The US-Taiwan Agreement on Reciprocal Trade Signed on 12 February 2026 Commits Taiwan's Government to No New Quantum, AI or Biotech Research Arrangements With Authorities of Countries of Concern
In November 1949 the United States and six Western European governments agreed to set up the Coordinating Committee for Multilateral Export Controls, CoCom, which began operating in January 1950 from an annex of the American embassy in Paris. CoCom had no treaty. Its lists of restricted goods and technologies were confidential, its decisions were taken by consensus among officials, and no legislature ever ratified it. Until it was wound up in 1994, that informal body decided which machine tools, computers and scientific instruments could cross the line to the Soviet bloc. The United States already had its Export Control Act of February 1949; the multilateral layer of allied technology control began as an executive arrangement among officials, and the published lists and the Wassenaar Arrangement that succeeded CoCom in 1996 came later.
The instrument under which Taiwan's government has committed to no new quantum research arrangements with "authorities of countries of concern" belongs to that older tradition. It is Article 5.2, paragraph 9 of the Agreement on Reciprocal Trade between the United States and Taiwan, signed in Washington on 12 February 2026 under the auspices of the American Institute in Taiwan and the Taipei Economic and Cultural Representative Office, the two bodies through which the United States and Taiwan deal with each other in the absence of diplomatic relations. The agreement is signed; its own Article 7.5 makes entry into force depend on each side notifying that its internal procedures are complete, and the February USTR fact sheet says the Taiwan side will submit it to its legislature for review first. USTR's press-release listings from March to September 2026 carry no entry-into-force announcement that this post could find, so the obligations below are described as signed commitments. The clause reached Canadian readers on 14 September through an analysis in The Logic of what Ottawa can expect after its own talks with Washington collapsed, built on a paper by Carlo Dade and Alex Giordano of the University of Calgary's School of Public Policy. This post reads the Taiwan clause first, then the paper.
Practical takeaway. A government's commitment to enter no new science and technology arrangements with authorities of countries of concern, quantum computing named among the fields, is carried here by a bilateral trade arrangement rather than by an export-control regulation. Anyone who maps their obligations from the Export Administration Regulations, the EU dual-use regulation or the Wassenaar lists alone will miss it. Trade agreements, investment memoranda and their termination clauses belong on the same compliance map as the control lists.
What Article 5.2, paragraph 9 says, and what the agreement leaves undefined
Section 5 of the agreement is headed Economic and National Security. Article 5.2, "Export Controls, Sanctions, and Related Matters", has ten paragraphs. The ninth reads, in the agreement's own formula, that TECRO through its designated representative "shall adopt and maintain robust research security measures and not enter into any agreements, partnerships, or other similar arrangements involving science and technology cooperation with authorities of countries of concern, especially in critical and emerging technologies such as artificial intelligence (AI), biotechnology, and quantum computing." Three features of the drafting deserve attention.
First, the obligation is TECRO's, acting through its designated representative, which under Article 1.1 is the Taiwan side's own executive; it runs to authorities of countries of concern, and the agreement does not define that phrase in Article 5.2 or in its general definitions. It is a commitment about what the Taiwan side itself agrees to, and the text says nothing directly about individual researchers or universities. The nearest anchor for the country list sits seven paragraphs earlier: paragraph 2 requires measures against the diversion of advanced semiconductors, machine tools and advanced computing items to "covered nations" as defined in 10 U.S.C. § 4872, the US defense procurement statute whose list is China, Russia, Iran and North Korea. Whether a university-to-university memorandum with a Chinese provincial academy counts as an arrangement with "authorities", and whether the commitment reaches institutions the Taiwan side funds, is left to implementing practice. Second, the list of fields is illustrative, "such as", so the restriction is wider than the three technologies it names. Third, the clause pairs the commitment with a positive duty, "robust research security measures", which imports the vocabulary of US research-security policy into an obligation the Taiwan side will have to give content to, without specifying a standard.
The eight paragraphs around it: FDPR alignment, the Entity List, subsea cables, cranes and ground stations
Paragraph 9 sits inside a complete export-control chapter. Paragraph 1 commits Taiwan to align with export controls in force in the United States and to ensure they are not "backfilled or undermined". Paragraph 3 aligns Taiwan's rules with the US Foreign Direct Product Rule on semiconductors and related technology, the rule that extends US jurisdiction to foreign-made chips produced with US tools or software. Paragraph 5 requires legal reforms permitting catch-all controls for national-security purposes beyond weapons of mass destruction, and controls on intangible technology transfers. Paragraph 6 requires cooperation on restricting transactions by Taiwanese nationals with persons on the Commerce Department's Entity List and the Treasury's SDN and consolidated sanctions lists. Paragraph 7 requires Taiwan to phase existing technology from countries of concern out of its critical digital infrastructure, listing 5G and 6G networks, subsea cables, cloud systems, data centers, imaging equipment at airports and seaports, and port cranes. Paragraph 8 forbids hosting satellite ground stations for such countries.
Article 5.3 adds inbound and outbound investment screening on national-security grounds. Article 5.4 commits Taiwan to a defense budget above three percent of GDP, restated from an earlier Taiwanese announcement, and to shipbuilding cooperation among market economies. Enforcement, under Article 7.4, is by each party reviewing the terms and acting "in accordance with its law" after consultations when practicable. Article 7.3 allows amendment only by written agreement, entering into force sixty days after the later party notifies approval under its own legal procedures, while Article 7.6 lets either side terminate on six months' written notice. The trade side of the bargain, per the Calgary paper and the Global Taiwan Institute's reading of 25 February and the USTR fact sheet, is a signed US commitment to apply the higher of the most-favored-nation rate or 15 percent to originating Taiwanese goods, with identified product exceptions, down from the 32 percent announced in April 2025, alongside a companion investment agreement of 15 January 2026 for $250 billion in direct investment and $250 billion in credit guarantees.
How the Calgary paper of August 2026 places Taiwan among 21 Agreements on Reciprocal Trade
Dade and Giordano's paper, "The New ART of the Deal", was reviewed externally in June 2026 and published in the School of Public Policy's Research Workshop Series in August, days after Canada broke off its negotiations. Its subject is Canada, and its method is comparison: 21 Agreements on Reciprocal Trade concluded since 2025, four of them with countries that already had ratified free trade agreements in force. The authors' central finding is structural. The ARTs "range from formal FTAs to frameworks to memorandums to news releases", and all are "executive arrangements negotiated, implemented, modified and ended by President Donald Trump with no congressional participation, let alone approval". Nearly every one contains provisions constraining China's role in the partner economy, which the paper sums up as "a co-ordinated architecture of economic exclusion, built partner by partner".
Within that set, the paper places Taiwan and Indonesia as the most strategically significant, "geopolitical instruments as much as they do trade arrangements". Taiwan's FDPR alignment, it finds, is "a FDPR alignment commitment not matched in the other ARTs reviewed". Malaysia's deal, signed in October 2025, went furthest on breadth: a commitment to align with all US export controls "carrying equivalent restrictive effect", alongside $150 billion in purchases of US semiconductors, aerospace components and data-center equipment. Cambodia's agreement gave Washington the right to terminate if Phnom Penh entered any third-party arrangement that "undermines this Agreement or otherwise poses a material threat" to US economic or national security. Argentina's prioritizes the United States for copper and lithium over "market manipulating economies or enterprises". The Calgary authors read these as one template with local variation, and Dade told The Logic that Canada, which accepted a comparable non-market-economy review clause in Article 32 of CUSMA in 2018, "agreed already" to American influence over its trade policy.
Why a trade arrangement can carry a research commitment that export-control law never issued
The published US export-control rules on quantum computing control items and technology. The September 2024 interim final rule from the Bureau of Industry and Security added quantum computers, their components and certain cryogenic and error-correction technology to the Commerce Control List, with license requirements and a deemed-export dimension for foreign nationals, and the rule went through the Federal Register with public comment. None of those rules, so far as the public text goes, tells a partner government which research arrangements it may enter. A commitment of that kind by a partner government is a different kind of act. It sits outside the Export Control Reform Act, outside the Wassenaar Arrangement's consensus lists, and outside the notice-and-comment cycle. The instrument is the trade deal, and the deal's own enforcement clause, review and action "in accordance with its law", plus six-month termination, gives Washington the leverage and Taipei the exposure.
Quentir has watched the location of a control change before. When the Bureau of Industry and Security moved the United Arab Emirates between country groups in July 2026, one export-control table opened a new AI compute route, and the whole change lived inside the Export Administration Regulations, in a supplement any counsel could look up. The Taiwan clause moves in the opposite direction: a control that rulemaking has not issued, placed in a trade text outside the regulations altogether. The economic logic is set out in China's own effort to will a quantum industry into existence through state procurement and provincial programs; if the Taiwan side's implementing rules reach its laboratories and foundries, the commitment would touch exactly the pipeline those programs draw on. Prof. Mauritz Kop's August conversation on The Prode made the point that quantum export controls only work when allies share the perimeter; the ARTs are that perimeter being drawn one partner at a time, by contract rather than by regime.
What this means for a physicist in Hsinchu, and for Europe's research-security rules
For a quantum group at a Taiwanese university the clause raises implementation questions rather than an immediate rule. Once the agreement is in force, the Taiwan side will have to decide how "robust research security measures" reach publicly funded institutions, and whether a joint workshop, a shared testbed or a co-supervised doctoral student with a mainland institute affiliated to a state academy counts as an arrangement with "authorities of countries of concern". The text does not answer those questions; the implementing regime will. Quantum science is small enough that its collaborations are personal and its instruments are scarce; a dilution refrigerator slot or a photonic chip run is often shared across borders because no single lab can afford idle capacity, which is why the answers will matter to individual careers. The clause also arrives while Taiwan commits to a defense budget above three percent of GDP under the same agreement, so the science and the security spending move together, paid for by the same taxpayers.
Europe reached research security by a softer road. The Council of the European Union adopted a Recommendation on enhancing research security on 23 May 2024, addressed to member states and their universities, with due diligence and risk assessment as its tools and no named countries. The EU's own framework agreement with the United States of 21 August 2025 commits it to address "non-market practices, unfair competition and lack of reciprocity in public procurement with respect to third countries", language the Calgary paper places at the mild end of the spectrum. The Council text itself points to the restrictions that already apply, the dual-use regulation and national rules among them, so a European university is not unregulated. The difference is in the instrument. A European partner of a Taiwanese quantum center will be working with an institution whose government is bound, by trade contract, to the American perimeter, while the European side's research-security duty is framed at EU level as a recommendation on top of whatever binding national and dual-use rules apply. Two research-security regimes will meet in the same consortium, one contractual and enforceable by tariff, the other a non-binding recommendation that sits alongside the binding EU and national rules that already apply.
How Quentir Reads It
The first thing this clause tells us is where the control is. The quantum export-control literature, including work Prof. Kop co-authored on quantum technology export controls, has focused on the Commerce Control List, the Wassenaar dual-use list and the EU regulation. The AIT-TECRO agreement shows that a research-cooperation commitment can be written without touching any of them. A compliance office that reads only the lists will report, accurately, that no control-list entry bars a joint quantum program between its Taiwanese partner and a Chinese academy, and it will have missed the question the partner's government has agreed to answer.
The second is durability. CoCom lasted forty-four years without a treaty, but it had a multilateral consensus behind it. An ART has one signature on each side, review "in accordance with its law", written amendment under Article 7.3 and a six-month exit under Article 7.6. The Calgary paper's warning to Canada, that existing agreements "do not guarantee protection against new U.S. demands", cuts the other way too: a research commitment made by executive arrangement can be amended by written agreement or ended by notice, and a Taiwanese laboratory planning a five-year program has no way to know which. The Calgary authors note that Washington preserves earlier agreements "only if the terms align with current American interests", so the ban and the six-month exit have to be read together.
The third is the precedent for everyone else. The paper counts 21 partners; Mexico and Canada are described as the 22nd and 23rd in waiting. If the Taiwan clause becomes the template for technology-heavy partners, the map of which governments may enter which quantum research arrangements will be written in trade texts signed in Washington, and the European recommendation of May 2024 will be the softest instrument in the room. The question for a European research council is simpler than the trade law: when its Taiwanese partner asks whether a third institution counts as an authority of a country of concern, who in Europe is going to answer? The earlier Quentir reads this post draws on, the export-control route analysis, the Defense Monitor's account of China's quantum industrial policy and the Prode conversation, sit in one archive; the All-access membership is the single subscription to all of it, refreshed as the next ARTs land.
Sources: Office of the United States Trade Representative, Agreement Between the American Institute in Taiwan and the Taipei Economic and Cultural Representative Office in the United States on Reciprocal Trade Between the United States of America and Taiwan (Section 5, Articles 5.2 to 5.4; Articles 7.4 to 7.6; Annex 3), and "Ambassador Greer Oversees Signing of U.S.-Taiwan Agreement on Reciprocal Trade", press release, 12 February 2026. Carlo Dade and Alex Giordano, "The New ART of the Deal: Trump Administration Agreements on Reciprocal Trade and Canada", University of Calgary School of Public Policy, Research Workshop Series, August 2026 (PDF RWS-ArtofDeal-Final; executive summary, "ART Agreement Analysis" pp. 6-10, "What the ART Agreements Reveal" pp. 10-13). The Logic, "The 'red line' in U.S. trade talks that sets Canada apart from other countries", 14 September 2026 (Dade quotations; 19-deal count as reported there). Riley Walters, "What's In the New US-Taiwan 'Agreement on Reciprocal Trade'?", Global Taiwan Institute, 25 February 2026 (signing date, tariff rate, 15 January 2026 investment companion). Council of the European Union, Council Recommendation of 23 May 2024 on enhancing research security (EUR-Lex 32024H03510). All public pages read 14 September 2026.
Published intelligence, built to inform your own decisions. Published: September 14, 2026.