Financial Times writer June Yoon says Dutch kit king ASML is untouchable despite shipping only 40 of its most advanced chipmaking machines a year.
After an 80 per cent share price jump over the past six months, the Dutch supplier is valued at more than $500bn, which is a lot of swagger for a firm that only flogs a handful of machines.
Yoon argues the real power in the AI supply chain is not always with the chipmakers everyone bangs on about. Even elite foundries still fight it out the old-fashioned way, sweating fabs, scaling output and chasing yields while customers can still walk.
ASML is playing more like a one-team sport because it is the only company that can build extreme ultraviolet lithography tools for the most advanced chips. Each EUV machine starts at $220mn, and there is no commercially credible alternative supplier.
That sort of dominance should normally invite copycats, especially across Asia and the US, where chipmaking talent, cash and industrial policy are hardly in short supply. Taiwan leads in cutting-edge logic, Korea runs memory, Japan owns key materials, and China will spend whatever it takes to close a gap.
Yet the Financial Times points out that there is still no number two, and the usual excuses miss the point as “In EUV, second place just does not work.”
The reason is the tech chain is a string of absurd engineering tricks that all have to behave at once. EUV light does not occur naturally, so engineers blast microscopic droplets of molten tin with lasers, turning them into plasma hotter than the Sun’s surface.
That creates a pulse of extreme ultraviolet light, then optics take over. The beam is bounced through mirrors made with atomic-level precision, which take months to produce, before the patterns finally land on the silicon wafer.
Yoon said that the optics are the nightmare bit, and the EUV-grade mirrors come from a single supplier, Carl Zeiss SMT. Decades of tightly coupled work between Zeiss and ASML make it a club you cannot just buy your way into.
Even if a rival could copy the science, the economics are savage. A new entrant would sell too few machines to repay development, while being expected to deliver near-perfect reliability from day one.
Chip fabs run continuously, and downtime is the kind of “oops” that can turn into board-level panic. The Financial Times notes that at a company like TSMC, with annual sales of more than $120bn, a single day of lost output can cost hundreds of millions of dollars.
That makes “almost as good” EUV a non-starter, no matter how pretty the price tag could be. Chipmakers will not gamble volume production on unproven tools, which means a challenger will never gather the field data needed to improve.
ASML shipped its first EUV machine in 2006 and its first production-capable system in 2013, and the operating hours now run into the millions. The Financial Times links that real-world gap to why Nikon and Canon walked away from EUV more than a decade before it became commercially viable.
Governments love the fantasy that subsidies can magic up a rival, and China is the stress test, throwing money at domestic chipmaking tools after export controls cut Huawei off from advanced chips in 2020. The Financial Times is not buying the easy fix, arguing ASML shows that “beyond a certain technological threshold, markets no longer correct monopolies.”
