Supply Chains
How modern production works, what COVID revealed about its fragility, and whether reshoring is real or rhetorical.
(WTO; about 70% of world trade runs through global value chains, on OECD estimates)
(Semiconductor Industry Association and Boston Consulting Group, 2021)
(in 2022 a cyberattack on one supplier stopped all 14 of Toyota's Japanese plants for a day)
A note on framing. Supply chains became visible to the general public during the COVID-19 pandemic, when shortages of everything from semiconductors to toilet paper made clear that modern economies depend on a physical network that most consumers never think about. This page tries to describe how that network actually works, why it was designed the way it was, what the pandemic revealed, and what the post-2020 restructuring looks like in practice rather than in rhetoric. Sources include the WTO, the OECD, US Census trade data, the New York Fed and IMF PortWatch.
How modern supply chains work
A modern supply chain is not a chain - it is a network. A typical smartphone contains components from dozens of countries: cobalt mined mostly in the Democratic Republic of Congo, rare earth elements mined and refined mostly in China, chips fabricated in Taiwan, final assembly in China, India or Vietnam, and software developed in the US and India. No single company controls the entire process. Instead, "lead firms" (Apple, Samsung, Toyota) coordinate tiers of suppliers, with Tier 1 suppliers (direct component makers) managing their own Tier 2 and Tier 3 suppliers, often with limited visibility beyond the immediate relationship.
The system was optimized over decades for efficiency. Just-in-time (JIT) manufacturing, pioneered by Toyota in the 1960s-70s (the Toyota Production System), minimizes inventory by delivering parts exactly when they are needed on the production line. Lean manufacturing extends this logic: every buffer, every warehouse, every redundant supplier is a cost to be eliminated. The system works brilliantly under stable conditions - it reduced waste, lowered costs, and helped push down the prices of manufactured goods from the mid-1990s to the 2010s. It works less well when conditions are not stable.
The COVID stress test
The pandemic exposed exactly what JIT efficiency had traded away: resilience. When Chinese factories shut down in early 2020, the effects cascaded globally within weeks. Auto manufacturers could not get chips. Electronics makers could not get components. Pharmaceutical companies could not get active ingredients. Container shipping rates spiked from roughly $1,500 to over $10,000 per forty-foot container. Ports backed up. The "bullwhip effect" - where small demand signals amplify into large swings at each tier of the supply chain - turned modest disruptions into months-long shortages.
The semiconductor shortage was the most consequential. Global auto production lost around 7.7 million vehicles in 2021 alone to chip shortages, on the consultancy AlixPartners' September 2021 forecast for the year. That figure is worth reading alongside its own history: the same firm had forecast 3.9 million lost vehicles in May, about half as many. The disruption was badly underestimated in real time by the people closest to it. The root cause was a convergence of factors: demand surged for consumer electronics during lockdowns, auto manufacturers had canceled chip orders early in the pandemic (expecting a downturn), and semiconductor fabrication has extreme lead times (building a new fab takes 3-5 years and costs $10-20 billion). The concentration of advanced chip manufacturing in Taiwan (TSMC produces roughly 90% of the world's most advanced chips) meant that a single point of geopolitical or natural disaster risk affected the entire global electronics supply chain.
Just-in-case: the post-pandemic shift
The post-2020 mantra in supply chain management has been a shift from "just-in-time" to "just-in-case" - building in buffers, diversifying suppliers, holding more inventory, and shortening supply chains where possible. The question is how much of this shift is real versus rhetorical.
The evidence is mixed. US business inventories were rebuilt in 2021-22, but by July 2026 their ratio to sales (1.30) was below its pre-pandemic level of 1.43 (Census). Companies talk about "dual sourcing" and "nearshoring" but the underlying economics have not changed: manufacturing in China and Southeast Asia remains substantially cheaper than in the US or Europe for most products. What has changed is the geopolitical risk calculus. US-China trade tensions, the Taiwan contingency, the pandemic and now the Hormuz crisis have made supply chain concentration a boardroom-level strategic concern rather than a back-office logistics question.
Reshoring and friend-shoring: rhetoric versus reality
"Reshoring" (bringing manufacturing back to the home country) and "friend-shoring" (relocating supply chains to geopolitically aligned countries) have become policy priorities in the US, EU, and Japan. The US CHIPS Act (2022) set aside $39 billion for chip-plant subsidies; the EU says its 2023 Chips Act has drawn over €52 billion in investment. TSMC has made chips in Arizona since late 2024 and is building fabs in Japan and Germany; Intel has dropped its German fab and slowed its Ohio one. These are real investments with real employment consequences. Washington has also leaned on tariffs: after the Supreme Court struck down its emergency-powers tariffs in February 2026, a 10% Section 122 stopgap gave way on July 24 to Section 301 tariffs of 10-12.5% on 60 economies. The average effective US tariff rate was 6.7% in July, up from 2.3% in January 2025 (Penn Wharton Budget Model).
But the broader reshoring picture is more complex. China's share of US goods imports fell from 21% in 2018 to 9% in 2025 (Census). Imports from Mexico, Vietnam and India rose by more than those from China fell, suggesting diversion rather than reshoring; WTO research finds that rerouting through third countries explains only a small part of the fall in direct US-China trade. China's overall manufacturing output has continued to grow. Mexico has been a major beneficiary of "nearshoring" - its share of US imports rose from 13.6% in 2018 to about 17% in January-July 2026, driven partly by the US-Mexico-Canada trade agreement and partly by Chinese companies setting up Mexican operations to access the US market.
India has positioned itself as an alternative manufacturing base, with Apple now assembling about a quarter of its iPhones there (some 55 million in 2025, on Bloomberg's count). Vietnam has attracted semiconductor packaging and electronics assembly. Whether any of these shifts represent a fundamental restructuring of global manufacturing geography or a redistribution within existing patterns is one of the most contested questions in trade economics.
Digital supply chains
Technology is reshaping supply chain management itself. Internet-connected sensors enable real-time tracking of containers, pallets, and individual products through the entire logistics chain. Digital twins - virtual models of physical supply chains - allow companies to simulate disruptions and test responses before they occur. Blockchain-based provenance tracking aims to create tamper-proof records of where products and components have been, though adoption has been slower than early hype suggested.
The most consequential digital shift may be in visibility. Historically, lead firms had limited knowledge of their Tier 2 and Tier 3 suppliers. The pandemic revealed that this ignorance was itself a risk. Regulation has been pushing in the same direction, though less decisively than it first appeared. The German Supply Chain Due Diligence Act has been in force since 2023, but it requires large companies to check mainly their direct suppliers for human rights and environmental risks. The EU's Corporate Sustainability Due Diligence Directive was adopted in 2024 and looked like the bigger step, but the bloc then scaled it back: the 2026 Omnibus package limits it to firms with more than 5,000 employees and delays application to 2029. Where the rules do bite, they create a compliance cost but also give many companies their first comprehensive picture of who actually makes their products.
Where analysts disagree
Is deglobalization happening? Trade-to-GDP ratios have plateaued since 2008 but not declined. Global trade volumes continue to grow in absolute terms. Some economists (like Richard Baldwin) argue that what is changing is the composition of trade (more services, more digital, more regional) rather than its volume. Others see a real break in the split into US-aligned and China-aligned blocs: the WTO estimates that US-China trade grew about 30% more slowly than each country's other trade in 2018-24.
Can reshoring work at scale? Skeptics point out that the US, Europe, and Japan lack the skilled manufacturing workforce, the supplier ecosystems, and the cost structure to replicate what China built over three decades. Advocates argue that automation, AI-driven manufacturing, and strategic subsidies can close the gap. The early evidence from CHIPS Act fab construction (delays, cost overruns, worker shortages) suggests that rebuilding manufacturing capacity is harder than announcing it. US data cut both ways: factory construction spending roughly tripled from 2019 to 2024, then fell by a third, and factory jobs are slightly below their pre-pandemic level (Census Bureau, Bureau of Labor Statistics).
How much resilience is worth? Every buffer, every redundant supplier, every unit of safety stock is a cost. The efficiency gains of the JIT era were real and benefited consumers through lower prices. The resilience question is ultimately a pricing question: how much are societies willing to pay - in higher prices, in reduced efficiency, in taxpayer-funded subsidies - for the insurance value of a more robust supply chain? The answer depends on assumptions about how frequently severe disruptions will occur, and on those assumptions, reasonable analysts disagree.
Modern supply chains are the circulatory system of the global economy. They are invisible when they work and catastrophic when they fail. The pandemic and the 2026 Hormuz shock have made them visible as a strategic concern, but the underlying tension - efficiency versus resilience, cost versus security, globalization versus national control - has no clean resolution. Every choice involves a trade-off, and understanding the trade-offs is a prerequisite for evaluating the reshoring claims that now dominate policy debate.
The paths from here
Dated September 2026, amid the Hormuz shipping crisis and new US tariffs. The restructuring under way now is slower and more partial than the announcements suggest. Each path below is one realistic shape the next decade could take, and they are not mutually exclusive - different sectors are already moving down different ones.
Selective de-risking rather than decoupling
Firms and governments hold buffers and second suppliers in a narrow set of categories judged strategic - semiconductors, pharmaceutical inputs, critical minerals, defense components - and leave the rest optimized for cost. Trade volumes keep growing; the map of who supplies what shifts at the margin.
Will it happen? This is the base case and is largely what has happened since 2020. Aggregate world trade has not fallen - the WTO recorded merchandise volumes growing 4.6% in 2025 and forecasts 1.9% for 2026 despite the Iran war - so the restructuring is happening inside a still-expanding system. What has changed is concentration in specific categories, and the willingness of governments to pay for redundancy in them.
Bloc formation
Supply chains reorganize around political alignment rather than cost, with a Western-aligned bloc, a China-centered one, and a large set of countries trading with both. Duplicate capacity gets built on each side. Consumer prices rise modestly and permanently; resilience improves for the categories that were duplicated.
Will it happen? Partially, and it is furthest advanced in semiconductors, critical minerals and defense-adjacent goods. China's April 2025 rare-earth export controls still apply, though a US-China truce, now extended to January 10, 2027, suspends its wider October 2025 controls. The obstacle is that most of the world's middle-income countries have no interest in choosing, and their refusal is what keeps the blocs porous.
Reshoring stalls on cost and labor
The announced factories are built more slowly than planned, run below capacity, and struggle to find workers with the specific skills required. Political attention moves elsewhere before the subsidies mature. The network reverts toward its cost-optimized shape with a thin resilience layer on top.
Will it happen? Partly, already. Several flagship projects have slipped, among them TSMC's first Arizona fab (delayed in 2023 for lack of skilled workers) and Intel's Ohio fab (now due in 2030 or later). The Semiconductor Industry Association's 2023 projection is that the US chip industry will add about 115,000 jobs by 2030 and that roughly 67,000 of them risk going unfilled. The subsidies have a political clock attached that is shorter than a factory's construction time.
Another shock resets the debate
A major disruption - a Taiwan crisis, a chokepoint closure, a pandemic, a cascading cyber incident - exposes a dependency that current planning did not cover. The response is faster and more expensive than anything undertaken voluntarily, and it is aimed at the specific failure rather than at resilience generally.
Will it happen? It is happening. Since March 2026 the Iran war has restricted shipping through the Strait of Hormuz, the route for a fifth of the world's oil and liquefied natural gas and a third of traded fertilizer; tracked transits fell from 85.5 a day in 2025 to about 4 in September (IMF PortWatch). Brent crude topped $100 a barrel, urea prices rose over 80% by April, and container freight rates more than doubled (Drewry), yet the New York Fed's supply-chain pressure index peaked at under half its 2021 high. The pattern after 2020 suggests the response will be sharp, narrow, and aimed at this failure rather than the next one.
The base case is selective de-risking that leaves most of the network intact. The thing worth watching is not the volume of trade but the number of single points of failure in categories that matter, which is a different measurement and one that has improved less than the spending would suggest.
What this means for you
If you run or work in a business that buys things
The useful question is not "where does my supplier operate" but "where does my supplier's supplier operate," which is the tier most firms cannot see. Mapping beyond the first tier is unglamorous and is rarely funded until after a disruption. It is also the measure most consistently identified in post-2020 reviews as separating firms that coped from firms that did not.
If you follow the reshoring debate
Announcements and construction are different things, and construction and steady-state production are different again. A useful discipline is to note the date a project was announced and check what it is producing three and five years later. The gap between the two is where most of the political argument is actually being conducted.
If you are trying to read price movements
Shortages that reach consumers usually originate several tiers upstream and months earlier, in a component nobody outside the industry can name. When a price spikes, the question worth asking is which input became scarce and whether the scarcity is a capacity problem or a logistics one. The two resolve on very different timescales.
If you are weighing the efficiency argument
Just-in-time was not a mistake. It delivered decades of falling prices and rising variety, and the firms that adopted it outcompeted those that did not. What it traded away was a buffer whose value only appears in the rare years when it is needed. That is a trade-off rather than an error: proposals to unwind it should say who pays the standing cost, and defenses of it, who pays when the buffer is missing.
The mechanics behind this
Supply chains are an applied case of three mechanisms covered from first principles elsewhere on this site. If the analysis above rests on ideas you want to see built up properly, start here.
