The Entire AI Industry Runs Through One Stumbling Company

The most powerful company in the AI race doesn’t build models, train algorithms, or ship software. It makes what they all depend on. ASMLASML

is the sole supplier of the machines behind nearly every AI chip. ChatGPT, Gemini, and Claude all depend on it — so when ASML stumbles, the world feels it.

• MicrosoftMSFT
  , MetaMETA
  , and peers are collectively spending $600B+ on AI infrastructure this year — and that flows upstream to the EUV lithography monopolist.
• To keep pace, ASML is ramping production 36% this year to at least 60 machines — while investing ~$2.2B across new US, European, and South Korean facilities.

It’s not enough: Even with a 20% bump in capital expenditures, money can’t simply buy precision at speed. Each school-bus-sized EUV machine requires months to build, and draws on hundreds of specialized suppliers — making rapid scaling genuinely hard. Normally, a supply crunch is a pricing opportunity, but at ~$400M, ASML’s next-gen machines haven’t found many takers. When the world’s most important supply chain runs through one bottleneck, “we’ll figure it out” isn’t exactly reassuring.