On February 12, 2026, Taiwan's Ministry of Economic Affairs added 18 items across three categories — advanced 3D-printing equipment, advanced semiconductor equipment, and quantum computers — to its Strategic High-Tech Commodities (SHTC) export-control list. Exporters of items such as CMOS fabrication tools, cryogenic cooling systems, scanning electron microscopes, and low-temperature wafer-probing equipment must now obtain a license from the International Trade Administration before shipping abroad. It is the latest move in a regime that has, over the past year, blacklisted Huawei and SMIC and rewritten the rulebook governing where TSMC may build its most advanced chips.
Two very different instruments are at work here, and conflating them has muddied the debate. One is a targeted, internationally coordinated licensing system. The other is a blanket cap on technology generations that TSMC may deploy outside Taiwan. The first is proportionate regulation done well. The second is industrial leverage wearing the costume of national security — and Taipei's own behavior over the past three months proves it.
The 18-item list is the right kind of control
Start with the strongest case for Taipei. Taiwan sits at the center of the world's most strategically sensitive supply chain. It has an obvious interest in ensuring its tooling does not flow to entities that would use it for weapons proliferation or military end-use, and in keeping its rules legible to the partners it trades with. The Ministry was explicit that the February revision was made to keep Taiwan's controlled items consistent with the Wassenaar Arrangement and the Nuclear Suppliers Group — the same multilateral frameworks the US, EU, and Japan use (Taiwan News).
This is export control as it should be done. It is item-specific, tied to identifiable end-use risk, and administered through a permit process rather than a prohibition. As the Ministry stressed, regulation is not a ban: licenses are issued once the International Trade Administration confirms the goods will not be used to proliferate weapons of mass destruction (Taipei Times). The underlying SHTC system, run by the ITA, already screens designated entities and military end-users (Taiwan ITA). Adding Huawei and SMIC to that entity list in June 2025 was a narrow, defensible decision aimed at named counterparties, not a sweeping market closure (Focus Taiwan).
The node-generation cap is a different animal
The second instrument is harder to justify. In April 2025 Taiwan amended Article 22 of its Industrial Innovation Statute, creating a prior-approval mechanism for outbound investment that the central authority may refuse — in whole or in part — where it judges national security or economic development to be at stake (Industrial Innovation Statute, Art. 22). Layered on top is the informal "N-1" rule: TSMC may run only one process generation behind its domestic leading edge at foreign fabs. With TSMC now moving its 2nm-class N2 node into high-volume production in Taiwan, its Arizona fabs are confined to older 5nm-class N4/N5 work (heise online).
The steelman here is the "silicon shield": keeping the frontier node exclusively on the island is said to raise the cost of any blockade or invasion and to preserve Taiwan's indispensability to its allies. It is a serious argument, and we should not pretend otherwise. But it does not survive contact with the policy's design or its trajectory.
A generation-cap is a poor security instrument because it controls the wrong variable. It does not ask whether a given customer poses an end-use risk — the question the SHTC list correctly asks. It instead freezes allied capacity at an arbitrary distance behind the frontier, slowing the very supply-chain diversification that reduces the world's single-point dependence on the Taiwan Strait. Officials have floated tightening it further to an "N-2" rule, which a deputy minister at the National Science and Technology Council confirmed could limit overseas fabs to two generations behind — potentially a two-to-four-year lag (Tom's Hardware). That would push Arizona's Phase 2, slated for 3nm-class output by 2027, back out of compliance the moment it opened.
The tell is that Taipei is already relaxing it
The clearest evidence that the cap is leverage rather than law is what happened after TSMC committed $100 billion to new US facilities: reporting indicates Taiwan moved to drop the restriction rather than enforce it (Yahoo Finance). A genuine security firewall does not dissolve in response to a capital-expenditure announcement. A bargaining chip does.
The lesson is not that Taiwan should abandon export control. It is that the two instruments should be kept distinct. The SHTC list — end-use-tested, multilaterally aligned, entity-specific — is the model: it constrains genuine proliferation risk while leaving legitimate allied commerce free to flow. The node-generation cap should be retired in favor of that approach. Taiwan's deterrent rests on being the most capable, most trusted node in a resilient allied network — not on hoarding a lead that diffuses anyway. Proportionate controls protect that position. Blanket ones quietly erode it.