1. Introduction
The release of DeepSeek-R1 in January 2025 marked what some described as a ‘Sputnik moment’ for the US.Footnote 1 Built on DeepSeek-V3, R1 arguably rivalled frontier US systems, with disclosed compute-equivalent costs of about US$5.9 million, far below Western estimates, although the disclosed figure has been contested as an incomplete measure of its true development cost.Footnote 2 The Nasdaq lost over \$1 trillion in market value, while Nvidia saw a record single-day decline of nearly $600 billion in capitalisation, the largest such loss in US stock market history.Footnote 3
DeepSeek’s rise also exposed a harder question: Had China, as widely reported, continued to access restricted hardware notwithstanding the progressive tightening of US export controls? Unlike bulky chipmaking equipment, chips are compact, easily concealed within end-use products, and comparatively difficult to detect and physically intercept.Footnote 4 Such structural vulnerabilities are compounded by deliberate circumvention. Chinese firms have reportedly exploited advance stockpiling,Footnote 5 design-out strategies,Footnote 6 and shell companies that route restricted hardware via ostensibly compliant supply chains.Footnote 7 Containing that diversion has become Washington’s pressing concern.Footnote 8 In April 2025, the House Select Committee on the Chinese Communist Party concluded that DeepSeek seemed to be powered by tens of thousands of Nvidia chips restricted from export to China, acquired through third-country intermediaries.Footnote 9 The Committee’s findings document persistent circumvention of US export controls; its calls for expanded controls implicitly concede that more remains to be done across the globalised semiconductor supply chain that no single jurisdiction can govern alone.Footnote 10
The Committee’s subsequent investigation in 2026 documented an escalating pattern of circumvention, identifying a series of prosecutions involving sophisticated smuggling networks that route restricted hardware through front companies and permissive transit jurisdictions, with Singapore, Malaysia, and Thailand appearing repeatedly as routing points.Footnote 11 A Federal prosecution in March 2026 makes the point explicit. The US Attorney for the Southern District of New York indicted three individuals affiliated with server manufacturer Supermicro for allegedly conspiring to divert to China, via a Southeast Asian company, servers worth at least $510 million containing export-controlled Nvidia GPUs.Footnote 12
Since 2018, the US has progressively expanded the scope of its semiconductor export controls, moving from entity-based controls toward a technology- and country-based architecture encompassing advanced chips, semiconductor manufacturing equipment, electronic design automation (EDA) tools, and, most recently, AI model weights.Footnote 13 Various rounds of rulemaking by the US Department of Commerce’s Bureau of Industry and Security (BIS), including the October 2022 interim final rule,Footnote 14 the October 2023 updates,Footnote 15 and the December 2024 Foreign-Produced Direct Product Rule (FDPR) expansions,Footnote 16 have sought to deny China access, not only to controlled US-origin items but also to specified foreign-produced items that satisfy the product-scope and destination, end-user, end-use or knowledge conditions of the applicable FDPR.Footnote 17 China has formally contested these measures in the WTO, alleging inconsistencies with the GATT 1994, the TRIMs Agreement, the TRIPS Agreement, and the GATS. The US accepted China’s requests to enter consultations without prejudice to its position that the challenged measures concern national-security issues not susceptible to review or capable of resolution through WTO dispute settlement.Footnote 18 Consultations were held in March and October 2023, but the dispute has not progressed to panel proceedings.Footnote 19
What the WTO episode reveals is more instructive than its outcome. For the US, extraterritorial export controls are an instrument of strategic competition; their legality under international trade law is treated as secondary to their effectiveness as policy. However, for third-country jurisdictions, the experience is structurally different. The extraterritorial reach of the US export controls is increasingly experienced not as a bilateral US–China matter but as direct pressure on their own regulatory autonomy, pushing domestic law and practice toward alignment with US objectives notwithstanding divergent sovereign preferences. The trade-law debate has largely mirrored the Geneva frame, asking whether US export controls can be squared with WTO law. What it leaves out are the jurisdictions pressed to follow the US lead. Transactions routed through third countries run under two layers of law at once: US controls, which follow the item extraterritorially through licence conditions and the compliance practices firms adopt in anticipation of them, and the transit State’s own regime. Because the two layers can be designed for different ends, the space between them generates both pressure to align and room to circumvent. The dynamics themselves are no secret in the WTO. What cannot be seen from there, however, is how third countries regulate in practice, and how far their domestic laws bend, or hold, under such pressure. It is that detail which will shape how trade rules governing such measures are negotiated and practised, day to day and away from Geneva.
How that pressure operates, and what it demands, depends critically on a jurisdiction’s position in the semiconductor supply chain. It also depends on the doctrinal categories through which the receiving jurisdiction’s law constructs the regulable transaction in the first place, an inquiry that requires comparative analysis of the two regimes’ jurisdictional architectures. The existing literature has not drawn this distinction with sufficient precision. The academic debate has been organised primarily around what some have coined ‘chokepoint economies’.Footnote 20 A small group of economies occupies structurally irreplaceable positions in the semiconductor supply chain: Taiwan in advanced logic chip fabrication, the Netherlands in extreme ultraviolet lithography, and Japan in manufacturing equipment and materials, for instance. For these economies, the US concern is supply-side: the ask is that they restrict what they produce and export from those nodes.
A second category of economies occupies a different position in the supply chain. They may be termed ‘hub economies’: jurisdictions that do not control the most-advanced fabrication or equipment nodes but whose logistical, financial, and corporate infrastructure makes them significant transit and distribution points for the global semiconductor trade.Footnote 21 For hub economies, the US concern is not supply-side but transit-side. The ask is not ‘restrict what you make’ but ‘intercept what passes through you’. These economies attract US scrutiny not because of what they produce but because of what they enable. The regulatory demands that flow from this position are structurally different from those that chokepoint alignment analysis describes, and the existing literature has not examined them as such.Footnote 22
Singapore is the paradigm case. Singapore accounts for around 10% of global chip output, 20% of global semiconductor manufacturing equipment production, and emerges as ASEAN’s premier destination for semiconductor foreign direct investment, attracting major FDI from the US, Taiwan, and Europe.Footnote 23 That footprint is concentrated in mature-node and specialty fabrication, assembly and testing, and equipment segments outside the most advanced controlled categories. Even so, Singapore remains the premier semiconductor location in Southeast Asia, its strategic position and mature legal and financial infrastructure making it the region’s most integrated node in the US-led semiconductor ecosystem. More crucially for present purposes, it is the place through which advanced hardware moves: routed, transhipped, and redistributed across the region’s supply chains. The same position explains why Singapore has been repeatedly reported as one of the principal conduits for chip diversion to China.Footnote 24 That characterisation has attracted more accusation than rigorous legal analysis. While Singapore is not a chokepoint economy in the Taiwan or the Netherlands sense, the absence of serious legal scrutiny is striking for a jurisdiction of this weight and exposure.
This article aims to fill that gap through three interrelated inquiries. Section 2 first compares the Singapore and US export-control regimes across four dimensions – legislative purposes, controlled modalities, controlled items, and the treatment of entities and destinations – showing that they rest on different jurisdictional premises. That divergence frames two further questions addressed in Section 3. The first is what tools, if any, Singapore law provides to address US concerns over semiconductor diversion, and how effective they are when the regulatory object shifts from physical goods to intangible knowledge. The second is, then, how far those tools can remain viable amid Singapore’s deeper integration into the US-led semiconductor ecosystem and the expanding reach of US controls. Section 4 concludes.
2. Two Regulatory Architectures: Singapore and the US Compared
Export-control regimes may regulate the same items for different purposes and through different jurisdictional connections. The comparison below is, therefore, deliberately confined to those portions of the Singaporean and US dual-use controls directed toward non-proliferation and security ends; it neither treats either system as monolithic nor asserts equivalence between them as a whole. Within that defined field, the analysis considers both what each regime controls and when its jurisdiction attaches. Singapore’s regime generally requires a territorial nexus to the goods, technology, or brokering activity. The US regime is likewise territorial, but its jurisdiction may also follow an item abroad because of its US origin, controlled US content, or satisfaction of an applicable FDP rule.
The resulting divergence is not a deficiency in Singapore law; it is part of the premise of the US extraterritorial turn. Yet US extraterritorial controls do not displace the domestic law of jurisdictions through which regulated items move. Unless Singapore chooses that alignment, its response to diversion must therefore operate through categories supplied by its own law. Sections 2.1–2.4 compare the regimes’ legislative purposes, controlled modalities, items, and treatment of entities and destinations, while Section 2.5 relates their designs to their respective positions in the semiconductor supply chain. Together, these comparisons supply the foundation for asking, in Section 3, what tools, if any, Singapore law provides to address diversion-related conduct without formal alignment with US export controls, and where their limits lie.
2.1 Legislative Purposes: Multilateral Non-Proliferation against National Security
Singapore’s primary export control instrument is the Strategic Goods (Control) Act 2002 (SGCA), effective from 1 January 2003. Its long title describes it as an act ‘to control the transfer and brokering of strategic goods, strategic goods technology’, including goods and technology capable of being used ‘to develop, produce, operate, stockpile or acquire weapons capable of causing mass destruction, and missiles capable of delivering such weapons; and for purposes connected therewith’.Footnote 25 The SGCA is driven by two distinct rationales, as the Singaporean Parliamentary record reveals. The regime was designed to fulfil Singapore’s non-proliferation obligations ‘without impeding normal legitimate trade’.Footnote 26 Beyond security, compliance with global non-proliferation efforts was framed as a source of competitive advantage: a robust export control system would ‘secure greater access to high-tech equipment and technology from more advanced economies’ and ‘safeguard the reputation and integrity of Singapore’s status as a major international shipping hub’.Footnote 27 The SGCA’s export control provisions were substantively extended once, by the Strategic Goods (Control) (Amendment) Act 2007, in response to UN Security Council resolutions on Iran and the Democratic People’s Republic of Korea (DPRK).Footnote 28 Administration of the SGCA is vested in Singapore Customs, designated as the national authority by the Ministry of Trade and Industry (MTI).
The SGCA is primarily an ‘item-based’ instrument in what it controls: it identifies goods and technology by reference to their technical characteristics and weapons of mass destruction (WMD) applicability. The jurisdiction to control them, however, attaches territorially. Destination- and entity-based controls exist, as examined below, but they operate as supplementary mechanisms rather than as the regime’s organising logic. Specifically, its substantive scope is defined by Section 4A, which empowers the Minister for Trade and Industry (the ‘Minister’) to prescribe any ‘military’ or ‘dual-use goods’ as ‘strategic goods’, and any ‘military or dual-use technology’ as ‘strategic goods technology’, for the purposes of this Act.Footnote 29 A permit is required to export a controlled item to any destination, irrespective of end user, because the item falls within the controlled list. Notably, while Singapore does not join any of the four principal multilateral export control regimes (the Wassenaar Arrangement, the Australia Group, the Missile Technology Control Regime, and the Nuclear Suppliers Group), it has voluntarily aligned with these regimes.
As the Minister of State for Trade and Industry stated in Parliament in 2007, ‘as a responsible transhipment hub, we should do our part for global security by implementing safeguards to curb the illicit movement of such goods’ and it was ‘in Singapore’s interest to tighten our present regime and adopt the control list covered under the four multilateral export control regimes’ in order ‘to keep our export control system updated and aligned with the international non-proliferation regimes’.Footnote 30 That choice reflects a convergence of non-proliferation obligations, hub-economy reputation, and national commercial interest. It also marks a deliberate boundary at the core of what this article calls ‘doctrinal independence’: Singapore’s controlled-items list tracks the multilateral non-proliferation consensus rather than unilateral US controls. Doctrinal independence, in this sense, is a feature of the regime’s design rather than a lacuna in it. The divergence traced below hence reflects the progressive movement of US controls beyond the multilateral baseline to which Singapore remains anchored, rather than any shortfall in Singapore’s regime.
SGCA’s operational detail is elaborated in the Strategic Goods (Control) Regulations (SGCR),Footnote 31 governing permit applications, brokerage registration, and record-keeping obligations. The Strategic Goods (Control) Order (SGCO) is issued by the Minister under the SGCA, prescribing the technical scope of the regime. The SGCO 2025 is the operative instrument at the time of writing.Footnote 32
The SGCA does not operate in isolation. It functions as lex specialis within Singapore’s broader trade control architecture: a specialised regime governing strategic goods and technology, embedded within the general framework established by the Regulation of Imports and Exports Act 1995 (RIEA).Footnote 33 The RIEA empowers the Minister to regulate all goods imported into, exported from, transhipped in, or in transit through Singapore,Footnote 34 with day-to-day administration vested in the Director-General of Customs.Footnote 35 The relationship between the two instruments is doctrinally explicit: the Regulation of Imports and Exports Regulations (RIER) prohibits the import, export, or transhipment of any goods prohibited ‘under any other regulations made under the Act or any other written law’, providing the cross-statute mechanism through which SGCA prohibitions are enforced through the RIEA.Footnote 36
RIER separately prohibits the import, export, transhipment, or transit of goods contravening applicable UN Security Council (UNSC) sanctions.Footnote 37 The RIEA defines ‘relevant decision of the United Nations Security Council’ as any decision made under Chapter VII of the UN Charter.Footnote 38 Its transit-specific enforcement provisions are confined accordingly: enforcement powers over goods in transit are activated only where a UNSC inspection requirement exists.Footnote 39 The general power in section 3 is broader, as the Russia measures discussed in Section 2.4 show, but the framework contains no conduit for automatic incorporation of another State’s unilateral technology-competition controls.
The regulatory divergence between Singapore and the US is evident in their statutory objectives. While the SGCA, operating within the broader RIEA framework, is fundamentally a non-proliferation instrument anchored to multilateral regimes and UNSC resolutions, the US Export Control Reform Act 2018 (ECRA) adopts a substantially more expansive mandate, framed throughout in terms of US national security concerns.Footnote 40 ECRA Section 4811 identifies a wide range of purposes, including not only the proliferation of WMD and conventional weapons, but also preserving the qualitative military superiority of the US, strengthening its defence industrial base, and carrying out foreign policy.Footnote 41 Driven by national security grounds, the ECRA further directs the US government to maintain its STEM leadership, including foundational technology essential to innovation,Footnote 42 and mandates a regular process to identify and regulate emerging and other types of critical technologies of concern, regardless of the nature of the underlying transaction.Footnote 43
Indeed, the ECRA itself lists compliance with ‘obligations and commitments under international agreements and arrangements, including multilateral export control regimes’ as one of its stated purposes,Footnote 44 and expresses a preference for multilateral coordination over unilateral controls applied to items widely available from foreign sources.Footnote 45 In operational practice, however, several rounds of BIS rulemaking have relied on unilateral measures targeting China’s access to advanced chips, semiconductor manufacturing equipment, and AI-relevant technologies. These measures have introduced different parameters that reach well beyond any multilateral consensus, a departure from Singapore’s multilateralism-oriented export control framework.
2.2 Controlled Modalities: Territorial Activity against Item-Following Reach
The SGCA regulates three categories of conduct: physical movement of goods through Singaporean territory, intangible transfer of technology with a Singaporean connection, and brokering activities conducted from Singapore. Each is examined in turn against its counterpart under the Export Administration Regulations (EAR).
2.2.1 Physical Movement: The Logic of Logistical Intermediation
The SGCA regulates three categories of physical movement. Section 5(1)(a) prohibits any person, absent a permit, from exporting, transhipping, or bringing in transit any strategic goods. Each is a statutorily distinct activity defined separately.Footnote 46 While often grouped together, each of these modalities reflects different regulatory rationales closely tied to Singapore’s role as a global logistics hub. ‘Export’ represents the most conventional modality. It captures the outward movement of goods from Singapore and is triggered by the act of taking goods out of the territory by land, sea, or air.Footnote 47 In this respect, the SGCA follows the classic territorial model of export control: jurisdiction attaches because the goods originate from Singapore.
Transhipment, by contrast, reflects Singapore’s critical role as a global redistribution hub.Footnote 48 It involves unloading goods from one conveyance and reloading on to another within Singapore for onward export, often documented via a through bill of lading or manifest.Footnote 49 In practice, this typically occurs within Free Trade Zones (FTZs), where goods may be temporarily stored, consolidated, or reconfigured before being sent onward without entering Singapore’s domestic market.Footnote 50 This modality is crucial for this city-state: a significant share of cargo moving through Singapore is not destined for local consumption but rerouted across global supply chains. Transhipment, therefore, introduces an intermediary stage in global circulation, where goods are physically handled within Singapore even though it is not their ultimate origin or destination. This physical intervention, the offloading and reloading itself, may create a window where goods may be diverted or repackaged. The SGCA’s robust licensing scheme is thus crafted to ensure that the hub’s infrastructure is not exploited for illicit transfers.
Transhipment must be distinguished from ‘bring in transit’.Footnote 51 Goods brought in transit remain on the same conveyance and pass through Singapore without operational handling; transit represents a thinner form of jurisdictional contact, closer to supervised passage than logistical intermediation. Because transit cargo never leaves the original vessel or aircraft, the opportunity for diversion in Singapore is, in theory, lower than in transhipment. The SGCA’s inclusion of these modalities reflects a layered approach to hub governance: it regulates not only goods originating in Singapore, but those utilising its logistical handling (transhipment), while extending a lighter, observation-based control to goods that merely traverse its territory (transit).
The EAR has no equivalent to transhipment or transit as distinct controlled activities. Instead, it addresses the onward movement of controlled items through the broader concept of ‘reexport’, defined as the shipment or transmission of an item subject to the EAR from one foreign country to another.Footnote 52 The SGCA has no comparable activity-level control for foreign-to-foreign physical movement. For instance, a controlled chip transferred from Taiwan to Malaysia requires a US licence determination as a reexport. This rule applies even though the transaction has no US territorial nexus, as long as the chip is subject to the EAR. The same transfer falls outside the SGCA’s physical-movement controls entirely unless the goods pass through Singaporean territory and are listed on the SGCL. The SGCA can reach such a transfer only indirectly, through the brokering regime under Section 6, and only where a Singapore-based intermediary arranges the transaction, a mechanism examined below.
The US architecture extends its conception of physical control further still by regulating ‘in-country transfers’,Footnote 53 defined as a change in end-use or end-user of a controlled item within a single foreign country. For Singapore, ‘movement’ is geographic: the SGCA is triggered when goods cross a territorial boundary and is indifferent to a transaction between two Singapore-based entities so long as the controlled item does not leave the country. For the US, ‘movement’ is relational: a change in the party who controls, possesses, or uses a controlled item constitutes a regulable transfer, even where the item remains physically stationary. The EAR may, for instance, require a US licence for the sale of a controlled chip from one Singapore-based firm to another if the new end-user is on the Entity List or the transfer involves a change to a military end-use. US physical control does not stop at the border; it penetrates the domestic commerce of the hub, regulating changes in possession even when the goods remain stationary within the jurisdiction.
2.2.2 Intangible Transfer of Technology (ITT): Territorial Anchors
The SGCA regulates two further technology-related modalities, both targeting strategic goods technology rather than goods: the export of any document in which such technology is recorded, stored, or embodied,Footnote 54 and the transmission of the technology itself.Footnote 55 Of these, transmission is the modality whose territoriality must be expressly defined, since electronic transfers lack the intuitive territorial nexus that physical movement supplies. Under SGCA, technology is transmitted only if it is sent ‘in Singapore by electronic means’ (e.g. email or fax) or where it is made available ‘in Singapore on a computer’ so as to be accessible to a person in a foreign country. Singapore Customs’ operational guidance confirms this, clarifying that the ITT regime captures not only active electronic transmission but also situations in which controlled technology is hosted on a server in Singapore and accessed remotely from abroad.Footnote 56
The formulation therefore encompasses both push-based transfers (where technology is actively sent out of Singapore) and pull-based access (where the recipient retrieves or accesses it from outside Singapore), but in both cases the regulatory trigger remains anchored in Singapore’s territory. The decisive factor is the location of the technological release, not the identity of the person who receives it. The SGCA is indifferent to the nationality of a person who accesses controlled technology within Singapore’s borders. What matters is whether the technology crosses an outbound boundary, not who encounters it domestically.
The EAR’s counterpart operates on the opposite premise. It distinguishes between two forms of constructive transfer based on where the release occurs. Where controlled technology is released to a foreign national within the US, the release is a ‘deemed export’ to that person’s most recent country of citizenship or permanent residency.Footnote 57 Where the release occurs outside the US, it is a ‘deemed reexport’ to the foreign person’s most recent country of citizenship or permanent residency.Footnote 58 In both cases, ‘release’ is defined broadly to include visual or other inspection, by a foreign person, of items that reveal controlled technology to that person, and oral or written exchanges of technology with a foreign person.Footnote 59 The major jurisdictional trigger is the identity of the recipient, not the geographic location of the release.
The distinction matters most acutely in the Singapore-based scenario. For instance, a Chinese national, employed as an engineer at a semiconductor fabrication facility in Arizona who is granted access to EAR-controlled technology, has received a deemed export to China.Footnote 60 However, for Singapore, the more important case is the same engineer employed at an equivalent facility in Singapore. If the technology is subject to the EAR, US law treats that domestic release as a deemed reexport to China, because the EAR layers a deemed-release rule on top of its physical-movement trigger, keyed to the recipient’s most recent country of citizenship or permanent residency rather than to any cross-border movement of the technology itself. Whether a licence is required then turns on classification, the destination’s status under the Country Chart, and any applicable exception. The SGCA, by contrast, does not treat domestic access by a foreign national as an intangible transfer merely because of the recipient’s nationality. Unless the (controlled) technology is transmitted from Singapore to a person in a foreign country, the ITT provision is not engaged. The result is a doctrinal mismatch: what Singapore law treats as domestic access within its territory, US law treats as a controlled reexport by legal fiction.
The practical consequence is that the SGCA’s intangible modality leaves unregulated an entire category of technology access captured by the EAR: the release of controlled knowledge to foreign nationals present within the regulating State’s own territory. For a hub economy that hosts major multinational semiconductor operations staffed by internationally mobile engineers, this regulatory mismatch could be significant. The fabrication, design, and R&D facilities that Singapore has attracted precisely due to its integration into the global semiconductor ecosystem are staffed by workforces drawn from across the region and beyond. The deemed-export mismatch is embedded in the ordinary operating conditions of those facilities. Section 3.2 returns to these implications.
2.2.3 Foreign-to-Foreign Controls: The Brokering Regime
The SGCA reaches foreign-to-foreign movement through the so-called ‘brokering’ regime under Section 6, a feature that distinguishes it from many comparable export control regimes.Footnote 61 Section 6(1) establishes a bifurcated structure. Subsection (1)(a) applies to ‘goods’, prohibiting the arrangement or negotiation of contracts for their acquisition or disposal where the person knows or has reason to believe that the transaction will result in their removal from one foreign country to another.Footnote 62 Subsection (1)(b) extends the same prohibition to ‘technology’ and to documents embodying such technology, but in broader terms, covering not only removal but also transmission. Technology is thus treated more expansively than goods, reflecting its capacity for intangible transfer.Footnote 63
The scope of this control is further defined by two parallel provisions: the first applies to categories of strategic goods and strategic goods technology specified by the Minister, reflecting a list-based approach.Footnote 64 The second extends control to any goods or technology linked to ‘relevant activity’, including situations where the person has been notified, has actual knowledge, or has reasonable grounds to suspect WMD-related end-use.Footnote 65 This introduces a ‘catch-all’ mechanism expanding the regime beyond the control list, albeit only through a knowledge-based trigger.
The authorisation structure, however, is differentiated. Persons intending to broker specified strategic goods must be registered with the Director-General of Customs, and registration may be subject to conditions and ongoing oversight.Footnote 66 By contrast, brokering involving all technology, whether listed or catch-all, as well as catch-all goods and documents embodying technology, is routed through the permit regime.Footnote 67 In effect, the SGCA establishes a substantive brokering prohibition, while separately institutionalising control over a defined class of brokers dealing in listed goods. This shifts the focus of regulation from transactions to intermediaries, but only selectively: technology-related and risk-contingent brokering remains transaction-specific and permit-based.
However, Section 6’s reach is circumscribed by a set of statutory exemptions, including strategic goods technology in the public domain, transactions authorised by permit, and persons whose involvement is limited to ancillary services, including transportation, financing, insurance, and certain forms of advertising.Footnote 68 Section 6’s jurisdiction is itself territorial in practice: it captures brokering activities conducted from Singapore, but does not extend to activities conducted entirely outside Singapore by persons with no Singapore nexus.
The EAR does not contain a standalone brokering regime comparable to SGCA Section 6. Instead, it captures intermediary conduct through a bundle of overlapping doctrines. The broadest is the reexport control: because US jurisdiction attaches to the item itself, the EAR can regulate any transfer of a controlled item between two foreign countries as a ‘reexport’, regardless of who arranges the movement or where that person is located. The intermediary’s identity and location do not determine whether the item is subject to the EAR.Footnote 69 That question is resolved by the item’s origin, composition, or production history.
Participation liability reinforces this coverage: the EAR extends liability to any person who causes, aids, or abets a violation, or who participates in a prohibited transaction with knowledge or reason to know of the violation.Footnote 70 A broker who arranges a transaction involving an Entity List party or a prohibited end-use can, therefore, be liable even without physically handling the goods. This is the closest functional analogue to the SGCA’s brokering control, but the resemblance is structural rather than doctrinal: EAR participation liability is derivative, attaching to an underlying export, reexport, transfer, or prohibited end-use violation, while SGCA Section 6 creates a freestanding offence in which brokering activity itself is the regulated conduct. In practice, the EAR’s participation liability, combined with the Office of Foreign Assets Control (OFAC)’s prohibition on facilitation of sanctioned transactions, may approximate a brokering regime in functional terms, but assembles it from general prohibitions rather than constructing it as a distinct regulatory category.Footnote 71
The two regimes address foreign-to-foreign transfers through inverse mechanisms. The SGCA regulates brokering directly through ex ante authorisation of the intermediary; the EAR reaches broker-like conduct indirectly, because US jurisdiction follows the item and may attach to intermediaries through participation or knowing-involvement liability. The SGCA’s hook is territorial; the EAR’s is item-based. Where both are present, the regimes converge: a Singapore-based broker arranging a foreign-to-foreign sale of a controlled multibeam echosounder system may be caught by the SGCA if the item is listed on the SGCL, and by the EAR if it falls under a corresponding ECCN (e.g. 6A001). Where only one hook is present, they diverge: absent a Singapore-based broker, the SGCA has no role, while the EAR may still apply because its jurisdiction runs with the item.
2.3 Controlled Items: List-Based Convergence and Unilateral Departure
The SGCA regulates items listed in the Strategic Goods Control List (SGCL), promulgated under the SGCO and periodically updated, most recently, through the 2025 Order.Footnote 72 The SGCL is structured in two parts covering military goods and dual-use goods, adopts a numbering structure derived from the multilateral export control regimes, which is also reflected in the EU dual-use regulation, and draws its substantive content from the four multilateral export control regimes noted earlier.Footnote 73 Singapore is not a formal member of any of these regimes, yet the SGCL mirrors their control lists in substance.
Importantly, the SGCL adopts a classification structure analogous to the US Commerce Control List (CCL) administered through the EAR. The CCL classifies controlled items by Export Control Classification Number (ECCN) across ten categories and five product groups corresponding to the nature of the item, ranging from hardware and materials to software and technology. Singapore’s SGCL uses a closely parallel coding structure, often adding the prefix ‘DL’ to dual-use list entries, so that US classifications such as ECCN 5A002 can be cross-referenced against Singaporean entries such as DL5A002.Footnote 74 On this score, the SGCL and the CCL are cross-readable, reflecting their shared derivation from the multilateral regime lists that form the common language of global strategic trade.
The SGCA supplements the SGCL with a WMD catch-all control. Where an item is not listed but a person knows, has reasonable grounds to suspect, or has been informed by an authorised officer that it is intended or likely to be used for the development, production, or delivery of weapons of mass destruction or missiles capable of delivering such weapons, a permit is required.Footnote 75 The catch-all extends the SGCA’s reach beyond the fixed list to items whose control is triggered by knowledge of end-use. However, it is primarily calibrated to WMD concerns. It operates as an end-use control, but its scope is functionally confined to WMD non-proliferation concerns and does not extend to national security or industrial-policy considerations of the kind that animate the EAR’s end-use controls.Footnote 76
Even at the catch-all level, then, a structural divergence persists: both regimes supplement multilateral control lists with end-use controls, but the EAR adds a distinct person-based overlay in Section 744.6, regulating specified support activities of US persons worldwide, including for items not otherwise subject to the EAR in specified end-use and end-user contexts.Footnote 77 The SGCA has no equivalent. At this baseline level, the two regimes reach broadly similar outcomes in their treatment of traditional dual-use and WMD-related concerns, though through structurally different regulatory architectures.
The divergence becomes even sharper where the EAR departs from multilateral convergence. BIS’s October 2022 interim final rule created new ECCNs 3A090 and 4A090 for specified high-performance integrated circuits and computers containing them.Footnote 78 These controls did not simply mirror preexisting multilateral semiconductor entries. Rather, they relied on performance-based thresholds tied to processing capability and interconnect speed, aimed at addressing immediate national security and foreign policy concerns arising from China’s use of advanced computing, supercomputing, and semiconductor manufacturing capabilities for military modernisation.Footnote 79 The threshold-based design created room for firms to redesign products for the Chinese market. Nvidia’s A800 and H800 were reduced-capability variants designed to fall below the 2022 thresholds, before later US updates moved to close that workaround.Footnote 80
In October 2023, BIS tightened ECCN 3A090 by replacing the earlier performance test with the ‘total processing performance’ metric and adding performance-density criteria to capture chips that had fallen below the 2022 threshold yet remained capable of advanced AI training.Footnote 81 The December 2024 updates deepened this unilateral turn still further.Footnote 82 They expanded country-wide chip controls to cover advanced HBM, broadened restrictions on semiconductor manufacturing equipment (SME) and significantly expanded the FDPR’s scope.Footnote 83
These measures are not derived from any multilateral regime. They are unilateral technology-denial controls designed to restrict China’s access to frontier computing capabilities. As of this writing, the SGCL contains no equivalent category; nor has Singapore internalised those US parameters into domestic law. That non-incorporation is, on this reading, a deliberate choice. Singapore’s April 2025 joint advisory on advanced semiconductors and AI acknowledges the practical significance of foreign unilateral controls, urging firms to comply with applicable foreign controls, but it reiterates that its own regime remains anchored to the major multilateral export control regimes and UNSC sanctions.Footnote 84
The SGCL does not have a ‘gap’ compared to the EAR; the two regimes perform different functions and rest on different rationales. The SGCL remains a multilateral list-based instrument: it identifies goods and technology that Singapore has chosen to control in line with non-proliferation consensus.Footnote 85 The EAR, by contrast, has evolved since October 2022 into a dynamic instrument of US technology strategy, layering unilateral controls onto the multilateral baseline in response to perceived strategic threats and technological change.
2.4 Controlled Entities and Destinations: Sanctions Implementation against Strategic Denial
The divergence between the two regimes is also visible in their treatment of actors and jurisdictions. The SGCA does not generally operate via a standing list of prohibited or presumptively denied entities comparable to the US Entity List. Its logic remains primarily permit-based and item-based: regulatory burden is triggered when a controlled item moves, rather than by the identity of the counterparty alone.Footnote 86 The SGCA’s catch-all controls are, as discussed, activated by knowledge or notice of a prohibited WMD-related end use;Footnote 87 thus, counterparty identity matters mainly as evidence of diversion risk towards a WMD-related end use, not as an independent ground for restriction, as it is under the US Entity List.
Singapore’s treatment of destinations is similarly distinct from the US model of strategic denial. Destination-specific trade prohibitions do exist, but they are imposed principally via the RIEA/RIER, which Singapore Customs describes as the instruments for implementing UNSC trade-related sanctions and related trade prohibitions.Footnote 88 The clearest example is the DPRK: Singapore prohibited all commercially traded goods to or from North Korea, including transhipment and transit, through the Seventh Schedule to the RIER.Footnote 89 Singapore’s destination controls are framed primarily as sanctions implementation and trade prohibition, not as a freestanding architecture for unilateral technological denial against a strategic rival.
The case of Russia marks a rare example in which Singapore imposed targeted export controls without a direct UN mandate, yet even these measures remained tied to a defined multilateral baseline. In March 2022, Singapore prohibited exports, transhipment, or transit to Russia of all items on the Military Goods List and of the Category 3, 4, and 5 portions of the Dual-Use Goods List.Footnote 90 The Singaporean Government stated that the purpose was to constrain Russia’s capacity to conduct war against Ukraine.Footnote 91 The form of the measure, however, remained bounded by preexisting list architecture. Singapore mapped the Russia prohibition onto the preexisting list categories under the SGCO,Footnote 92 which Singapore Customs explains is aligned with the multilateral export control regimes. Even where Singapore acts unilaterally, it still does so through multilateral categories.
2.5 Different Mandates, Different Architectures
The divergence documented in Sections 2.2 through 2.4, across controlled modalities, items, and entities alike, is not explained by regulatory lag, administrative incapacity, or inattention to diversion risks. It reflects two regimes designed for fundamentally different purposes, operating from different positions in the semiconductor supply chain.
The US sits at the apex of the supply chain’s architecture. On one account, the US-based companies maintain approximately 68% of global value-added shares in electronic design automation (EDA) and core intellectual property, 65% in logic chip design, and 47% in semiconductor manufacturing equipment.Footnote 93 Because virtually every advanced chip design passes through US-origin EDA tools and semiconductor manufacturing relies on a highly concentrated equipment and materials ecosystem dominated by American, Dutch, and Japanese firms,Footnote 94 these upstream nodes constitute what Farrell and Newman described as the central nodes of a networked system.Footnote 95 Such an asymmetric networked structure enables the State exercising jurisdiction over those hubs to project regulatory authority over downstream products through chokepoint effects, which allows a privileged State to limit or penalise the use of these nodes by third parties. The FDPR serves as a powerful legal instrument converting this industrial centrality into extraterritorial regulatory reach.
Singapore is a crucial node, but not a chokepoint in this sense. The city-state accounts for around 10% of global chip output and roughly 20% of global semiconductor manufacturing equipment production and is home to many semiconductor companies spanning IC design, wafer fabrication, packaging, and testing.Footnote 96 Notwithstanding, Singaporean firms do not produce the foundational EDA tools or core IP that enable the EAR’s extraterritorial reach. Singapore’s role lies downstream, in fabrication, packaging, and redistribution, where upstream technologies developed elsewhere are deployed, integrated, and routed onward.Footnote 97 It is a crucial node in the semiconductor supply chain, but not a true chokepoint.
The regulatory designs reflect these positions. The EAR has been progressively repurposed since October 2022 into a tool of technology competition, expanding both the substantive scope of controlled items and the extraterritorial reach of the FDPR with each successive rulemaking. The SGCA was enacted in 2002 as a non-proliferation instrument anchored to multilateral consensus and calibrated to Singapore’s hub function. Its logic is multilateral, not unilateral: controlled items track the four multilateral export control regimes; destination-based prohibitions are principally confined to UNSC sanctions, the Russia measures being the bounded exception; and there is no domestic mechanism through which another State’s technology-competition parameters can be automatically absorbed into Singapore law.
Singapore’s export control regime is territorial by design. Its controlled modalities all require a Singaporean nexus: goods crossing Singapore’s borders, technology transmitted from or made accessible through Singapore-based infrastructure, or brokering conducted from Singapore. SGCA Section 6 is the regime’s main hub-specific extension. It reaches certain foreign-to-foreign transactions that would otherwise fall outside Singapore’s territorial frame, but only where a Singapore-based intermediary arranges a transaction involving items already subject to Singapore’s strategic-goods controls. Brokering expands the nexus, not the substantive control universe. The SGCA, therefore, regulates Singapore-mediated transactions, not the global afterlife of controlled items. Singapore lacks the industrial leverage that would let its law follow the item abroad; whether it internalises another State’s parameters is another matter, one that fits the purpose and its national interests. The regulatory design (or redesign) is essentially a political call, one that Singapore has so far declined to take. The next question, then, is what tools its existing laws nevertheless supply to answer US diversion concerns and the alignment pressure behind them. Section 3 takes up that question.
3. The Challenge Facing a Hub Economy
Two commitments pull against one another at this point. Singapore must respond credibly to US diversion concerns, on which its continued access to advanced computing substantially depends, while preserving the coherence of a regime built on multilateral categories and the standing that coherence confers on an open hub. In theory, nothing prevents Singapore from expanding export controls: its Russia measures show that it can impose autonomous controls when it chooses. The cost is another matter, and it is not conjectural: the neutrality it would forfeit is the declared position of Singapore’s foreign policy. The Foreign Minister has stated in Parliament that aligning with a single dominant power would render Singapore ‘a vassal state’, and that Singapore will work with both the US and China where interests coincide while ‘courteously stand[ing] up’ when they diverge from Singapore’s own interests.Footnote 98 The Prime Minister has framed this position as Singapore’s and ASEAN’s alike: declining to choose sides between the major powers, keeping the region ‘open and inclusive’, and remaining friends with China, the US, and Japan.Footnote 99
Incorporating unilateral advanced-computing parameters would reorient the SGCA from non-proliferation toward technology denial, embed great-power rivalry in Singapore law, and surrender control over the categories its own law applies. It would also cut against the value proposition on which Singapore’s standing as a hub rests: a ‘trusted global hub’ for trade, finance, and advanced manufacturing, whose value a fractured world accentuates precisely because goods, data, and capital flow through it without a geopolitical test at the border.
The response has instead taken the form of selective accommodation. Where Section 2 traced the doctrinal independence through which Singapore retains control over the classifications recognised by its law, Section 3 turns to how it addresses conduct falling outside those classifications. This article uses ‘doctrinal substitution’ to describe the use of existing domestic legal categories to reach independently wrongful conduct associated with diversion, without incorporating the US regime’s classifications or export-control logic. Doctrinal independence and substitution are thus mutually reinforcing dimensions: the former preserves Singapore’s control over its legal categories, while the latter enables it to act within that boundary. The repeated official characterisation of such enforcement as independently undertaken under domestic law, as discussed below, gives that relationship practical expression: substitution addresses diversion-related conduct while reaffirming that Singapore is enforcing its own law rather than reacting to the US pressure. Section 3.1 shows this relationship in operation through Singapore’s recent enforcement practice. Section 3.2 tests the limits of substitution as the regulatory object shifts from physical goods to intangible knowledge. Section 3.3 then asks how far the strategy can remain viable amid Singapore’s semiconductor upgrading and the expanding reach of US controls, including pending legislation.
3.1 The ‘Depoliticised’ Substitution: Fraud Law as Export Control Surrogate
The structural mismatches mapped in Section 2 have a direct enforcement corollary. Where an item is controlled under the EAR but not listed in the SGCL, the SGCA cannot itself supply the enforcement vehicle for diversion through Singapore. Since 2022, BIS has created and repeatedly tightened advanced-computing controls, including ECCNs 3A090 and 4A090, to capture high-performance GPUs, advanced computing systems, HBM, and related hardware. Singapore’s SGCL, by contrast, remains tied to the multilateral dual-use architecture and does not incorporate an equivalent post-October 2022 advanced-computing category. An AI chip that triggers a US licence requirement may, thus, move through Singapore without engaging the SGCA merely by virtue of that US control. The result is a design mismatch: unilateral EAR controls over advanced-computing items do not automatically transform those items into controlled items under the SGCA, unless they independently satisfy an SGCL entry or the WMD-related catch-all.
This mismatch surfaced in early 2025, when Singapore charged three individuals over the alleged fraudulent procurement and onward transfer of servers from Singapore to Malaysia. Two of the accused, subsequently identified as officers of Aperia International Pte Ltd, allegedly represented to Dell Global B.V. (Singapore) that the items would not be transferred beyond the authorised consignee, while concealing their intended onward transfer of the servers to Malaysia.Footnote 100 By March 2025, prosecutors had further charged two of the accused with an additional count each of conspiracy to commit fraud against Super Micro Computer, Inc., on materially identical false representation grounds. Singapore framed the matter as an independent domestic fraud investigation rather than a confirmed US export-control circumvention case. Minister Shanmugam rejected speculation that the case was linked to US export controls for advanced Nvidia chips, stating that ‘[t]his case is unrelated’ and that it was an ‘independent investigation conducted by Singapore’.Footnote 101 The charges were brought under Section 120B read with Section 424B of the Penal Code 1871, framing the conduct as criminal conspiracy to commit fraud by false representation, rather than as a violation of the SGCA.Footnote 102
The significance extends beyond the individual prosecution. The fraud framing captures the deception in the procurement chain, where the false end-user representation was made to the Singapore supplier, but not the diversion of controlled technology as such. Singapore’s official account underscores this distinction: the State was enforcing domestic fraud law within an international supply chain, not directly implementing US export-control classifications. The legal hook lay in the alleged misuse of a Singapore entity and inaccurate destination representations, not in the automatic domestic incorporation of US semiconductor controls.
In practice, diversion through Singapore is unlikely to occur without deception: a truthful end-user declaration would create the evidentiary trail that exposes the transaction to US enforcement, where the total penalty package, including denial of export privileges that can exclude a person or entity from the US technology ecosystem for the duration of the denial order, is substantially more severe than anything Singapore’s fraud-based prosecution carries.Footnote 103 The wrongdoer who employs false representations is, in effect, engaging in jurisdictional arbitrage: by misrepresenting end-use to a Singapore entity, the transaction is channelled into Singapore’s domestic fraud regime, a regime that carries no equivalent exclusionary sanction. Conversely, exposure under US law carries the risk of effective expulsion from US-origin technology supply chains and even financial networks, making diversion via deception a comparatively rational strategy.
The contrast with SGCL-listed items is instructive. In September 2023, Hydronav Services (Singapore) Pte Ltd was fined over S$1.1 million under SGCA Section 5 for exporting to Myanmar a multibeam echosounder system procured from Norway, without the requisite permit.Footnote 104 Singapore Customs found the system included sub-systems controlled under the SGCO 2017. However, because the items were listed, the regulatory breach of the SGCA could be prosecuted directly. Two Hydronav employees were also fined under the Penal Code for conspiracy to cheat under Section 417 read with Section 109, based on false end-user representations made to the Norwegian supplier.Footnote 105 The point is not that fraud displaces export-control enforcement. Where the item falls within the SGCL, Singapore can proceed under the SGCA and, where deception is involved, the Penal Code as well. Where the item falls outside the SGCL, the SGCA route is unavailable, leaving general criminal law to do the enforcement work.
The framing has a depoliticising effect. By prosecuting diversion as fraud rather than as an export-control violation, Singapore can address the conduct without importing US unilateral export-control parameters into domestic law and without framing the case as part of the US–China technology rivalry. The Government’s public posture is consistent with this account. When asked whether Singapore would tighten laws on high-tech hardware movements, the Minister stated that ‘the issue is not with the law’ but ‘with enforcement’, while leaving open future legal change if facts warrant it.Footnote 106
Domestic criminal law thus carries the enforcement work while the export control regime remains formally unengaged. The substitution is not equivalent in kind. Fraud liability operates ex post and transaction by transaction; it supplies none of the prospective instruments an export-control regime provides, namely licensing, end-user screening in advance of shipment, conditions restricting access or onward transfer, and denial orders excluding a party from future trade. Whether that domestic prosecution adequately addresses the underlying strategic concern, as opposed to merely performing regulatory responsiveness, is a question the substitution strategy leaves formally open: the Penal Code penalises the deception but does not exclude the wrongdoer from re-entering the semiconductor trade, and it may not generate enforcement data that feeds into US export-control compliance systems. The substitution lets Singapore answer, at least in part, Washington’s diversion concerns without signalling alignment to Beijing. But that balance is contingent on how often Singapore can credibly characterise each case as independent. If the US regulatory perimeter expands and diversion via Singapore becomes more visible, the depoliticising frame may become harder to sustain.
3.2 Where Substitution Falters: Knowledge, Trade Secrets, and Regulatory Reach
The substitution strategy reaches its sharpest limit when the regulatory object shifts from goods to knowledge. Knowledge travels in the heads of engineers, in electronic transmissions, and in access permissions granted to employees of multinational facilities. Where controlled technology is transmitted out of Singapore, the SGCA’s intangible transfer provisions reach the conduct directly. The harder problem, however, lies in what those provisions do not reach: the release of controlled technology to a foreign national within Singapore. The EAR’s deemed-export rule treats that domestic release as a controlled reexport keyed to the recipient’s nationality; the SGCA, anchored to outbound transmission, is indifferent to who accesses controlled technology within Singapore. For a hub economy hosting internationally mobile R&D workforces, this is the harder regulatory problem.
The Computer Misuse Act (CMA) might appear to fill this gap, since the conduct in question often involves accessing technology held on Singapore-based computer systems. The CMA criminalises unauthorised access to computer material, access with intent to commit a further offence involving property, dishonesty, or causing harm; unauthorised modification of computer material, and unauthorised use or interception of computer services.Footnote 107 Where the offence involves a ‘protected computer’, that is one used in connection with Singapore’s security, defence, international relations, or essential services, such as banking, public utilities, and emergency services, substantially enhanced penalties apply: a fine of up to S$100,000 or 20 years’ imprisonment, compared with a standard maximum of two years for ordinary unauthorised access.
However, these enhanced penalties are attached only when the access itself was unauthorised; where a person accesses the system lawfully and subsequently misuses that access, the conduct falls, if at all, under Section 4, which carries a separate maximum of 10 years and is not subject to Section 11’s enhancement.Footnote 108 The Act also has extraterritorial reach where the accused was in Singapore at the material time; the computer (or data) was in Singapore at the material time; or the offence caused or created a significant risk of ‘serious harm in Singapore’, defined to include damage to national security, defence, or foreign relations, and disruption of essential services, such as banking, finance, and public utilities.Footnote 109
An individual who accesses proprietary chip design files on a Singapore-based server without authorisation and transmits them abroad may, at first glance, fall within the CMA, particularly where the system qualifies as a ‘protected computer’ and enhanced penalties apply.Footnote 110 The difficulty, however, is one of doctrinal fit. The CMA is fundamentally a cybercrime statute, structured around conduct that is wrongful because it circumvents access controls imposed by the system owner.
This issue becomes problematic where access is initially lawful. A person who accesses chip design data within the scope of employment and subsequently transmits the data to an unauthorised foreign recipient will not ordinarily engage in ‘unauthorised access’ under Section 3, which requires access ‘without authority’. While Section 4 extends liability to access data undertaken with intent to commit a further offence (including where the access itself is authorised), the provision is directed at the purpose of the access rather than the subsequent use of what is accessed, and may not readily capture cases where the decision to disclose arises after the data has been retrieved.
Conversely, the US EAR treats the release in Singapore of technology subject to the EAR to a foreign person of another country as a deemed reexport.Footnote 111 Where diversion is effected through fraudulent misrepresentation in computer-mediated transactions, the CMA may provide a viable, if indirect, prosecutorial hook. However, where the concern is the transfer of controlled technology to foreign nationals who are otherwise authorised users of the system, the CMA may have its limitations for such downstream disclosure. The CMA targets wrongful access; it does not systematically regulate the misuse of lawfully accessed technology.
The mismatch is compounded on the civil side by the absence of a standalone trade secrets statute. Unlike the US, which supplemented the criminal provisions of the Economic Espionage Act of 1996 with a Federal civil cause of action under the Defend Trade Secrets Act of 2016,Footnote 112 Singapore relies on the common law action for breach of confidence, supplemented by contractual non-disclosure agreements.Footnote 113 The limitations for semiconductor governance are twofold. The action is private, with no State-driven counterpart to EEA prosecution.Footnote 114 More crucially, it protects proprietary rights, not regulatory interests. A firm that voluntarily shares its semiconductor process technology with a foreign partner under a licensing agreement has not committed a breach of confidence, even if the transfer would require a deemed export licence. The regulatory wrong is not coextensive with the civil wrong.
Some jurisdictions have responded by elevating semiconductor technology protection to the level of national security. Taiwan’s 2022 amendments to its National Security Act introduced the concept of ‘national core key technologies’, establishing a new economic espionage offence targeting the misappropriation of designated technologies for use in foreign regions, including Mainland China, with enhanced criminal penalties and outbound investment review.Footnote 115 The EEA distinguishes between ordinary trade secret theft under 18 USC Section 1832 and economic espionage benefiting a foreign government under Section 1831, the latter carrying penalties of up to 15 years’ imprisonment.Footnote 116 In both cases, the State claims a direct interest in the protection of semiconductor-related knowledge, independent of the private interests of the technology holder.
For Singapore, this path is strategically costly. First, Singapore would need to occupy a supply chain position where the technology generated or hosted within its territory sustains a ‘national security’ characterisation. Its current role lies primarily in fabrication, packaging, and redistribution; but its upgrading into higher-value segments, examined below, may progressively alter that calculus. More delicately, any statutory framework designating semiconductor technology as a national security matter necessarily implies a ‘threat’ assessment. A law that names specific countries (as Taiwan’s does, with its express reference to ‘Mainland China’) would align Singapore with the US technology-denial architecture in ways the depoliticisation strategy is designed to avoid. A law that does not identify particular countries may lack the operational specificity required for effective enforcement. The choice cannot be made on legislative technique alone. It is a question of how visibly Singapore is willing to take a side, and the trajectory of its semiconductor upgrading will increasingly force the question to the surface.
3.3 The Substitution Strategy under Strain: US Legislative Expansion
The substitution strategy described in Sections 3.1 and 3.2 operates against a moving target. Two concurrent developments threaten to compress the room within which substitution remains viable: Singapore’s own industrial upgrading, which deepens its integration into the US-led semiconductor ecosystem, and US legislative developments that would expand Washington’s capacity to monitor the downstream use of its technology abroad.
Singapore’s current semiconductor upgrading trajectory is the latest phase of a role it has played for more than half a century. From 1968, Singapore began attracting US firms such as Texas Instruments and National Semiconductor to establish chip assembly operations,Footnote 117 then systematically pushed the industry up the value chain via priority industry listings from 1975 onward, and a deliberate high-wage policy in the early 1980s to force out firms unwilling to invest in skill and technology transfer.Footnote 118 The strategy’s most significant product was Chartered Semiconductor Manufacturing,Footnote 119 once the world’s third-largest pure-play silicon foundry, alongside TSMC and UMC, before its late-2009 acquisition by GlobalFoundries.Footnote 120
The loss of Singapore’s only indigenous foundry did not halt the city-state’s upgrading trajectory; it somewhat altered its character. The investment wave that followed has been driven primarily by multinational corporations headquartered in the US or US-allied jurisdictions, deepening Singapore’s integration into the semiconductor supply chain. Between 2024 and early 2025 alone, Singapore attracted announced semiconductor investments exceeding US$18 billion across advanced wafer fabrication, high-bandwidth memory packaging, semiconductor equipment, and chip design R&D.Footnote 121 The investing firms read as a roster of the US-aligned ecosystem: Micron and Applied Materials (US), TSMC-associated Vanguard, UMC, and MediaTek (Taiwan), NXP (Netherlands), Advanced Substrate Technologies (AST), a TOPPAN Holdings subsidiary (Japan), and Siltronic (Germany), reinforcing an existing base anchored by GlobalFoundries (US), Infineon (Germany), and STMicroelectronics (France/Italy).Footnote 122
These investments carry a normative corollary. The firms making them are headquartered across multiple chokepoint jurisdictions, each operating its own export control regime. The EAR, including the FDPR, functions as the global anchor, but it is not the only layer: Taiwan’s tightening export controls and outbound investment screening and the Netherlands’ advanced semiconductor equipment restrictions, for instance, each apply to their respective firms’ Singapore-based operations insofar as controlled items are involved.Footnote 123 As Singapore hosts more of these operations, it becomes subject not to one regulatory architecture but to several, operating concurrently and extraterritorially. The compliance burden falls on the firms; Singapore’s regulatory framework, for the time being, remains formally unchanged. Yet the aggregate effect is progressive entanglement via private ordering. As companies operating within Singapore’s semiconductor ecosystem internalise the export control obligations of the chokepoint jurisdictions where they originate, Singapore’s industrial base becomes bound by a multilayered regulatory architecture it did not design and over which it exercises no legislative control.
In this increasingly entangled environment, US law remains the global anchor shaping the semiconductor supply chain, and that anchor is being actively reinforced. Several pending bills would expand Washington’s capacity to monitor and control the use of US-origin technology abroad. The Remote Access Security Act (RASA), which has passed the House, would amend the ECRA to bring ‘remote access’ to US-controlled items within the export control regime.Footnote 124 It defines ‘remote access’ as access ‘through a network connection, including the internet or a cloud computing service’,Footnote 125 by a foreign person to an item subject to US jurisdiction from a location other than where the item is physically located, where the Secretary of Commerce determines that such use could pose a ‘serious national security or foreign policy risk’.Footnote 126
Crucially, the bill would authorise the President to control and license ‘the remote access of items subject to the jurisdiction of the United States by a foreign person’,Footnote 127 extending existing powers over exports, reexports, and in-country transfers to remote computational use itself. The bill seeks to close the perceived ‘cloud loophole’ by treating access to computing power, rather than only the physical movement of chips, as a potential export-control event.Footnote 128 For Singapore’s growing data-centre industry, this could bring Singapore-hosted cloud compute within US export controls even where the chips never cross Singapore’s border. Cloud providers may, therefore, face stricter customer screening, end-use checks, contractual safeguards, and access monitoring, especially for China-linked users. Ordinary cloud service provision could become an export-control compliance function.
The proposed Chip Security Act (CSA) would complement RASA by shifting control from access to physical traceability. It would require covered advanced chips to include security mechanisms, including location-verification tools, capable of detecting diversion, theft, tampering, or unauthorised relocation.Footnote 129 Licensees would also have to report evidence that a controlled chip had moved from its authorised destination or end user. Export control would, therefore, no longer end at customs clearance, but follow the chip through its operational life. In March 2026, the House Foreign Affairs Committee advanced the CSA in full committee markup, presenting it as a tool against illicit smuggling of advanced AI chips to China and other foreign adversaries.Footnote 130 For Singapore, this would intensify pressure to prevent chip diversion through its territory, while sitting uneasily with the SGCA's list-based model: where US-controlled items fall outside the SGCO and do not engage the SGCA's WMD-related catch-all, their US-controlled status alone supplies no legal basis for action under the SGCA.
BIS has already extended Entity List and Military End User List restrictions to unlisted foreign affiliates owned 50 per cent or more by listed entities through its September 2025 Affiliates Rule, though enforcement was suspended for one year from November 2025. The proposed STOP Shells Act would codify this standard in statute and additionally require BIS, before listing an entity, to assess whether applying the FDPR would advance US national-security or foreign-policy interests, embedding an FDPR trigger directly into the listing process.Footnote 131 This reflects a broader US shift from regulating named entities to policing corporate affiliation and indirect access channels. For Singapore, the cumulative effect would be to make China-linked investment through regional headquarters and intermediary entities harder to insulate from US controls: a Singapore corporate presence may no longer dilute those controls, but instead trigger scrutiny as a potential evasion channel.
Taken together,Footnote 132 these proposals could risk eroding Singapore’s depoliticisation strategy identified above: treating diversion cases as domestic fraud or cybercrime rather than export-control violations. As US controls increasingly target movement, access, and corporate structure, Singapore will face growing pressure not merely to prosecute deception after the fact, but to prevent diversion before it occurs.
4. Conclusion
The semiconductor export-control literature has largely focused on US–China rivalry and on the chokepoint economies that occupy the supply chain’s indispensable nodes. This article has made two related contributions from a comparative-law perspective. The first is to distinguish the ‘hub economy’ from the ‘chokepoint economy’ as a regulatory category. Unlike a chokepoint economy, a hub economy does not derive its strategic significance from a structurally irreplaceable position in a single technology or production node. Its significance lies instead in bringing together manufacturing and connective functions across the supply chain, from fabrication, packaging, and testing to logistics, transhipment, finance, corporate structuring, and onward distribution. The regulatory demands it attracts differ accordingly, directed less at what it produces than at what it enables. Singapore is paradigmatic of this position. Other economies, depending on their own value propositions within the supply chain, may face analogous challenges, though the argument advanced in this article is established only in relation to Singapore. Nevertheless, because Singapore’s regime reflects a multilateral, Wassenaar-derived baseline shared by many trading jurisdictions, its experience may illuminate how similarly anchored regimes absorb comparable pressure, though further comparative work is needed to test the argument’s broader applicability.
The second is the relationship between doctrinal independence and doctrinal substitution. Singapore’s regime remains territorially grounded and anchored to multilateral non-proliferation categories rather than to the unilateral technology-denial objectives of any single upstream State; doctrinal independence describes that retention of control over the classifications its own law recognises. The tension with US controls is not evidence of legal incapacity: Singapore could adopt more expansive controls, and its Russia measures show as much. What it declines is the cost, which would convert a non-proliferation regime into an instrument of another State’s technology denial, thus putting itself in the cross-fire of major power rivalry. Doctrinal substitution describes what occurs within that retained boundary: where the SGCA does not reach, the Penal Code and the Computer Misuse Act do some of the work against independently wrongful conduct associated with diversion. The Aperia prosecution illustrates the pattern: diversion is addressed as fraud, the export-control regime remains formally untouched, and Singapore avoids internalising US classifications. This is not regulatory failure. Functionally, the pattern arguably amounts to selective legal adaptation, answering Washington’s diversion concerns, at least in part, without signalling alignment to Beijing. But substitution remains partial: retrospective and transaction-bound, it supplies none of the prospective instruments of an export-control regime.
The space within which this balance operates may be narrowing. Singapore’s movement into higher-value semiconductor and computing activities deepens its integration into the US-led ecosystem and its exposure to the regulatory obligations multinational firms internalise. At the same time, the House-passed RASA and the proposed CSA and STOP Shells Act would, if enacted, extend US regulatory reach beyond physical shipments to remote computing access, chip location, and corporate affiliation. As US controls increasingly target movement, access, and ownership structure, Singapore’s room for treating diversion as ordinary domestic wrongdoing narrows, and the pressure shifts from punishing deception after the fact to preventing diversion before it occurs.
The Singapore case, thus, reveals a dilemma that other hub economies may share: how to remain inside the US-anchored semiconductor ecosystem without converting domestic law into an enforcement arm of unilateral US controls. Sovereign agency, for a hub economy, does not consist in escaping US extraterritoriality altogether; it lies in retaining control over the domestic legal categories through which external pressure is translated into enforceable obligations. Singapore’s response has been neither capitulation nor defiance: doctrinal independence marks that boundary; doctrinal substitution describes how Singapore has so far navigated within it.
Pax Silica adds an emerging institutional development to this picture. Singapore has, among the first cohort,Footnote 133 associated itself with a non-binding AI and technology-supply-chain initiative framed around ‘economic security’,Footnote 134 a vocabulary closer to the expanded mandate of the US regime than to the non-proliferation categories on which the SGCA rests. Whether, and through what legal mechanisms, it will shape Singapore’s export-control framework or wider technology-policy posture remains open.
Acknowledgements
The author thanks the anonymous reviewers for their constructive comments and Yitao Qian, Joia Ng, Wan Yi, and Claudia Toh for excellent research assistance. Earlier ideas were presented at an expert forum on global export controls organised by the Institute of Law for Science and Technology, National Tsing Hua University, on 12 June 2026, and at the Thailand Development Research Institute (TDRI) on 18 September 2026; thanks are due to Ching-Fu Lin, Gunn Jiravuttipong and colleagues at TDRI, and the participants at both events. This research is supported by the Ministry of Education, Singapore, under its Academic Research Fund Tier 1. Any opinions and errors are strictly my own and cannot be attributed to the funder or my institution.
Open access
