Supply chain decoupling from China has become the rallying cry of Western industrial policy. “De-risking,” “friend-shoring,” “supply chain sovereignty”: the terminology varies, but the promise is consistent. Sever the dependencies. Bring production home. Insulate the economy from geopolitical risk. The problem is that the math does not work.
The economic reality, as a Center for European Policy Analysis senior fellow put it bluntly in March 2026, is that “decoupling (full, clean, and enforceable) is impossible.”[s] This is not a matter of political will or funding levels. It is a function of industrial structures built over three decades that cannot be unwound in three years, or even ten. The structural vulnerabilities we saw during the 1973 oil embargo are back, but this time the concentration is in minerals, chemicals, and manufacturing processes rather than petroleum.
The Rare Earth Chokepoint
China’s grip on rare earth supply chains is often cited, but the numbers remain poorly understood. China accounts for roughly 70% of global rare earth mining.[s] That is manageable; other countries have reserves. The real leverage comes downstream. China controls close to 90% of the world’s rare earth refining and processing capacity.[s] And it produces 93% of global permanent magnets, the finished components that go into everything from F-35 fighters to electric vehicles.[s]
U.S. rare earth magnet capacity remains narrow. In October 2025, CSIS identified Noveon Magnetics as the only U.S. rare earth magnet manufacturer; by 2026, MP Materials and Vulcan Elements were also reporting U.S. finished magnet production or planned capacity expansions.[s][s][s] The Department of Defense has invested $400 million in MP Materials and signed long-term offtake agreements. But ramping these capabilities will take years, during which China retains significant leverage over supply chains crucial for national and economic security.[s]
How long to build alternatives? At a US Capital Access Forum in Singapore, mining executive Mick McMullen told Fortune that “China has been at this for more than 30 years”; Fortune reported that it could, optimistically, take countries a decade to build their own rare earth industry.[s]
Beijing understands this leverage. In October 2025, Chinese authorities unveiled extraterritorial export restrictions that would apply to any product containing 0.1% or greater Chinese-origin rare earth content, whether produced domestically or abroad.[s] TD Economics concluded that “replacing Chinese supplies will likely be nearly impossible before the 1-year truce expires” in November 2026.[s] Any realistic supply chain decoupling timeline starts after that deadline passes.
The Pharmaceutical Dependency Nobody Mentions
While rare earths dominate headlines, the pharmaceutical supply chain represents an equally acute vulnerability. India is often cited as the alternative “Pharmacy of the World.” But India’s pharmaceutical prowess depends on Chinese inputs. Approximately 70% of India’s Active Pharmaceutical Ingredient needs, and up to 90% for critical antibiotics like penicillin, are met by Chinese imports.[s]
Direct statistics need care. Brookings researchers reviewing US drug exposure to China in 2025 found Chinese API in perhaps a quarter of drug volume sold in the United States, and warned that higher estimates often overstate API reliance by blending APIs with upstream precursors. The dependency grows farther upstream, where Chinese Key Starting Materials and intermediates feed Indian and other manufacturers.[s]
The bottleneck is at the bottom of the chemical hierarchy: Key Starting Materials and intermediates. An April 2026 Council on Geostrategy report estimated that roughly 90% of global 6-APA production capacity is located in China, making penicillin-class antibiotics unusually exposed to upstream interruption.[s]
Supply chain decoupling in pharmaceuticals faces the same structural barrier as rare earths. The “dirty chemistry” that produces Key Starting Materials, processes involving toxic byproducts, was pushed out of Western countries by environmental regulations over the past two decades. China absorbed that capacity. Rebuilding it would require not just capital but regulatory tolerance for chemical plants that American and European communities rejected a generation ago.[s]
The Semiconductor Capability Gap
The CHIPS Act committed $39 billion to domestic semiconductor manufacturing. Headlines celebrated the investment as a turning point. But capital is not capability.
Justin Gillebo, a former Microsoft Azure supply chain executive, draws the distinction sharply: “Capacity is what you own. Capability is what you can execute.”[s] Taiwan Semiconductor Manufacturing Company controls nearly 70% of the global foundry market. “They built that dominance over four decades: optimizing yields, establishing a supplier ecosystem, and refining maintenance routines over millions of production hours. You cannot replicate that ecosystem by building a building.”[s]
The capability gap is already showing. TSMC’s first Arizona fab required over 1,000 technicians flown in from Taiwan to train local workers. Intel’s 18A process node has faced yield problems. And the workforce pipeline is empty. The US semiconductor workforce is down 43% from its 2000 peak, and the Semiconductor Industry Association projects a 67,000-worker shortfall by 2030.[s]
Policy has adapted to reality. The Trump administration shifted to case-by-case licensing for certain AI chips to China. TSMC received a license to ship American chipmaking tools to its Chinese fabs. Export controls “have not disappeared; they have become conditional and administratively dense.”[s] The bright lines between permitted and prohibited, civilian and military, ally and adversary have blurred.
What Supply Chain Decoupling Would Actually Cost
European Central Bank economists built a novel input-output model isolating 129 green-transition products to simulate what supply chain decoupling between a US-centric West and a China-centric East would actually produce. The results: global trade in targeted green products would fall 10 to 20%, global welfare would decline by up to 3%, and annual greenhouse gas emissions would rise by approximately 50 million tonnes. Over 20 years, the cumulative additional emissions would be comparable to a year’s emissions from Japan or Brazil.[s]
The ECB researchers found that previous models systematically underestimated these costs because standard input-output tables bundle green and non-green products together, masking the specific interlinkages. Many key input products have low substitution elasticities: they are hard to replace, so disruptions ripple through global supply chains with amplified force.[s]
The climate irony is stark. Supply chain decoupling pursued in the name of security would undermine the green transition by making clean-energy technologies more expensive and slowing their adoption.
The Integration That Cannot Be Unwound
The scale of economic integration defies the decoupling narrative. Bureau of Economic Analysis data shows US affiliates in China generated over $640 billion in revenue in 2023.[s] Critically, 70% of that production, roughly $441 billion, was sold in the Chinese market, not exported back to the United States.[s]
This is not offshoring in the traditional sense. American companies are not just using China as a factory; they are serving Chinese consumers. Forcing them out would cede that market to Chinese competitors while destroying the technology ecosystem lock-ins that currently benefit American interests.
The Information Technology and Innovation Foundation argues that “the discussion of forcing companies to exit the Chinese market is an overreaction and fails to account for ways in which having American companies in China serves the U.S. national interest.”[s]
A Decade of Managed Competition
Supply chain decoupling as currently conceived is not a strategy. It is a slogan. The 2024 House version of the BIOSECURE Act included a grandfather clause allowing some existing contracts with named Chinese companies to continue until January 1, 2032; a July 2025 revised Senate FY2026 NDAA draft narrowed that structure, with no grandfathering for companies already on the Department of Defense 1260H list and a five-year grandfathering period for later designated companies.[s] A June 2024 L.E.K. survey found that 26% of life sciences companies were looking to shift away from current Chinese partners, but only 2% had taken actual steps to unwind relationships with companies named in the US BIOSECURE Act.[s]
UN Trade and Development reports that nearly two-thirds of global trade takes place within value chains, and since 2020, around 18,000 discriminatory trade measures have been introduced.[s] The measures are multiplying, but the dependencies remain.
Semiconductor supply chains “remain interconnected, capital-intensive, and geographically distributed. Attempts to sever them have proven costly and destabilizing.”[s] The decade ahead will not bring supply chain decoupling. It will bring managed competition: conditional controls, selective tariffs, targeted investments in marginal alternative capacity, and continued dependence on the industrial structures that three decades of globalization built.
The math does not lie. The political calendar does not match the industrial calendar. And no amount of rhetoric will compress thirty years of integration into three years of unwinding.
