivals' failure to commit the capital and the partners that EUV demanded, at the very moment ASML was locking those partners up.

Isn't this just a great product winning? The honest counter.

The fair objection is that none of the financial engineering would have mattered if the machine didn't work. And that's true — ASML's first product, the PAS 2000, was a flat-out failure, and the company only found its footing with the PAS 5500 in the early 1990s.⁷ EUV is the hardest manufacturing problem in modern industry, and ASML spent roughly EUR 4.0 billion on R&D in 2023 alone, recognizing revenue on just 53 EUV systems that year.⁸ You cannot co-invest your way to a working light source; you have to build it. So yes — technology was necessary. The point is that it was not sufficient. Plenty of superior technologies have been strangled by a better-funded competing standard. What the 2012 program did was guarantee that ASML's superior technology would never face one. The capital structure didn't replace the engineering. It insured it — by removing the only people who could have funded the alternative.

There is a founding anecdote ASML still tells about itself — a small team starting out in a leaky shed in Eindhoven. It is good lore, and it flatters the romance of the better mousetrap. But the company's defining move was not made in a shed. It was made in three SEC filings over seven weeks in 2012, when ASML looked at the only firms capable of arming its enemy and handed each of them a share of itself instead. The lithography standards war is usually told as a story about light. It was really a story about ownership. ASML didn't out-shine Nikon and Canon. It out-structured them — and then bought the one bulb everyone else would have needed.