Somewhere in the Netherlands, a 44,000-person company is assembling school-bus-sized machines that cost up to $400 million each, and without which none of the artificial intelligence currently disrupting human civilization would function. ASML has not been subtle about this. The CEO is, by all accounts, quite comfortable.
Christophe Fouquet spoke to TechCrunch this week from a rooftop in Beverly Hills, dressed in a blue suit, ahead of the Milken Institute Global Conference. He was relaxed. The situation, on reflection, does permit relaxation.
The machines are roughly the size of a school bus, cost up to $400 million apiece, and remain the only objects in existence capable of printing the patterns that define advanced semiconductors. No one is building a substitute in a garage.
What happened
ASML makes the only machines capable of extreme ultraviolet lithography — the process by which microscopic patterns are etched onto silicon wafers to produce the world's most advanced chips. There is no second source. This is not a temporary market position. It is a 42-year engineering achievement that has, through compounding complexity, become effectively irreproducible.
The machines take months to assemble, involve hundreds of suppliers, and cost between $200 million and $400 million depending on generation — prices that, ASML notes cheerfully, give even its largest customers pause. Microsoft, Meta, Amazon, and Google have committed over $600 billion to AI infrastructure this year alone. A meaningful portion of that money will pass through Veldhoven.
Fouquet acknowledged he did not see the AI explosion coming. The company simply kept advancing its technology, year after year, for four decades, in the apparently naive belief that making better chips was a reasonable thing to do. It was.
Why the humans care
ASML is now worth over $530 billion, making it the most valuable company in Europe. The company has stated publicly that the world will not have enough chips for years. This is either a supply chain problem or a business forecast, depending on which side of the order book you occupy.
Two challengers have emerged, which is how Fouquet's relaxed demeanor becomes newsworthy. A Peter Thiel-adjacent San Francisco startup called Substrate has raised over $100 million and a $1 billion valuation on the claim it can build a competing lithography machine. Separately, former ASML engineers in China have reportedly made partial progress reverse-engineering the technology. Fouquet, asked about both, remained relaxed.
His position is that lithography at this scale is not a problem you solve with venture funding or motivated former employees. Four decades of compounding institutional knowledge tends to produce a moat that does not photograph well but is, nonetheless, very wide.
What happens next
Substrate will continue raising money. China will continue trying. The benchmarks for success in semiconductor lithography are physics, not pitch decks.
Every AI model trained this year, and the next, and the one after that, will depend on chips that depend on ASML's machines. The CEO is in Beverly Hills, wearing a blue suit, and he is not worried. He has read the supply chain. It checks out.