—·
By moving from item lists to entity watchlists, China effectively imposes end-use licensing and re-export liability rules that raise tech supply-chain compliance costs.
On February 24, 2026, China’s Ministry of Commerce restricted exports of dual-use goods to 40 Japanese entities split across an export control list (20) and a separate watch list (20)—and the structure matters because it turns compliance into a determinate “where you can route components” question, not merely a “whether you can ship” checkbox. (AP News: apnews.com)
In plain supply-chain terms, this is what a technology trade agreement looks like when it is enforced not through tariffs or “mutual recognition,” but through licensing choreography: documented end-use, entity verification, and—crucially for multinational firms—re-export risk that travels with the component through subcontractors and overseas processing.
That is why the more consequential development is not the diplomatic narrative around Japan’s “remilitarization.” The more technical—and more durable—shift is the compliance logic: China is using dual-use export controls to impose a de facto agreement on downstream behavior across corporate networks, even where goods are not explicitly itemized for a specific AI capability.
To appreciate the mechanism, you have to read these rules the way compliance teams do: as a layered regime of (1) controlled “dual-use items,” (2) restricted destinations/end users, (3) licensing and documentation duties, and (4) penalties and legal consequences that create deterrence for both exporters and third parties.
The February 24 measure is structurally important: 20 entities are placed on an export control list (where exporters face a ban on importing/using the controlled dual-use items with those entities), while 20 others sit on a watch list requiring individual export license applications, alongside risk assessment reports and written pledges that dual-use items will not be used by Japan’s military. (AP News: apnews.com)
What the tiering does—beyond sorting names—is create two different operational clocks for firms. An export control list triggers “stop-and-replace” behavior because the baseline is prohibition for transactions involving the named end user with relevant dual-use items. A watch list, by contrast, forces firms into an evidence-and-timing game: the entity can remain a plausible counterparty, but only if the exporter (and often upstream suppliers supporting the application package) can produce a regulator-credible narrative that the end use will not drift into disallowed military support.
In other words, the practical compliance distinction is less about optics (“ban” versus “license”) and more about what compliance teams have to scale across the supply chain:
The implication for AI-adjacent supply chains is that “dual-use” is not confined to obvious defense hardware. Dual-use language is broad enough to sweep in components that are foundational to modern manufacturing and sensing—items that firms routinely treat as civilian-procurement normal. Once these items are subject to an end-use/end-user framework tied to Japan military capability enhancement, the compliance posture changes from “spec-based procurement” to “relationship-based procurement.”
In other words, this is less like a single trade restriction and more like a compliance-and-end-use regime that compels firms to build a routing discipline: which vendors, which subsidiaries, which intermediaries, which geographies—and which documentation trails—are acceptable.
Japan’s own system also emphasizes end-use verification and licensing under the Foreign Exchange and Foreign Trade Act framework, with exporters needing licenses when transactions are designated as obstructing peace and security, and with the possibility of being required to obtain documents from intended end users. (METI: meti.go.jp) Even though this is not China-specific, it matters because it shows the shared global pattern: end-use licensing has become a governance substrate across jurisdictions—so restrictions elsewhere can be “felt” as operational constraints at the interface layer.
China’s dual-use approach is increasingly end-use/end-user centered rather than purely item-list centered—meaning the compliance cost is less about classification of a single part number and more about demonstrating that the whole transaction chain is consistent with permitted end use.
The core technical reason is that end-use licensing forces firms to prove negative: that a component will not be used in a disallowed manner. That proof is built from documentation, contractual flow-downs, and ongoing compliance review—work that does not scale linearly with the volume of transactions, because it scales with the uncertainty companies must manage.
China’s December 1, 2024 implementation of its “Regulation on the Export Control of Dual-Use Items” codifies a licensing regime that covers transfers and includes provisions for end-use certification and legal liability frameworks. (Orrick summary of the Regulation effective December 1, 2024: orrick.com) This matters because the regime is not just an “announcements-based” crackdown; it is anchored in a more systematic administrative architecture.
When end-use licensing is operationalized, “tech supply chain compliance” becomes an engineering constraint. For multinational manufacturers, a component might be technically capable of multiple functions; the end-use licensing standard turns that latent multifunctionality into an evidentiary and legal burden. The moment a destination entity is categorized as contributing to enhanced military capabilities, firms lose the default assumption of benign civilian use and must treat procurement as an audited compliance pathway.
And for re-export liability, the burden becomes structural. A multinational that buys a Chinese-origin dual-use component may not be “the exporter” in the first transaction, yet it can still face downstream risk if the legal framework asserts responsibility for transfers or provision of controlled items to listed entities or for end-use violations. China’s regime also contemplates legal liability for aiding or abetting circumvention of export control regulations. (Orrick: orrick.com)
Consider the January 6, 2026 escalation reported by AP: China banned exports of dual-use goods that could serve military purposes to Japan, stating that violating individuals/organizations face legal consequences “regardless of where they are from.” The notice did not list specific product categories, but it explicitly framed the restrictions around military users and end uses that enhance Japan’s military power. (AP News: apnews.com)
This is an essential piece of the compliance story because it demonstrates how the regime can shift quickly from a more “list-like” posture to a broader end-use/end-user standard. When firms hear that exports to military users and those end users/uses that contribute to enhanced military capabilities are prohibited, the practical question becomes: Do we risk being interpreted as facilitating a prohibited end use even if the buyer claims otherwise?
That is where “cessation requirements” enter as a compliance discipline. The February 24 announcement for the export control list entities uses language that ongoing related activities must stop immediately for those entities and that foreign organizations or individuals are banned from providing China-origin dual-use items to those entities. (AP News: apnews.com) That effectively forces immediate supply-chain reconfiguration, not phased remediation.
For re-export liability, compliance teams need to model not only the initial export but also subsequent transfers, processing, and component integration that could be interpreted as enabling prohibited end-use. Legal commentary on China’s dual-use export-control framework highlights that re-export and extraterritorial aspects are addressed through the regulatory system and the mechanisms under the Export Control Law and its implementing regulations. (Squire Patton Boggs consolidated/implementation discussion (Dec 1, 2024 effect) PDF: squirepattonboggs.com)
Even for firms that believe their transactions are benign, the regime introduces a compliance “friction tax”—the costs of classification review, end-use document generation, entity screening, contract redlining, and audit trails.
The global baseline is that firms already operate under end-use certificate workflows; Japan’s METI materials show how licensing may require detailed end-use confirmation, and Japan publishes end-use certificate forms intended for presentation to the export control authority. (METI Form 2 / end-use certificate: meti.go.jp) That same infrastructure—if practiced with higher frequency and stricter thresholds due to foreign restrictions—becomes the operational cost that a de facto technology trade agreement imposes.
Three numbers make the compliance implications tangible:
40 Japanese entities are affected in the February 24, 2026 measure: 20 on an export control list and 20 on a watch list that requires individual export license applications, risk assessments, and written pledges. (AP News: apnews.com)
The January 6, 2026 ban is framed as prohibiting exports to Japanese military users and other end users/end uses that contribute to enhancing Japan’s military capabilities—the point being that the compliance standard is not confined to a named set of products; it is tied to end use and end-user interpretation. (AP News: apnews.com)
China’s dual-use export control regulation regime takes effect in a unified way from December 1, 2024, with an implemented Regulation on the Export Control of Dual-Use Items and an administrative framework that includes licensing and legal liability structures. This anchoring matters because it makes today’s announcements operate within a durable compliance system rather than a purely ad hoc enforcement posture. (Orrick: orrick.com)
The quantitative takeaway is not “how many companies were named” in isolation; it’s how naming plus tiering plus licensing requirements creates a compliance map that firms can treat like a governance protocol. Once such maps exist, supply-chain design decisions follow: which BOM (bill of materials) components have China-origin risk, which suppliers can produce equivalent parts without crossing the same threshold, and which logistics or integration steps trigger re-export liability concerns.
Below are concrete, documented examples that illustrate how compliance regimes translate into outcomes with timing and identifiable entities.
AP reports that the export control list includes multiple business subsidiaries of Mitsubishi Heavy Industries involved in shipbuilding and aircraft-engine/maritime machinery, and that these exports are restricted such that “ongoing related activities must cease immediately” for those listed entities. (AP News: apnews.com)
Why this matters operationally: “cessation immediately” converts a sanctions-style constraint into a supply-chain redesign problem: the compliance response cannot wait for end-use paperwork because the problem is the counterparty relationship itself. Firms commonly respond by (i) swapping suppliers for any China-origin dual-use components used in those programs, (ii) revising engineering BOMs to preserve performance while avoiding controlled routing, and (iii) renegotiating contracts with flow-down language that blocks restricted destinations.
Outcome: Export access is curtailed for China-origin dual-use items tied to those entities.
Timeline: February 24, 2026.
AP reports that the watch list includes entities such as Subaru Corporation and Mitsubishi Materials Corporation; for these, Chinese exporters must submit individual export license applications, risk assessment reports, and written pledges that the dual-use items will not be used by Japan’s military. (AP News: apnews.com)
Why this matters operationally: A watch list does not merely delay shipments; it creates a documentation bottleneck that must be repeated per deal or per defined licensing scope. In practice, that pushes firms to (i) pre-build “license-ready” end-use packages for specific buyer programs, (ii) tighten internal controls around who can request exports, (iii) ensure subsidiaries and distributors do not re-route components into the wrong end use, and (iv) keep traceable records for audit or enforcement.
Outcome: Business is not necessarily shut, but it becomes conditional on licensing and documented end-use discipline.
Timeline: February 24, 2026.
AP includes the Institute of Science Tokyo among the entities listed on the watch list. (AP News: apnews.com)
Why this matters operationally: Research institutions often sit downstream of general-purpose equipment and components with multiple legitimate applications. Placing a research organization on a watch list signals that the compliance assessment will focus on end-use intent and downstream conversion—meaning procurement departments may need to coordinate with principal investigators and technology-transfer offices to prevent “benign” scientific use claims from appearing implausible under a military-capability-enhancement lens.
Outcome: Even research organizations can fall under dual-use end-use licensing discipline, which typically raises procurement friction for lab equipment and components that are technically dual-capable.
Timeline: February 24, 2026.
Japan’s METI describes how the Foreign Exchange and Foreign Trade Act empowers licensing requirements when export transactions are specified as having possibilities of obstructing peace and security, and notes exporters may need to obtain documents from intended end users and submit them to METI. (METI: meti.go.jp) Japan also provides an end-use certificate form used for presentation to METI. (METI Form 2: meti.go.jp)
Why this matters operationally: This is the institutional baseline firms must “satisfy” to operate cross-border. When China’s licensing posture tightens on named entities, Japanese importers and integrators cannot rely on domestic documentation alone; they often need an end-use narrative that remains coherent when tested against China’s dual-use end-use standard.
Outcome: This shows the compliance interface is already institutionalized—so China’s end-use licensing posture can “plug into” existing Japanese compliance practices, amplifying the overall compliance cost across the tech supply chain.
Timeline: METI framework is ongoing; the specific form and guidance materials are current and represent the operational baseline used by exporters.
Together, these cases show the real business effect: companies do not merely respond by changing destinations; they re-engineer procurement logic, documentation systems, and third-party routing decisions to avoid triggering cessation or penalty risk.
This is the heart of the de facto “technology trade agreement” framing. Under an end-use licensing regime with re-export liability considerations, firms route by control relevance, not by logistics convenience.
Practically, firms must treat these categories as routing constraints:
Compliance tooling becomes the operating layer. While many companies use internal screening, the architecture typically depends on:
Even if a firm does not publicly name its tools, the underlying tasks are consistent: origin tracking, end-user mapping, and evidentiary retention. And when the legal framework explicitly contemplates licensing and legal liability under the Regulation, compliance is no longer optional “paperwork”—it becomes the decisive constraint on how a semiconductor/AI-adjacent supply chain can keep moving.
A recurring expert lesson from export-control practice is that end-use licensing shifts cost from “one-time classification” to “ongoing verification.” Japan’s METI materials and guidance emphasize verification and end-use confirmation in licensing practice. (METI: meti.go.jp)
Under China’s dual-use Regulation framework, legal liability provisions and administrative licensing duties expand the compliance footprint. Commentary on the Regulation highlights how the licensing regime and legal liability mechanisms exist to manage circumvention and end-use/end-user changes. (Orrick: orrick.com)
The export-control compliance community also treats entity screening and end-use documentation as “system” problems rather than “document” problems. BIS guidance in the U.S., for instance, stresses that license applications and end-use checks require complete, accurate information upfront to fit regulatory time limits—an operational logic that mirrors the end-use licensing reality across jurisdictions. (BIS guidelines: bis.gov)
That is why framing China’s actions as a de facto technology trade agreement is analytically useful. A trade agreement sets rules and disciplines outcomes. Here, the “rules” are export licensing and end-use/end-user compliance requirements; the “outcome” is which components can be routed and under what documentary conditions.
If this regime continues along the same design principles evidenced by the February 24, 2026 40-entity tiering (20 list, 20 watch) and the January 6, 2026 end-use framed ban logic around military capability enhancement, then the next predictable shift is not a single headline restriction—it is the consolidation of compliance expectations into standardized internal processes.
Forecast (Q3 2026): By Q3 2026, many semiconductor and AI-adjacent suppliers operating in multi-geo supply chains are likely to treat China-origin dual-use component risk as a standing procurement gate—moving from ad hoc screening toward formal BOM origin mapping, documented end-use workflows, and license-ready contracting. This forecast is grounded in the demonstrated durability of China’s dual-use Regulation effective December 1, 2024 and the operationalization shown in early 2026 through entity tiering and end-use/end-user standards. (Orrick: orrick.com; AP News: apnews.com)
Policy recommendation (concrete actor): The Government of Japan (METI) should publish an “end-use licensing interface” workbook targeted at exporters and integrators on how to document and evidence end-use/end-user intent under Japan’s export control system when counterparties face extraterritorial dual-use restrictions. Japan already provides end-use certificate materials and explains licensing triggers under the Foreign Exchange and Foreign Trade Act framework; the next step is to connect those documents to the operational needs of multinational end-to-end supply chains in a way that reduces rework and licensing back-and-forth—specifically by adding:
After all, the real lesson is not that technology trade agreements have returned. It’s that compliance-and-end-use governance can already function as a trade agreement—without signatures—by defining the conditions under which firms may route components, re-export risk, and certify end-use.
Export licensing is becoming a supply-chain audit standard. The next policy step is to make evidence portable, reduce friction, and protect against diversion risk.
The rescission of a proposed AI-chip export rule marks a pivot: from one-size compliance to tiered, partner-coordination licensing—where “governance commitments” quietly replace tariff-style deals.
Export controls are forcing buyers and assemblers to rebuild procurement and verification across borders, with ASEAN intermediaries becoming the compliance stress test.