Image Credit
13 minutes
Summary
This article argues that the American de-risking approach has not eliminated dependence on China so much as rerouted and obscured it. While China’s direct share of US imports has fallen sharply, many supply chains now pass through connector economies such as Vietnam, Mexico, and India while continuing to rely on Chinese inputs, suppliers, machinery, and process know-how. We call this condition “triangulated interdependence”: a structure in which production or sourcing is geographically diversified across third countries while critical upstream capabilities remain concentrated in, or controlled by, a competing power.
Trade statistics tell only part of the story. They can lead Washington to overstate how far diversification has gone, whereas enforcement agencies struggle to trace embedded Chinese content and connector economies take on more of the compliance burden. Beijing also retains influence over how production moves through its control of technology, rare earths, and outbound industrial capacity. Washington should start with better measures of exposure. Bilateral trade data should be paired with sector-specific estimates of Chinese content and control. When firms seek tariff preferences, procurement contracts, or strategic subsidies, they should provide auditable supply-chain disclosures. Washington should also develop compatible transparency and resilience tools with its allies. Three questions deserve further study. How much of the decline in direct imports from China reappears as China-linked content in connector-economy exports? How much of the new capacity remains Chinese-owned or technologically controlled? And do measures of embedded exposure predict disruption better than bilateral trade shares? De-risking may continue, but a cleaner map is not necessarily safer terrain.
Introduction
In a factory in northern Vietnam, Foxconn assembles electronics worth billions of dollars for export to the United States and Europe. On paper, the goods are Vietnamese. In reality, most of their components still come from China, South Korea, and Taiwan. Local Vietnamese value-added amounts to roughly 7.8 percent.[1] A nearby BYD facility making iPads for Apple totals just 4.5 percent. Although the products have swapped passports, their production base has hardly moved an inch.
This is what much of de-risking now looks like behind the scenes. Washington might not be ending its dependence on China so much as making that dependence harder to see. Where controls carry real penalties, as with chipmaking equipment, dual-use systems, and sensitive data networks, the situation is clear-cut and firms cannot simply engineer around them. Countries have blocked transactions and barred technology transfers in an effort to draw security lines meant to hold. As a result, China’s direct share of US imports fell from about 21 percent in 2017 to 9 percent by mid-2025, near levels last seen around China’s entry into the World Trade Organization.[2] In corporate boardrooms, efficiency is no longer the only watchword. Companies that once treated China as the unavoidable center of global manufacturing now speak of resilience, diversification, and China-plus-one strategies.
That said, across much of the economy the headline numbers belie a more complex reality. Beijing’s global trade surplus hit record highs at $1.19 trillion even as direct trade with America fell.[3] In the first quarter of 2026 alone, Vietnamese imports from China topped $50 billion, up around 32 percent from a year earlier, with the sharpest rises recorded in electronics components and production machinery.[4] American officials have repeatedly accused Hanoi of serving as a transshipment point for relabeled Chinese goods that preserve exposure in a less visible form.[5] Chinese firms, in the meantime, have continued to pour capital into Southeast Asian plants. A factory leaving China does not necessarily mean its supply chain has escaped China. And a product shipped through a connector economy may look different in trade statistics while important upstream dependencies remain tied to Chinese suppliers. In such cases, interdependence is being redirected instead of severed, a pattern we call “triangulated interdependence.”
Where the Walls Hold
Separation and entanglement can coexist within the same supply chain. At the frontier of semiconductors, export controls and allied restrictions sever access to tools and chips that China cannot easily replace. Decoupling is genuine at these chokepoints. Such controls, however, govern a handful of irreplaceable nodes, leaving the massive supply base beneath them intact. Decoupling can hold at the frontier without holding at the foundation.
The picture is messier in autos and broader electronics. Final assembly often moves while Chinese inputs and supplier networks remain central. Electric-vehicle (EV) manufacturers expanding into Mexico’s northern production corridor still rely on drivetrain components and process technologies allied suppliers cannot yet match. Chinese suppliers and EV firms have meanwhile pursued footholds of their own in or around the same North American automotive clusters, most visibly through CATL’s battery-technology licensing arrangement with Ford for its battery plant in Michigan.[6]
Clean energy is the limit case. There is little yet to reroute because the underlying dependence remains direct. Industry estimates compiled in 2024 put Chinese-owned firms in control of the overwhelming majority of worldwide solar panel manufacturing capacity across polysilicon, wafers, cells, and panels, with more than 95 percent of solar wafers made in Chinese-owned factories.[7] The US Energy Information Administration reports that Chinese firms account for roughly a quarter of global lithium-mining capacity and over two-thirds of cobalt and lithium processing.[8] Allied producers cannot reproduce these positions any time soon.
The Rise of Triangulated Interdependence
Between genuine separation and persistent direct dependence lies triangulated interdependence. It is a structure in which production or sourcing is geographically diversified across third countries while critical upstream capabilities remain concentrated in, or controlled by, a competing power. Ordinary global value-chain fragmentation does not qualify. The resulting disconnect between the geography of trade and the geography of strategic dependence means that conventional trade measures can be a poor guide to actual exposure.
The process begins with choices made inside firms. When tariffs, sanctions risk, customer pressure, or political scrutiny raise the cost of direct exposure to China, companies have incentives to relocate the parts of production that are easiest to move. Final assembly may shift to Vietnam or Mexico, packaging and testing to another region, and alternative suppliers may be qualified elsewhere in Asia. For many industries, however, replicating China’s dense supplier ecosystems, specialized machinery, and accumulated production know-how is slower and more costly than moving final assembly. The result need not be cosmetic. New plants and suppliers can create genuine diversification. Even so, the customs identity of a product can change before the capabilities on which it depends do.
Connector economies do not all play the same role. Some flows amount to little more than “passport swapping” with minimal local transformation. In other cases, production genuinely relocates while specialized machinery, lower-tier inputs, supplier relationships, or process know-how remain tied to China. Elsewhere, local suppliers and productive capabilities develop alongside assembly. A 2025 IMF working paper draws a useful distinction between trade reallocation and trade rerouting. Vietnam shows evidence of the former, with rising domestic value-added in strategic-sector exports to America.[9] In aggregate, Vietnam, India, and Mexico saw an estimated $49 billion rise in US imports in 2025, nearly offsetting the $51 billion decline in direct shipments from China.[10] The countries Washington counts on to lower its reliance on China may simultaneously become channels through which China-connected manufacturing adapts to American pressure.
Building productive capability outside China leaves the question of dependence wide open. Geography is only part of the answer. Where new overseas capacity is owned or technologically controlled by Chinese firms, as with battery and component investment from Hungary to Indonesia, geographic diversification may overstate the degree of strategic separation. Friendshoring tracks where capacity is located, but it also needs to track who controls the capabilities embedded within it.
This distinction has consequences for how governments regulate trade. Rules of origin translate these complex production networks into legal statements about where a product is “from.” The Brookings analysis found that USMCA compliance on Mexican products entering the United States rose from 49.5 percent in December 2024 to 76.1 percent by July 2025, and Canadian compliance from 35.5 to 78.7 percent.[11] Firms adapt quickly when origin-based incentives change. A 2026 White House report describes US enforcement looking behind final-origin claims: It directs attention to factory capacity, input sourcing, local value added, and shipping patterns to distinguish genuine manufacturing from pass-through trade and origin shifting.[12] Importantly, the report acknowledges that shifts away from direct Chinese imports can reflect legitimate changes in production and sourcing rather than illegal transshipment.
Beijing is moving in a related direction. Its recent export controls reach Chinese-origin inputs and production technology in some foreign-made goods. The two systems differ sharply in purpose and institutional maturity. What they share is a growing recognition: where a product is finally made may reveal less than what it contains and which capabilities its production still rests on.
Tactical Success, Strategic Fallout
The problem with targeted de-risking runs deeper. US policy has focused on what it can readily observe—import reliance and bilateral trade concentration. Those measures show where direct exposure has fallen. They are less revealing about dependence that has moved into upstream processing and lower-tier suppliers.
That creates a risk of miscalculation. A government may conclude that a supply chain has been diversified, only to discover in a crisis that critical tooling, minerals, or process knowledge still run through China-linked networks. US Customs and Border Protection (CBP) requires importers to trace component origins through transaction records and supply-chain documentation because exposure often sits behind the final export platform. RAND researchers note how quickly lower-tier connections multiply and how limited firms’ visibility often is.[13] Even the Department of Defense still lacks meaningful visibility into the raw materials and parts suppliers beneath major systems.[14] Dependence can disappear from the bilateral data long before it disappears from the supply chain.
Enforcement is another source of strain. Under US origin rules, substantial transformation can change where a good legally counts as “from” even when large shares of its inputs come from elsewhere. Firms can adapt to that system faster than governments can map every input. CBP is already moving in that direction through the Enforce and Protect Act (EAPA). In July 2026, the agency reported more than $1 billion in additional duties owed, a first in the program’s 10-year history, and said investigators had inspected production facilities in Mexico, Thailand, India, New Zealand, and the United Kingdom to verify legitimate sourcing.[15] EAPA also covers misclassification and undervaluation, so the figure captures the program’s overall reach. CBP enforcement records illustrate that the largest stopped solar shipments came from Malaysia, Vietnam, and Thailand rather than China, and that most of their value was later released.[16] The focus is shifting to goods whose risk lies behind the final export label instead of in the label itself.
The politics are harder still. Partial decoupling pushes pressure onto the connector economies Washington needs most, while giving Beijing every incentive to manage how the rerouting unfolds. These economies do more than host factories: They also carry much of the compliance and enforcement burden. Vietnam committed last year to cooperate with the United States on duty evasion and export controls. Thailand has stood up a similar origin-certificate task force under tighter US scrutiny. China is not a bystander. Xi Jinping’s April 2025 tour of Vietnam, Malaysia, and Cambodia produced agreements spanning trade, technology, health, and education, deepening Beijing’s institutional presence in the region’s emerging manufacturing hubs. At home, Beijing is tightening control over its supply chains, from rare-earth export controls to the reported delay of BYD’s proposed Mexico plant over fears of technology leakage.[17] Chinese authorities now shape which capabilities can leave the country and which remain subject to domestic control, a second boundary alongside Washington’s own.
Tactical trade diversion opens strategic gaps—between data and reality, between enforcement demands and state capacity, and between Washington and the partners it needs most. The challenge is whether those gaps can be seen clearly enough to manage.
A Strategy of Opaque Interdependence
A serious strategy should replace metrics of movement with metrics of exposure. Falling direct imports from China answer only part of the question. The rest is how much China-linked content, tooling, and process knowledge remains embedded in strategic goods after they pass through countries such as Mexico and Vietnam. Headline import figures should therefore be paired with estimates of embedded dependence in critical sectors. Those estimates would track where goods are packed and how much Chinese input and know-how they still contain.
Market access and public support should be contingent on auditable disclosure of dependence. Washington already has precedents in this regard. The CHIPS and Science Act, for example, bars subsidy beneficiaries from materially expanding semiconductor production capacity in countries of concern for 10 years after receiving funding. The Inflation Reduction Act incorporated a related approach to clean-vehicle credits, making eligibility turn in part on battery-component and critical-mineral rules, including restrictions on inputs from foreign entities of concern. Extending this logic to tariff preferences, procurement, and strategic subsidies would shift the burden of proof from regulators to firms by conditioning benefits on verified-origin disclosures. In practice, firms are generally better positioned than regulators to document their own supply chains and identify where dependencies remain.
Neither move will be cost-free. Cheap imports, low-friction trade, low exposure to China, and high supply-chain transparency are unlikely to all be achieved at once. Building that resilience means more intrusive disclosure, higher compliance costs, and tougher negotiations with connector economies. Allies face the same triangulation problem. Washington should work with them to build interoperable supply-chain transparency and resilience tools, from the European Union’s battery passport and due-diligence requirements to Japan’s economic-security framework for specified critical products and stable supply plans.
Washington need not map the path of every screw in the global economy. It should concentrate disclosure and monitoring on the chokepoints where hidden dependencies would matter most in a crisis. Otherwise, superficial diversification may become a dangerous proxy for true resilience.
Notwithstanding the fall in direct exposure shown in trade statistics, those statistics fail to capture every technical input or routing arrangement that continues to connect strategic markets. The harder question is whether states can identify these vulnerabilities before they are weaponized, and whether their institutions can adapt when the political climate turns hostile. A trade barrier that produces diversion but not resilience is a failure of statecraft.
Interdependence is a structural reality that an executive order cannot dissolve. The danger today is a more opaque architecture of dependence, in which strategic exposure persists even as its statistical paper trail is obscured.
That leaves some questions worth exploring. How much of the fall in direct imports from China is offset by China-linked content in what connector economies ship to the United States? How much of the new strategic capacity in those countries is Chinese-owned or technologically controlled rather than merely located there? Moreover, do measures of embedded exposure predict disruption better than bilateral trade shares? Input-output data, customs records, and ownership databases can begin to answer those questions. If local value-added rises while Chinese input intensity and control fall, triangulation may prove more transitional than structural. If those trends do not materialize, the dependence is likely more durable than the bilateral data suggest. De-risking may continue, but it must stop mistaking a cleaner map for a safer terrain.
