ASML's 2030 order book is the canary in the AI hardware coal mine

ASML Is Selling Machines That Don’t Exist Yet

ASML handed investors something rare at its Wednesday earnings call: visibility. Some customers are already expressing interest in placing orders for the company’s Extreme Ultraviolet (EUV) lithography machines for 2029 and even 2030. Christophe Fouquet, ASML’s CEO, didn’t mince words about what this means.

“Our customers continue to increase their short- and medium-term equipment demand forecasts with us, driven primarily by their views on the size and duration of AI-related investments,” Fouquet said. “In light of their demand visibility, some customers are expressing an interest in placing orders with us for 2029 and even 2030.”

This is not a reassurance. It is a confession. ASML, the only company on Earth that produces EUV tools, cannot keep up. Supply planned for 2027 and 2028 will not fully meet demand. The company is investigating ways to increase capacity beyond the 30% boost it had previously forecast. In July, ASML raised its full-year sales and gross margin guidance. This quarter, it narrowed the revenue range upward—now expecting between 44 billion and 45 billion euros in annual sales, up from a previous lower bound of 43 billion. Shares have rallied 56% year-to-date.

The market was looking for confirmation that AI-driven chip demand wasn’t a flash in the pan. ASML just handed it proof. And the consequences stretch far beyond Veldhoven, the Dutch town where ASML’s headquarters sits.

Who’s Actually Ordering Machines For 2030?

Fouquet wouldn’t name names, but the roster is obvious. Samsung and TSMC committed in July to using ASML’s High NA EUV tools. These are the machines that will print the chips running every major foundation model, inference platform, and data-center workload for the next half-decade. High NA EUV isn’t incremental—it’s a generational leap in resolution, enabling the gate-all-around transistors and 2-nanometer-class process nodes that both companies need to stay ahead of Moore’s Law.

TSMC is the world’s largest foundry, fabricating Apple A-series chips, NVIDIA GPUs, and AMD processors. Samsung is simultaneously a memory powerhouse and a foundry player chasing TSMC’s leadership. Both are betting billions on processes that require ASML’s most advanced tools—tools ASML itself struggled to deliver on time for years.

When Fouquet says customers want 2030 machines now, he’s describing a supply chain in panic mode. These aren’t speculative orders. They’re insurance policies against being the only fabs without cutting-edge capacity when the next AI training workload doubles in size.

Korea’s Fab Bet Just Got Realer—and Riskier

South Korea is the second-largest economy in East Asia and home to Samsung Electronics and SK Hynix. Together, they dominate global memory chip production and are investing aggressively in advanced logic manufacturing. ASML’s 2030 order visibility changes the calculus for both companies.

For SK Hynix, the memory maker behind Hyundai Motor and a critical supplier of HBM (High Bandwidth Memory) for NVIDIA’s GPU stacks, ASML’s signals validate the company’s decision to invest heavily in advanced packaging and process development. HBM demand is exploding alongside AI server builds, and memory cycles move on 18-month to two-year horizons. SK Hynix needs to commit now or risk missing the next upcycle.

For Samsung Foundry, the implications are sharper. Samsung has been losing ground to TSMC in advanced-node adoption. Every wafer Samsung doesn’t produce at TSMC is revenue and market share ceded to a rival that already dominates the AI accelerator space. The company’s Ulsan Fabs—its largest semiconductor complex—require massive capital expenditure to expand. ASML’s capacity constraints mean Samsung can’t simply order more machines whenever it wants. It has to book them years in advance, locking in cash commitments before it knows whether its customers will still be buying.

This is the paradox of capital-intensive manufacturing during a technology inflection: the worst time to invest is when you’re most uncertain, and the best time is when everyone else is also investing, which means you can’t. ASML is telling both Korean champions that the window is open—but it won’t stay open forever.

The Supply Chain Crunch Nobody’s Pricing In

ASML’s earnings beat tells one story; its capacity planning tells another. The company expects fourth-quarter sales between 14.2 billion and 15.2 billion euros, with the upper end ahead of LSEG consensus of 14.2 billion. Net profit came in at 4.96 billion euros versus 4.1 billion expected. The numbers are strong. But the guidance reveals stress underneath.

ASML’s previous forecast assumed a 30% capacity increase by 2028. Now it’s “investigating ways” to go beyond that. That language matters. It means the company doesn’t yet have a plan—it has a problem. Building EUV lithography machines requires a supply chain that spans multiple continents, thousands of suppliers, and decades of accumulated know-how. You can’t fast-track that.

The bottleneck isn’t just ASML. It’s the companies that feed ASML: Trumpf, the German laser maker that produces the pulse sources for EUV tools; Cymer, ASML’s own subsidiary that makes the excimer lasers; Zeiss, the optics manufacturer that produces the mirrors inside ASML machines to sub-nanometer tolerances. Each of these firms faces its own capacity constraints. Trumpf is building a new facility in Germany to meet EUV laser demand. Zeiss is expanding its Jena plant. None of this happens overnight.

China’s DUV Ambition Is a Non-Starter—For Now

The earnings report included one notable detail: ASML is sticking to its forecast that China will account for 20% of total business. That’s significant because U.S. export controls have restricted ASML from selling EUV tools—and some DUV (Deep Ultraviolet) equipment—to China. The company has been navigating these restrictions carefully, and the 20% figure reflects that reality.

A July report from The Information suggested a Chinese company had begun manufacturing an immersion DUV lithography machine, citing people familiar with the matter. Analysts raised questions about whether China could realistically ramp production of its own DUV tools. The gap between immersion DUV and EUV is enormous—EUV operates at 13.5-nanometer wavelengths, while immersion DUV works at 193 nanometers. The physics alone make replacement impossible in the near term.

China’s semiconductor push, embodied in the Big Fund and state-directed R&D, is real. But it’s aimed at mature-node self-sufficiency, not cutting-edge lithography. ASML’s 20% China forecast acknowledges this: Beijing will buy what it can, but it can’t buy everything it wants.

What Happens When Demand Actually Arrives

Here’s the scenario nobody wants to discuss: What if ASML can’t increase capacity fast enough? What if the 30% boost falls short? What if Trumpf, Zeiss, and the rest of the supply chain can’t scale in sync?

The answer is already visible in pricing and allocation. ASML machines sell for $200 million to $350 million each, depending on configuration. High NA EUV tools are at the top end. Foundries that secure allocation get first dibs on capacity; those that don’t face delayed product launches, missed market windows, and competitive disadvantage. Samsung and TSMC are competing in the same space with the same equipment constraint. That competition will intensify, not ease.

For investors, the takeaways are straightforward. ASML’s moat is intact—the company is the sole producer of EUV lithography, and no credible challenger is emerging. The AI infrastructure build-out is multi-year, not multi-quarter. Demand visibility through 2030 suggests the capex cycle has further to run.

For Korea, the message is more complicated. Samsung Foundry and SK Hynix are both dependent on ASML capacity that may not materialize fast enough. Both companies need to secure machine allocations now or risk falling behind. The risk isn’t just commercial—it’s geopolitical. South Korea’s semiconductor industry is the backbone of its economy, and its positioning in the global AI supply chain depends on access to the same tools as everyone else.

The Bottom Line

ASML’s 2030 orders aren’t just good news for shareholders. They’re a signal that the AI hardware build-out has more runway than most analysts assumed. The question isn’t whether demand exists—it’s whether the supply chain can keep up. And that uncertainty is where the real risk lives.

For Korea’s fab ecosystem, the window is open. But it won’t stay open forever.