In 2023, the Netherlands unilaterally restricted exports of the world’s most advanced chipmaking equipment, months before Brussels had settled on any common approach. Within two years, the European Commission had diagnosed an EU-wide “patchwork” of similar national rules and moved to centralise enforcement of dual-use export controls — even as France and Germany, the bloc’s two largest economies, reportedly pushed back against exactly that move. This piece traces the sequence across four member states — the Netherlands, Germany, France and Lithuania — and asks a simple question: who actually pays the political price of acting first, and who ends up keeping control of the outcome? The answer complicates the usual story about “big states dominating small ones” in EU policymaking, and carries a direct lesson for anyone watching how the EU is likely to handle the next chokepoint technology — AI chips included.

In June 2023, the Dutch government quietly introduced a national licensing requirement on the export of advanced semiconductor manufacturing equipment.^[1] It was a national measure, adopted outside the EU’s ordinary legislative process, aimed — without saying so explicitly — at slowing China’s access to the machines that make the world’s most advanced chips. Two years later, that single national decision had triggered a European Commission White Paper diagnosing a bloc-wide “patchwork” of similar national rules,^[2] a voluntary coordination mechanism, and a delegated regulation rewriting the EU’s entire common export control list.^[3] No member state asked for that sequence of events at the outset. And at least two of the Union’s most powerful economies, France and Germany, are reported to have told the Commission bluntly that it “has no role to play” in this area of policy.^[4]

That is a genuine puzzle worth unpacking. How did one country’s national administrative decision end up reshaping the EU’s collective approach to a matter — economic and technological security — that formally remains a member-state, not a Brussels, competence? And once you look closely at how four different EU countries actually behaved, a more interesting and less predictable story emerges than the usual “Brussels versus the member states” or “big states push around small ones” narratives that tend to dominate commentary on EU economic security policy.

Semiconductor manufacturing equipment is a useful case precisely because it is one of the clearest examples anywhere in the global economy of a genuine chokepoint technology: a handful of companies, concentrated in one or two countries, make a product that the rest of the world’s chip industry simply cannot substitute. That concentration gives the states hosting those firms an unusual amount of structural leverage — and, as this case shows, an unusual amount of structural exposure to pressure from outside the EU as well. Understanding who actually carries that exposure, and who gets to decide what happens with it, matters well beyond this one dossier.

Extreme-ultraviolet, or EUV, lithography is the technology at the center of all of this: a process that uses light with an extremely short wavelength to etch circuit patterns onto silicon wafers small enough to pack billions of transistors onto a single chip. Building a machine capable of doing this reliably at industrial scale turns out to be one of the hardest manufacturing problems in the world — decades in development, involving components so precise that a handful of companies anywhere on Earth can supply them. That is what makes this a genuine chokepoint rather than merely an important industry: there is, at present, no realistic alternative supply chain a country like China could stand up quickly, which is exactly why access to this equipment has become such a live flashpoint in the broader US-China technology competition, and why the EU has found itself pulled into a dispute that is, at its root, about a rivalry between two other powers.

1. A National Rule with EU-Wide Consequences

The Dutch story starts with a letter, not a law. In March 2023, the Dutch Minister for Foreign Trade and Development Cooperation told parliament, plainly, that the government intended to introduce new export controls.^[5] The letter gave an unusually candid reason for going it alone rather than waiting for a multilateral deal: the Wassenaar Arrangement, the main international forum for coordinating this kind of export control, works by consensus — and Russia, as a member, could block any new restriction indefinitely. Rather than wait for a veto that would likely never lift, the Netherlands decided to act nationally, relying on a clause in the EU’s own dual-use regulation (Article 9 of Regulation (EU) 2021/821) that lets any member state impose additional controls on public-security grounds, on top of whatever the EU-wide list already covers.

The resulting rule, in force from 1 September 2023, required a license for exporting specific deep- and extreme-ultraviolet lithography equipment^[6] — the machines made almost exclusively by the Dutch company ASML, the sole global supplier capable of producing the tools needed to manufacture the most advanced computer chips. The rule was expanded twice: once in September 2024, adding further lithography equipment and introducing a general license (NL900) that pointedly excluded China from its list of eligible destinations;^[7] and again in January 2025, extending controls to testing and inspection equipment. That second expansion is where the Dutch government’s own language becomes genuinely revealing: its explanatory text grounds the measure explicitly in “the exceptional position of the Netherlands in the international semiconductor value chain”^[8] — an unusually direct admission, in a legal text, that this was about the country’s singular commercial position, not a generic, symmetrical non-proliferation concern applicable to any state.

By early 2025, then, the Netherlands had gone from a cautious, multilaterally-minded actor to a country running its own escalating national export-control regime — alone, and explicitly aware of just how alone it was. It is worth pausing on how unusual that sequence is. Most EU member states prefer to move in step with Brussels precisely to avoid the diplomatic exposure of being singled out by a major trading partner. The Netherlands took that exposure on directly, three times in under two years, before there was any EU-wide framework in place to share the cost.

2. Four Countries, Four Very Different Playbooks

The Dutch case looks less like a template and more like an outlier once compared with what three other member states actually did. All four sit somewhere in the same semiconductor supply chain or the same broader China-relations story, but their choices diverge sharply — in whether they legislated at all, in how loudly they justified what they did, and in how they positioned themselves on the question of EU-level coordination.

Germany is the most surprising case. Two German firms, Carl Zeiss SMT and TRUMPF, are every bit as indispensable to EUV lithography as ASML itself: Zeiss makes every mirror optic and TRUMPF every drive laser in every EUV machine sold anywhere in the world.^[9] Despite holding a chokepoint position fully comparable to the Netherlands’, Germany has to date passed no equivalent national law. Instead, German authorities quietly tightened licence screening from 2023 onward — with real effects (reported delivery delays for firms like SÜSS MicroTec)^[10] but no public justification, no named threat, and nothing for outside observers to point to. In 2026, when US regulators opened an enforcement inquiry into whether EUV components had reached China,^[11] the response came not from the German government but from Zeiss itself, whose semiconductor chief publicly denied the company exports EUV technology to China at all, while noting the firm was independently reducing its dependence on the Chinese market^[12] — a corporate denial standing in for the government statement that never came. It’s a genuinely odd asymmetry: two countries hosting equally essential technology, one of which has spent two years building an increasingly public legal architecture around it, and the other of which has said, officially, almost nothing at all.

France took a middle path. Its own national control list, adopted by ministerial order on 2 February 2024,^[13] covers cryogenic chip circuits, advanced etching equipment, and quantum-computing components. But the French text itself is almost entirely procedural — no mention of weapons risk, no named threat, nothing like the Dutch government’s blunt language about its own exceptional position. Commentary on the measure links its timing and scope directly to the European Commission’s own Economic Security Strategy, published a few weeks earlier,^[14]which had already flagged semiconductors and quantum technology as areas needing protection. France, in other words, let Brussels do the framing and simply implemented it quietly at home — getting the substance of a restriction without the domestic political cost of a public justification.

Lithuania has no chip industry to protect and virtually no trade exposure to China worth mentioning — roughly 1% of exports even before 2021.^[15] Yet Lithuania is arguably the most consequential of the four states in shaping the EU’s broader posture on this whole area of policy. After Vilnius allowed a “Taiwanese Representative Office” to open in 2021, Beijing responded with a sustained campaign of retaliation: Lithuania vanished from Chinese customs systems, and China quietly pressured third-country manufacturers against using Lithuanian-made components at all.^[16] Lithuania’s response was not a chip-specific national law, since it has no relevant industry to legislate for, but sustained diplomatic and legal advocacy for a general EU tool against economic coercion, which fed directly into the EU’s Anti-Coercion Instrument, in force since December 2023.^[17] Separately, and for an unrelated reason, Lithuania has also used the exact same legal provision the Netherlands used — but to restrict exports relevant to Russia’s war against Ukraine, not to China^[18] — a useful reminder that the same legal tool can be doing completely different political work in different countries, and that a state’s absence from one story doesn’t mean it’s absent from the underlying legal machinery altogether.

What the table doesn’t fully capture is the sequencing. The Netherlands moved first and alone, without any EU framework to lean on. France moved second, after Brussels had already supplied a ready-made justification. Germany has, so far, chosen not to move in any public, legally visible way at all. And Lithuania’s contribution long predates the chip dossier entirely — it comes from a completely different fight, which nonetheless supplied language Brussels would later reuse.

3. Why Brussels Stepped In — and Why Some Pushed Back

A quick word on the multilateral backdrop, because it explains why this whole saga happened at the national level in the first place rather than through the usual international channels. The Wassenaar Arrangement, founded in 1996, is where 42 countries — including all EU member states, the US, and, notably, Russia — are supposed to agree collectively on which dual-use and military-relevant goods need export controls. It works by consensus, which was a reasonable design when it was created but became a serious liability the moment one of its members had an interest in blocking exactly the kind of restriction the rest of the group wanted. Russia’s continued membership, even after 2022, meant that any attempt to add lithography equipment to the internationally agreed control list could simply be vetoed. That gridlock is the direct reason the Netherlands, and later the EU as a whole, ended up building a parallel, EU-specific route around a forum that was supposed to make this kind of unilateral action unnecessary.

By January 2024, the Commission’s own White Paper on Export Controls had spotted the pattern: more than half of member states had layered national controls on top of the EU’s common list, risking exactly the kind of “forum shopping” where an item blocked in one country simply gets exported through another that hasn’t yet restricted it.^[19] The White Paper’s own language is worth quoting directly, because it echoes the vocabulary the Netherlands and Lithuania had already been using in their own, separate contexts: the lack of a common EU voice, it warned, “exposes individual Member States to strong geopolitical pressures”.^[20] That is not a coincidence of phrasing — it is close to a direct restatement of the case both countries had already been making for their own, very different reasons.

Two concrete steps followed. In April 2025, the Commission issued a Recommendation encouraging — but not requiring — member states to coordinate their national control lists before adopting them.^[21] In September 2025, it went further, using its existing delegated powers to rewrite the EU’s entire Annex I control list in one go,^[22] explicitly in order to add items that Russia had been blocking at the multilateral Wassenaar level for years. Both steps matter — but so does the choice of legal vehicle. A Recommendation is voluntary and carries no legal obligation. A delegated act uses powers the Commission already had under the existing regulation, rather than requiring new legislation subject to a full vote by the Council and the European Parliament. Both are, relatively speaking, the least confrontational tools available to the Commission — a striking choice for a Commission that had just spent a year documenting, in its own White Paper, how serious a problem the national patchwork actually was.

That choice looks less like an accident once you know that France and Germany had, shortly beforehand, reportedly told the Commission directly that it “has no role to play” in export control policy on arms and dual-use goods.^[23] Two of the EU’s most powerful, most exposed member states pushed back hard against binding centralization — and got a Recommendation and a technical Annex update instead of a new law that might have bound them to a common standard on the same terms as everyone else.

4. Who Actually Benefits from “Centralization”?

Put the pieces together and a pattern emerges that doesn’t match either of the two stories people usually tell about this kind of EU policymaking.

It is not simply “Brussels grabbing power from member states” — the Commission’s actual tools here are notably soft, and were chosen that way deliberately. And it is not simply “big states imposing rules on small ones” either — if anything, something closer to the opposite happened. The Netherlands, a mid-sized economy, was the country that had to act first, alone, and take the diplomatic heat directly from Washington. Lithuania, a small state with zero material stake in semiconductors, supplied much of the vocabulary — “exposure,” “vulnerability to coercion” — that Brussels later adopted almost word for word to justify its own response. Germany and France, the two heaviest-weight economies in the comparison, ended up with an outcome that preserves rather than restricts their own freedom to act (or not act) as they see fit — and Germany achieved this without ever passing a public law on the subject at all.

There is a useful way to think about why this particular pattern is not a coincidence. Imagine a simple situation with two players who both host essential technology: someone has to publicly restrict trade in a hugely valuable, politically sensitive product, or the risk is that the United States imposes far harsher, unilateral restrictions on everyone’s firms regardless. Whoever moves first pays a real cost — diplomatic friction, a public confrontation with a major trading partner, the domestic political fallout of constraining a flagship national industry. Everyone else benefits once someone has moved, without paying that cost themselves. Economists sometimes call this a volunteer’s dilemma, and it predicts more or less exactly what happened here: the player facing the most acute, most immediate exposure — the Netherlands, given how directly and specifically Washington was leaning on it and on ASML by name — ends up volunteering, while the player with more diffuse, more easily deniable exposure — Germany, spread across two firms rather than concentrated in one globally famous household name — can afford to wait and see.

One qualification matters here, because it’s tempting to overstate the “Germany got away with it” reading of events. Germany’s lack of a public law did not mean zero cost — the quiet administrative tightening documented for firms like SÜSS MicroTec shows real material friction and was still imposed on German exporters throughout this period. What Germany avoided was not cost altogether, but the political and diplomatic cost of a public justification — the kind of cost the Dutch government took on explicitly, and that its own explanatory texts show it was fully, deliberately aware of taking on from the very first letter to parliament in 2023.

There is a broader lesson here for how to read EU integration stories generally. It’s tempting, whenever the EU centralizes a new area of policy, to reach for one of two ready-made explanations: either “the technocrats in Brussels are quietly expanding their turf,” or “the big member states are once again writing the rules for everyone else.” This case suggests a third possibility that gets less attention than it deserves: sometimes centralization happens because one exposed, mid-sized state runs out of good options and moves first, and the resulting EU framework ends up reflecting whoever spoke loudest early on — while the states with the deepest pockets and the most to lose retain enough leverage to keep the actual rulebook loose. Voice and power, in other words, don’t automatically travel together, and assuming they do will lead you to misread who actually shaped a given piece of EU policy.

A Live Case, and a Lesson for AI Chips

This is still an unfolding story, not a closed one. The April 2025 Recommendation is barely a year old at the time of writing, and it remains a genuinely open question whether the Commission will, in practice, use it to nudge member states toward something functionally closer to a binding rule — which would meaningfully change the balance described here. It’s worth revisiting in a year or two, once there’s enough implementation practice on the ground to judge whether “voluntary coordination” stays voluntary in substance, not just in name.

But the pattern already has an obvious application beyond lithography equipment. The same structural logic — a handful of firms concentrated in one or two member states controlling a genuine chokepoint technology, sustained US pressure falling hardest on whichever state is most exposed, and a Commission response built from whatever legal vehicle survives resistance from the largest member states — is very likely to repeat itself as the EU works out how to handle export controls on advanced AI chips, AI training hardware, and the software toolchains that go with them. Several EU member states already host firms with genuine chokepoint relevance to the AI hardware stack, even if none yet approaches ASML’s degree of global irreplaceability.

If the semiconductor equipment case is any guide, whichever member state hosts the next AI-relevant chokepoint firm should expect to be the one that ends up moving first, alone, and bearing the political cost of doing so — while the states with more diffuse, harder-to-pin-down exposure keep the most room to maneuver, and the states with no chokepoint industry at all end up supplying the language everyone else eventually borrows. That is a pattern worth keeping firmly in mind well before the next chokepoint dispute actually lands on Brussels’ desk.

These statements are attributable only to the author, and their publication here does not necessarily reflect the view of the other members of the AI-Regulation Chair or any partner organizations.

^[1] Regeling van de Minister voor Buitenlandse Handel en Ontwikkelingssamenwerking, Staatscourant 2023, nr. 18212, 30 June 2023; Letter of the Netherlands’ Minister for Foreign Trade and Development Cooperation to the Dutch Parliament, 8 March 2023 (AVT/BZ-230307-008).

^[2] European Commission, White Paper on Export Controls, COM(2024) 25 final, 24 January 2024.

^[3] Commission Recommendation (EU) 2025/683 of 8 April 2025 on coordination of national control lists, OJ L, 2025/683, 16.4.2025; Commission Delegated Regulation (EU) 2025/2003 of 8 September 2025 amending Regulation (EU) 2021/821, OJ L, 2025/2003, 14.11.2025.

^[4] Pugnet, A. (2025) ‘Germany and France to EU: Stay Out of Arms Export Policy’, Euractiv, 25 July 2025.

^[5] 5. Letter of the Netherlands’ Minister for Foreign Trade and Development Cooperation to the Dutch Parliament, 8 March 2023, above.

^[6] Regeling geavanceerde productieapparatuur voor halfgeleiders, Staatscourant 2023, nr. 18212, above.

^[7] Staatscourant 2024, nr. 29008, 6 September 2024.

^[8] Staatscourant 2025, nr. 1894, 13 January 2025.

^[9] Germann, J., Rolf, S., Baines, J. and Starrs, S.K. (2024) ‘A Chip War Made in Germany? US Techno-Dependencies, China Chokepoints, and the German Semiconductor Industry’, Politics and Governance, 12, Article 8265; see also Sovereign Magazine (2026) ‘European Semiconductor Companies the World Cannot…’, 15 May 2026, describing Zeiss/TRUMPF/ASML as an EUV ‘trinity’.

^[10] Reporting on SÜSS MicroTec licence delays, 2023–24 company filings.

^[11] 11. Tech Times (2026) ‘ASML EUV Components in China: US Cites Shipment Evidence as BIS Enforcement Looms’, 22 June 2026.

^[12] BigGo Finance (2026) ‘Germany’s Zeiss Denies EUV Technology Exports to China, Vows to Reduce Supply Chain Reliance’, 18 July 2026.

^[13] Arrêté du 2 février 2024, JORF n°0034 du 10 février 2024, texte n°8, NOR: ECOI2401482A.

^[14] Doussin, A., Estrabaud, P. and Allinson, T. (2024) ‘France Announces New Export Controls on Semiconductors and Quantum Equipment and Technology’, Hogan Lovells Engage, 15 February 2024; European Commission, Joint Communication on a European Economic Security Strategy, JOIN(2023) 20 final.

^[15] Andrijauskas, K., cited in Center for Strategic and International Studies analysis; Szyszczak, E. (2022) ‘Trade and Security: The EU’s Unilateral Approach to Economic Statecraft’, Briefing Paper 70, UK Trade Policy Observatory, October 2022.

^[16] Reynolds, M. and Goodman, M.P. (2022) ‘China’s Economic Coercion: Lessons from Lithuania’, Commentary, Center for Strategic and International Studies, 6 May 2022.

^[17] Regulation (EU) 2023/2675 of 22 November 2023 on the protection of the Union and its Member States from economic coercion by third countries, OJ L, 2023/2675, in force 27 December 2023.

^[18] Lithuania, Resolution No. 512 of 28 June 2023 ‘On the Application of National Control Measures Pursuant to Article 9 of Regulation (EU) 2021/821’, as amended by Resolution No. 888 of 15 November 2023.

^[19] European Commission, White Paper on Export Controls, COM(2024) 25 final, above.

^[20] Ibid.

^[21] Commission Recommendation (EU) 2025/683, above.

^[22] Commission Delegated Regulation (EU) 2025/2003, above.

^[23] Pugnet, A. (2025), above.