ASML Is Not Semiconductor Sovereignty: What Europe Still Lacks in the Chip Supply Chain

Editorial image generated by the author.

Europe occupies an unusual position in the semiconductor industry. ASML controls a technology without which the world’s most advanced chip factories would struggle to advance, while its long-standing partner ZEISS supplies the extraordinary optical systems inside its lithography machines. ASML describes ZEISS optics as a feature of all its lithography systems since the late 1980s. Europe also retains major semiconductor companies of its own, particularly in automotive, industrial and power electronics. Yet losing access to Taiwanese fabrication, American chip design, Korean memory or important parts of the Asian packaging ecosystem would still leave much of European computing exposed.

The paradox exposes the weakness in treating European semiconductor sovereignty as synonymous with domestic fabrication. Europe cannot sensibly recreate the entire semiconductor industry within its borders, and trying to do so would consume enormous resources duplicating capabilities that already exist elsewhere. A more useful objective is strategic interdependence: enough domestic fallback capacity to survive serious disruption, enough diversity among foreign suppliers that one country cannot shut Europe out, and enough indispensable European technology that dependency runs in both directions. ASML is the strongest existing example of that last category, but also a reminder that owning one extraordinary chokepoint is not the same as owning the semiconductor system.

The semiconductor supply chain is a system of chokepoints

The phrase “semiconductor industry” suggests something more coherent than the industry actually is. A modern chip passes through a long chain of specialised activities before reaching a computer, car or industrial machine. Someone designs its architecture. Electronic Design Automation software turns that design into manufacturable circuits. Intellectual property is licensed. Ultra-pure chemicals and silicon wafers are produced. Lithography, etching and deposition equipment create structures measured in nanometres. The chip is fabricated, cut, packaged, tested and finally incorporated into a larger system.

Each stage has developed its own specialists, sometimes dominating extraordinarily narrow technological niches. The United States leads much of advanced chip design, design software and semiconductor intellectual property. Taiwan has an exceptional position in advanced contract manufacturing. South Korea is particularly strong in memory. Japan retains important positions in materials and manufacturing equipment. Europe is unusually strong in lithography, optics, automotive and power semiconductors, and semiconductor research. The geography is uneven by design: decades of specialisation have produced a network of mutually dependent oligopolies rather than a collection of self-contained national industries.

Even countries with large semiconductor industries remain dependent on technologies produced elsewhere. Taiwan’s leading fabs need European lithography equipment and American design tools; European chipmakers use American software and intellectual property and Asian manufacturing; American chip designers depend heavily on Asian fabrication. Manufacturing share therefore tells us something about semiconductor power, but not enough. A region can manufacture large volumes of chips while remaining vulnerable to the loss of one foreign equipment supplier. Another can fabricate relatively few chips yet control technology without which everybody else’s factories struggle to advance. ASML demonstrates the difference.

Europe is strong upstream and weak where computing value is concentrating

European semiconductor policy should begin with the assets that already exist rather than with the assumption that Europe somehow lost the entire industry. ASML may be one of the most strategically consequential industrial companies in the world. Its EUV lithography systems combine light sources, mirrors, vacuum systems, software and precision mechanics at a level of complexity that is extremely difficult to reproduce. Infineon, STMicroelectronics and NXP occupy important positions in automotive, industrial and embedded semiconductors, while research institutions such as imec remain influential in developing future process technologies. These are not remnants of an obsolete European industrial base; they are serious positions in the modern semiconductor economy.

The strengths are nevertheless lopsided. The Commission’s 2026 semiconductor value-chain assessment estimates that the EU accounts for less than one per cent of global fabless semiconductor revenues and roughly four per cent of the worldwide chip-design workforce. The essential EDA tools and much semiconductor intellectual property remain predominantly American. The same assessment shows a much stronger European position in capital equipment, substrates and materials than in foundries or outsourced assembly and testing.

That gives Europe a peculiar industrial profile: powerful in selected upstream technologies but thin across several of the layers where AI and advanced computing are generating enormous value. The useful policy question is therefore not simply how Europe can manufacture more chips. It is which missing capabilities create dangerous dependencies and which merely reflect efficient international specialisation.

A fab is not an ecosystem

The TSMC-led European Semiconductor Manufacturing Company facility under construction in Dresden illustrates the difference. The plant is strategically useful. According to ESMC’s published plans, it will initially manufacture using 28/22 nanometre planar CMOS and 16/12 nanometre FinFET processes, technologies suited to the automotive, industrial, IoT and telecommunications markets in which Europe already has substantial demand.

These are not the process nodes used for the newest AI accelerators, but dismissing them as “old” technology would miss one of the lessons of the semiconductor shortages earlier this decade. A car factory gains little from having access to its most sophisticated processors if production stops because a far cheaper controller is unavailable. Resilience depends on the chips an industrial system actually consumes, not only on whatever currently represents the technological frontier. Dresden fills a genuine European need, but it does not create a European equivalent of Taiwan’s leading-edge foundry ecosystem.

Intel’s abandoned Magdeburg project shows why closing that gap is harder than constructing a factory. Intel planned enormous advanced-fabrication investments in Germany and Poland, supported by substantial public incentives, before deciding not to proceed. The company’s own explanation for its wider foundry retrenchment was revealing: Intel said it had invested “too much, too soon” without adequate demand and had created a fragmented and underused manufacturing footprint. Intel later cancelled the German and Polish projects.

The retrenchment was not a withdrawal from European advanced manufacturing. In July 2026, Intel announced a €5 billion expansion of its existing Leixlip campus in Ireland, using existing cleanroom space to expand Intel 3 production and research and development. The contrast reinforces the ecosystem problem: expanding an established manufacturing hub is different from creating a new leading-edge cluster from scratch.

Magdeburg does not prove that weak European demand killed the project; Intel had strategic and financial problems of its own. It does illustrate a constraint that subsidies cannot remove indefinitely. Leading-edge fabrication needs engineers, equipment suppliers, energy infrastructure, packaging capacity and customers capable of absorbing huge production volumes. Each technological generation then demands further billions in investment. A frontier fab without a dense economic ecosystem around it risks becoming an extraordinarily expensive monument to industrial policy.

Europe’s weakness in advanced fabrication is tied to its weakness in advanced chip design and, increasingly, to the geography of AI and cloud computing. The companies designing and consuming the largest volumes of frontier processors help create the demand that makes advanced foundries economically sustainable. This is closely related to Europe’s dependence on rented AI capacity: importing a capability is not inherently a strategic failure, but relying on providers that cannot be replaced or rapidly supplemented can become one. Industrial policy can help create an ecosystem. It cannot permanently substitute for one.

Manufacturing market share is a poor measure of resilience

The original European Chips Act emerged from a genuine problem. In 2022 the Commission set an ambition for Europe to reach at least 20 per cent of world production, by value, of cutting-edge and sustainable semiconductors by 2030. The policy was more sophisticated than the shorthand claim that “Europe must produce 20 per cent of the world’s chips”: the original Chips Act communication also spoke explicitly about balanced interdependencies and supply chains without single points of failure.

Manufacturing capacity plainly matters. The difficulty comes when production share becomes a proxy for security. Europe could dramatically expand mature-node manufacturing while remaining dependent on a handful of non-European suppliers for advanced design software, specialised materials, packaging technologies or particular pieces of fabrication equipment. Semiconductor output would rise while an unchanged single point of failure remained capable of disabling the system. Conversely, a relatively modest investment in an obscure material or technology might remove a dependency that could otherwise stop an entire industry.

Semiconductors make the problem unusually visible because the strategic importance of a component bears little relationship to the size of its market. A specialised chemical may generate little revenue compared with a processor company. An optical component may come from a supplier few consumers have heard of. A packaging technology may account for only a small part of the finished chip’s cost. Remove the wrong one, however, and billions of euros of manufacturing capacity become irrelevant.

A serious resilience policy therefore needs to ask what happens when a capability disappears, not merely how much of everything Europe produces.

Strategic interdependence needs an escape route

ASML points towards a more realistic form of technological sovereignty. Europe depends heavily on semiconductor manufacturing elsewhere, but those manufacturing centres also depend on European lithography. That creates reciprocal leverage: deliberate exclusion of Europe from one critical technology carries the risk that Europe can restrict something difficult to replace in return.

It would be a mistake, however, to turn ASML into a comforting myth of European independence. Reciprocal dependence does not imply reciprocal vulnerability on the same timescale. If access to large volumes of advanced Taiwanese chips disappeared, European users could feel the shortage quickly. If new ASML systems stopped reaching Taiwan, existing fabs would not suddenly cease operating; the pressure would emerge through maintenance, capacity expansion and successive technology generations.

Nor is ASML itself an entirely self-contained European capability. The company says that approximately 80 per cent of its bill of materials comes from a global network of suppliers. ASML’s own description of its supplier network makes the point neatly: even one of Europe’s greatest technological chokepoints is itself the product of international specialisation.

Strategic interdependence therefore needs an escape route. For technologies whose disappearance could rapidly disable critical European systems, maintaining some domestic production may be worth paying for even when it is commercially inefficient. Where domestic duplication would be prohibitively expensive, Europe can reduce risk by sourcing from several politically independent countries. Beyond those minimum safeguards, European industrial policy should cultivate technologies in which European firms are themselves difficult to replace.

The resulting strategy is neither autarky nor simple reliance on allies. It combines domestic fallback where failure would be catastrophic, diversification where substitution is realistic, and European technological positions where specialisation can create reciprocal dependence. That also provides a much better test for industrial subsidies. A project should not qualify merely because it belongs to a “strategic industry”. The relevant question is whether that particular investment materially changes Europe’s vulnerability.

Chips Act 2.0 moves towards the right question

The Commission’s Chips Act 2.0 proposal, published on 3 June 2026, moves noticeably in this direction. It explicitly identifies continued European dependence in advanced semiconductor manufacturing and design, gives greater attention to the broader value chain and supply-chain vulnerabilities, and speaks of maintaining Europe as an “indispensable” participant while reducing strategic dependencies. That diagnosis is closer to the real problem than a policy centred principally on adding fab capacity.

It is still a proposal rather than settled policy. Whether the change in diagnosis produces a corresponding change in how public money is allocated will matter more than the wording itself. Governments have powerful incentives to favour visible investments: large factories create construction jobs, ribbon-cutting ceremonies and impressive announcements about billions committed. A small materials plant, packaging facility or specialist equipment supplier may have far greater resilience value while producing much less political theatre.

European semiconductor policy should therefore begin with dependency mapping and work backwards towards investment. Which technologies could stop a critical industry if access disappeared? How concentrated is supply? Could Europe switch to producers in another allied country? How long would substitution take? Can the technology be stockpiled, or would reserve production capacity be more useful? Does Europe already possess neighbouring capabilities from which an alternative could realistically be developed? Where could European research or industry create the next position that other regions would struggle to replace?

The answers will sometimes justify a very expensive fab. In other cases they may justify advanced packaging, design infrastructure, specialised materials, strategic inventories, reserve manufacturing capacity or research funding. Some foreign dependencies will turn out not to require intervention at all. A component produced competitively in several friendly countries is foreign, but it is not necessarily a strategic vulnerability.

The same distinction extends beyond semiconductors. Europe is having similar arguments about artificial intelligence, cloud infrastructure, energy technology, defence, critical raw materials and strategic autonomy in space. Trying to reproduce every layer of every strategic industry inside Europe would consume capital and skilled labour on a scale that would itself weaken the continent. Treating all foreign dependence as dangerous is therefore almost as unhelpful as ignoring dependency altogether.

The semiconductor industry suggests a more discriminating standard. Europe should preserve the capabilities in which it is already exceptionally strong, close the relatively small number of gaps whose failure could disable critical systems, diversify external supply where duplication makes little sense, and keep investing in technologies that make Europe difficult to exclude from the systems on which it depends.

Europe does not need to manufacture every chip. It needs to make sure that no single outside power can decide whether Europe keeps computing.