Not long ago, Roger Dassen, CFO of Dutch lithography equipment manufacturer ASML, revealed the company’s strategic cards during a conference call in front of all of Europe.
The executive said: "We have almost no sales in Europe, and as other countries heavily invest in their domestic semiconductor industries, Europe risks falling even further behind." He immediately added: "The U.S. has asked if we can move half of our R&D there, while China and India are extending enthusiastic invitations."
Europe's largest tech company by market cap manufactures the world's only EUV lithography machines, yet none of its customers are European chip manufacturers.
ASML sells EUV lithography machines, with each bare machine priced between $150 million and $300 million; the next-generation High-NA models have already been quoted at $400 million per unit.
This can be considered the most expensive and complex single machine in human industrial history—an EUV machine has 100,000 components, 3 kilometers of piping, and hundreds of thousands of sensors; transporting it requires a specially modified Boeing 747, and installation and calibration at the factory take a full year.
Only a handful of chip manufacturers worldwide can afford, dare to buy, and use EUV lithography machines: TSMC, Samsung, and Intel. These three have secured ASML’s entire capacity for the next three years.
ASML itself states that it will deliver approximately 65 Low-NA EUV machines in 2026, with its full-year net sales guidance raised to €43–45 billion. In 2027, EUV lithography machine capacity will approach full utilization, with an additional 30% expansion to 85 units. By 2028, a significant number of orders are already in place.
The CFO is now more concerned about reducing the build time per unit from 22 weeks to 15 or 16 weeks than about sales. In this seller’s market with supply falling short of demand, how many EUV machines has it sold in Europe? The answer: zero.
Europe doesn't lack chips; it lacks advanced manufacturing processes.
Europe has ASML and Belgium's IM. EC With Infineon, NXP, and STMicroelectronics, how could not a single EUV machine be sold?
The answer is: The chips these companies produce don't require EUV at all.
Infineon produces automotive power semiconductors using mature processes such as 28nm and 40nm. NXP produces automotive-grade MCUs using mature processes as well.
STMicroelectronics manufactures sensors and power devices, using older-generation immersion DUV lithography machines, each costing tens of millions of dollars. Due to their low purchase volume, these machines contribute almost nothing to ASML’s global revenue and are nowhere near the scale of TSMC or Samsung, which deploy dozens of EUV machines.
What truly requires EUV is advanced logic chips below 7nm—Apple’s A-series, NVIDIA’s H100/B100 GPUs, AMD’s EPYC server CPUs, and Qualcomm’s flagship smartphone SoCs. In these areas, Europe has no domestic chip companies.
In Europe, chip design largely remains focused on "slow-paced" sectors such as automotive, industrial, and power devices. These markets offer stable profits and strong cash flow but do not require the most advanced manufacturing processes.
A chip design company without advanced process nodes doesn't need advanced process foundry services; without advanced process foundry services, there's no need for EUV.
This is the fundamental reason why ASML generates zero revenue in Europe: there are no facilities in Europe capable of utilizing EUV technology.
The EU allocated €43 billion, but the Court of Auditors itself says it’s unachievable.
Europe is not unaware of this issue.
In 2023, the European Union passed the Chips Act, allocating €43 billion with the goal of achieving 20% of global chip production capacity by 2030.
However, in April 2025, the European Court of Auditors issued a special report directly stating that the goal of the Chips Act was "very unlikely," and European Parliament members, in written questions, said the internal audit assessment suggested the actual share might be only 11%.
Where did the money go? The audit office reviewed the accounts and found that most of the €43 billion in subsidies went to "pilot lines" and "competence centers"—in other words, laboratories, research and development centers, and pilot production lines. Of the 13 potential wafer fab projects, only four received approval.
That’s the problem: Europe has poured money into R&D, hoping the market will naturally grow factories. But chip manufacturing is a scale business—building an advanced wafer fab requires at least $20 billion in investment, and without hundreds of billions in orders to sustain it, unit costs can never compete with TSMC.
IM in Belgium EC One of the world's leading microelectronics research centers, with R&D capabilities comparable to Bell Labs and IBM Watson. From IM EC Between the cleanroom and a mass-production wafer factory lies a bridge that policy, capital, and customers cannot cross.
Europe once attempted to build advanced factories. In 2023, Intel announced the construction of an advanced wafer fab in Magdeburg, Germany, with nearly €10 billion in subsidies from the European Union and the German government.
In 2024, Intel ran the numbers and found that costs had overrun, electricity prices were too high, and construction was too slow—so it scaled the project down to a smaller packaging and testing facility. The €10 billion in subsidies nearly went to waste.
Another example is Intel’s factory in Ireland. The Leixlip industrial park was originally slated for Intel 4 manufacturing, but due to Intel’s struggling foundry business and insufficient orders for advanced nodes, the project has been repeatedly delayed.
There aren’t enough local chip design companies in Europe to provide sufficient orders for these factories. TSMC’s factory in Arizona is backed by a lineup of customers including Apple, NVIDIA, and AMD.
Intel is building a factory in Germany, but the annual chip demand from German companies like Volkswagen and Siemens is not enough to fully occupy an advanced wafer fab.
Chip manufacturing is a chicken-and-egg problem: without design companies, there are no factory orders; without factories, design companies can't be attracted.
Europe lacks NVIDIA, Apple, and a large cluster of consumer electronics brands to absorb advanced process technologies. Demand for AI chips is entirely concentrated in the U.S., China, South Korea, and Taiwan—Europe can only watch.
There are no customers at the doorstep in Europe.
So ASML’s “warm invitation” translates to: the U.S. ties R&D through its chip bill, China and India tie factories through subsidies and land, while Europe comes up empty-handed.
ASML is headquartered in the Netherlands, with its R&D center in Eindhoven. But it knows well: over the next five years, the company’s fate will not be determined by how much subsidy Europe provides, but by whether TSMC, Samsung, and Intel expand production, and whether Europe can cultivate its own domestic industry.
Dassen’s comment on the call, “There is no sales in Europe,” highlights a concern that ASML itself is reluctant to disclose: ASML is currently the only player globally, but what happens when it is no longer the only one?
Lithography technology is not permanently monopolized; the history of technology shows that no industrial equipment can maintain a monopoly for more than fifty years.
Suppose one day in the future, other countries develop a lithography technology that can replace EUV, and all their major clients switch to the new equipment; ASML would need a fallback strategy—retreating to the European domestic market.
But what is the situation in Europe right now? Zero demand. Not a single advanced wafer fab, not even one foundry capable of running EUV.
This is ASML’s real concern: although it is currently thriving, its entire market exists on someone else’s land—ASML has not cultivated a single customer in its own backyard capable of purchasing its most advanced machines.
But its factory is still in Veldhoven, the Netherlands, and its thousands of engineers still work in Europe—this is the reason ASML says “zero sales in Europe” to increase Europe’s share of global chip production to 20%.
Europe is preparing for the worst-case scenario for the future: right now, global customers are rushing to place orders, and ASML is the only one. By the time it’s no longer the only one, it won’t even have a domestic base left.
This is another kind of overcapacity—Europeans spent decades developing the world’s most advanced lithography machines, yet not a single buyer can be found for them at home.
This is not a problem that can be solved by throwing money at it. It requires an entire ecosystem: chip design companies, foundries, consumer electronics brands, AI computing demand, and a venture capital system—none of which can be grown by a €43 billion subsidy bill alone.
ASML's CFO's statement, "We don't have system sales in Europe," is effectively the elegy for Europe's entire semiconductor strategy: you're the one selling shovels, but you don't own your own gold mine.
This article is from the WeChat public account "Hot Spot Commentary" (ID: redianweiping), authored by Wang Xinxǐ.