ASML CEO on His Monopoly: No One Will Take the Crown

Manufacturing the chips that run the most demanding AI models today—from Claude Opus 4.7 to the latest-generation accelerators—depends on lithography machines that only one company in the world knows how to build. ASML, headquartered in Eindhoven, has held that monopoly for years, and its CEO does not appear to be losing sleep over it.

Last Tuesday, May 5th, Christophe Fouquet sat down with TechCrunch on the terrace of his Beverly Hills hotel, before his appearance at the Milken Institute Global Conference. Relaxed, blue suit, no apparent competitive anxiety. When the conversation turned to potential rivals, his message was clear: no one is in a position to threaten ASML's position in the EUV lithography market. Not in the short term. Probably not in the medium term either.

What Makes ASML Irreplaceable

ASML's EUV machines—and their most advanced version, High-NA EUV—are the physical bottleneck of the semiconductor industry. Without them, there are no process nodes below 7 nm at commercial scale. Each unit contains more than 100,000 components, requires years of coordinated supply chain work, and is delivered in dozens of cargo trucks and aircraft. The price of a single High-NA unit exceeds 350 million euros.

That complexity is precisely the barrier to entry. This is not a software advantage that someone could replicate in eighteen months. It is accumulated knowledge built over decades, interlocking patents, and relationships with specialized suppliers that do not exist elsewhere. Fouquet knows this, and he says it without euphemisms.

China has spent years trying to develop a domestic alternative through SMEE (Shanghai Micro Electronics Equipment), but sector analysts agree the technology gap remains at least one full generation—possibly two—behind what ASML offers today.

Why This Matters to Anyone Working with AI

For teams deploying or developing with Claude or other large LLMs, the chain running from ASML to Anthropic is shorter than it appears: ASML manufactures the machines → TSMC and Samsung produce the chips → Nvidia, Google and others design accelerators → data centers run them → models operate.

If that first link were to tighten—through additional export restrictions, a geopolitical accident, or simply limited capacity—the effect would ripple upward within months. It happened partially already with U.S. export controls to China in 2023-2024, which forced ASML to revoke sales licenses for its most advanced equipment.

Fouquet did not elaborate on export restrictions during the published interview, but the geopolitical context is inseparable from any conversation about ASML in 2026. The company operates in a space where trade policy can reshape its customer base from one quarter to the next, regardless of what competitors do.

The Paradox of a Comfortable Monopoly

There is something striking about the serenity with which Fouquet speaks of his company's position. Technical monopolies tend to generate two external reactions: regulation or massive investment to break them. In ASML's case, both forces exist—the EU watches, and China invests—but neither has yet produced a tangible result.

What has changed is pressure on delivery timelines. Demand for AI chip manufacturing capacity has driven orders through the roof, and ASML has had a backlog for several quarters. That is not a sign of a sleeping monopoly: it is a sign of a monopoly at the limit of its production capacity, attempting to scale a manufacturing process that by definition does not scale easily.

For the AI ecosystem in general, the practical conclusion is uncomfortable: the computing abundance that today allows running models with context windows of a million tokens ultimately depends on a single Dutch company manufacturing a few dozen machines per year.

---

Editor's View: That the CEO of a company with such structural power can say in public "no one will catch us" without sounding boastful says a great deal about the actual state of the sector. It is a situation that should worry anyone thinking about the long-term resilience of the AI supply chain, not just regulators.