Switzerland is not typically associated with chip wars. Its semiconductor footprint runs more toward precision metrology, packaging, and niche ASIC design than high-volume fabrication. But on April 2, 2025, the Swiss Federal Council did something structurally significant: it amended the Goods Control Ordinance (GCO) to impose new export licensing requirements on advanced semiconductor technologies — not because the Wassenaar Arrangement directed it to, but precisely because the Arrangement could not. The changes took effect May 1, 2025.
What the New Controls Actually Cover
The amended Annex 2 of the GCO adds licensing requirements across four technology sectors: advanced semiconductor manufacturing, quantum computing, artificial intelligence applications, and additive manufacturing. Within semiconductors, the Export Control Numbers (ECNs) are specific. GAAFET (gate-all-around field-effect transistor) technology for advanced integrated circuits falls under ECN 3E901. Advanced electron microscopy equipment used in semiconductor inspection sits under 3B902. Cryogenic systems and low-temperature CMOS circuits are captured by the 3A901–3A904 series. Isotopically enriched materials for epitaxy production — the ultra-pure substrates on which chips are grown — fall under 3C901–3C903.
GAAFET is worth dwelling on. It is the transistor architecture that has displaced FinFET at the frontier nodes: Samsung deployed it in its 3nm process in 2022; TSMC followed. Placing GAAFET know-how under export licensing means Switzerland is controlling knowledge central to sub-3nm chip design. This reflects not just geopolitical caution but a precise reading of where the semiconductor supply chain's most dual-use-relevant knowledge actually resides — not in finished chips, but in process technology and architectural IP.
Why Switzerland Bypassed the Multilateral Channel
The Wassenaar Arrangement — the 42-member dual-use export control regime — requires unanimous consent to add any technology to its control list. That unanimity requirement worked tolerably in the 1990s. It has become a structural liability since 2022.
Following Russia's full-scale invasion of Ukraine, Russia has used its Wassenaar seat to block updates targeting technologies most relevant to military AI and precision-guided manufacturing. CSIS researchers Junusova and Reinsch, in a November 2024 analysis, documented the resulting institutional paralysis: adding a new technology under the consensus process "can take up to three years." With GAAFET architectures already at commercial scale and quantum computing advancing rapidly, that timeline is not a bureaucratic inconvenience — it is a policy failure.
Switzerland's answer was to amend domestic ordinances rather than wait. This aligns it with a coalition that has been assembling since 2024: France imposed national semiconductor and quantum controls in February 2024; the United Kingdom followed in March; the US Department of Commerce published an interim final rule targeting advanced semiconductors, quantum computing items, and additive manufacturing in September 2024. The European Commission then issued Recommendation 2025/683 in April 2025, formally coordinating national control lists across member states outside the blocked WA process. Switzerland, as a non-EU state, moved in the same direction on its own authority.
The Case for Multilateral Frameworks — and Why the Argument Has Run Out
Critics of the "Wassenaar Minus One" approach are not wrong in principle. The Arrangement's value always rested on universality: if all major exporters apply the same controls, leakage routes close. When leading Western economies control GAAFET know-how but others do not — even willing partners not yet aligned — controls are only as effective as the weakest link in the chain. There is also a legitimate concern that unilateral controls, freed from the discipline of multilateral negotiation, slide toward disguised protectionism.
These objections deserve serious weight. But they describe a world in which the multilateral mechanism is functional. That world does not currently exist. A consensus process with a structural veto held by a state that has demonstrated it will deploy that veto systemically is not a mechanism — it is a veto service. The question is not whether a functioning multilateral framework is preferable. It clearly is. The question is whether waiting for one is a tenable policy while frontier semiconductor architectures proliferate. For Switzerland in April 2025, the answer was no.
A Licensing Regime, Not a Prohibition
What makes Switzerland's implementation notable is its calibration. The GCO controls, it does not prohibit. Transfers require licenses; they are not automatically blocked. Critically, the regulations carve out basic scientific research below Technology Readiness Level 4 — the proof-of-concept threshold. ETH Zurich's internal guidance on the change stated plainly that research below TRL 4 "is generally not affected," but that physical prototypes and applied knowledge transfers trigger licensing requirements. This is a meaningful boundary: it preserves Switzerland's role as a global research hub while targeting precisely the commercializable knowledge that carries dual-use risk.
That design matters given Switzerland's simultaneous investment in domestic semiconductor capability. The SwissChips initiative — a CHF 48.6 million public-private collaboration led by ETH Zurich, EPFL, and CSEM, funded by the State Secretariat for Education, Research and Innovation — is building Swiss competence in chip design, metrology, space electronics, and biomedical systems through 2027. Tightening export controls while expanding domestic capability is not a contradiction; it is the industrial strategy that the US CHIPS Act, the EU Chips Act, and now Swiss policy have all converged on independently.
The Broader Signal
The Second Swiss Semiconductor Day, held in Neuchâtel in March 2026, made the tension explicit. Its closing panel addressed "reshoring, export-control pressures, energy constraints, and the global competition for talent and IP" as the defining challenges for Switzerland's chip sector. Compliance costs for licensing requirements land differently on a country whose semiconductor strengths are research-intensive niches — chip design, metrology equipment, packaging — rather than high-volume fabs with dedicated trade-compliance teams.
The broader implication is architectural. When a neutral, treaty-respecting, non-EU country like Switzerland determines that the formal multilateral export control channel is too slow and too compromised to be usable, it signals not a breakdown in Swiss foreign policy but a breakdown in the regime itself. The Wassenaar Minus One coalition is the best available substitute. It is not a replacement. Until a multilateral framework resistant to strategic vetoes is constructed, controls on advanced semiconductor technology will remain fragmented, imperfect — and worth maintaining regardless.