Semiconductor Supply Chain

How silicon moves through a global coordination system that no single entity controls.

Introduction

A supply chain is the sequence of transformations that turns raw materials — sand, metals, gases — into finished products like chips, phones, and cars, crossing organizational and geographic boundaries at each stage. In semiconductors, this sequence is defined by extreme specialization, capital intensity that excludes most participants, and lead times that prevent the system from responding to demand changes for years.

No single company on earth can build a chip alone. A modern semiconductor begins as purified sand and ends as a component in a phone, a car, or a medical device. Between those points lies a coordination system spanning multiple continents, dozens of specialized firms, and tolerances measured in atoms. Each node depends on others, and disruption at any point propagates through the entire structure.

What makes this supply chain structurally unusual is not complexity alone but the combination of extreme specialization, geographic concentration, and long lead times. Capacity cannot be added quickly. Alternatives rarely exist. Small disturbances produce large downstream effects.

The Structure of the Chain

Each stage of semiconductor production is performed by a different set of specialized firms, and at most stages, only a handful of companies in the world can do the work. This extreme specialization means every node is a potential bottleneck—there are no quick substitutes when one fails.

The chain begins with raw materials. Silicon must be refined to extraordinary purity—impurities measured in parts per billion. A small number of firms, concentrated in Japan and Germany, produce the silicon wafers that foundries use as substrates. Wafer production requires specialized equipment and expertise that cannot be replicated quickly.

Design happens separately from manufacturing. Companies like Apple, Qualcomm, and Nvidia design chips but do not fabricate them. They rely on foundries—primarily TSMC in Taiwan—to turn designs into physical products. This separation is structural, not accidental. It lets design firms avoid the capital intensity of fabrication, but creates dependency on a small number of manufacturing partners.

Fabrication requires lithography equipment, etching tools, deposition systems, and testing apparatus. Each has its own concentrated supplier base. ASML is the sole manufacturer of extreme ultraviolet lithography machines — the equipment required for leading-edge chips — because the physics of producing light at 13.5 nanometer wavelengths requires technology that no other company has been able to replicate. This single-source dependency means that the entire industry's capacity to advance is gated by one company's production schedule. Applied Materials, Lam Research, and Tokyo Electron supply other critical equipment on long production cycles — ordering today means delivery in one to two years.

After fabrication, chips are packaged and tested—often in yet another country. Assembly, testing, and packaging operations are concentrated in Southeast Asia, particularly Malaysia, Vietnam, and the Philippines. The finished chips then enter the supply chains of device manufacturers, distributors, and contract electronics assemblers.

Structural Patterns

• Geographic Concentration at Critical Nodes — Advanced chip fabrication is concentrated in Taiwan and South Korea. Lithography equipment comes from the Netherlands. Silicon wafers come from Japan and Germany. Each concentration point is a single point of vulnerability for the entire system.
• Long Feedback Delays — Building a new fabrication facility takes three to five years and costs tens of billions of dollars. This means the system cannot respond quickly to demand changes. When shortage signals arrive, the earliest possible capacity response is years away.
• Cascading Dependency — Each stage depends on the previous one. A shortage of neon gas (used in lithography) constrains chip production, which constrains automotive production, which constrains vehicle availability. The chain transmits disruptions forward with amplification.
• Inventory as Buffer — Because lead times are long and disruptions are costly, participants hold inventory as a buffer against uncertainty. The 2020–2022 chip shortage revealed that many downstream users had reduced these buffers in pursuit of efficiency, removing the system's shock absorption.
• Specialization Without Substitution — At the leading edge, there are no alternative suppliers for many inputs. If TSMC cannot produce, there is no equivalent substitute. If ASML cannot deliver, fabrication capacity cannot expand. Specialization increases efficiency but eliminates redundancy.
• Information Asymmetry Across the Chain — End consumers of chips (automakers, appliance manufacturers) often have limited visibility into the upstream supply chain. They order from distributors or Tier 1 suppliers and may not know which foundry produces their components until a disruption forces them to find out.

Flows and Constraints

Material flows are slow relative to demand changes. A sudden increase in demand cannot be met for months—fabrication schedules are set weeks or months in advance, and capacity is allocated through long-term contracts. Spot availability is limited and expensive.

Capital flows shape the chain's evolution. Building a fabrication facility requires sustained investment over years before any revenue — because the equipment operates near physical limits and must be replaced with each technology generation. This capital intensity means only a few firms can participate in leading-edge fabrication, reinforcing the same concentration that the physics of precision manufacturing creates at every other stage of the chain.

Government subsidies (the US CHIPS Act, European Chips Act, similar programs in Japan and South Korea) attempt to redistribute fabrication capacity more broadly. Whether these programs change the structure or merely add capacity at existing concentration points remains to be seen.

Information flows are uneven. Foundries have detailed visibility into their own capacity and order books. Design firms know what products they are planning. But the mapping between downstream demand and upstream capacity is imprecise. During the 2020–2022 shortage, double-ordering (placing orders with multiple suppliers for the same need) distorted demand signals, making it harder for the system to calibrate its response.

What the 2020–2022 Shortage Revealed

The global chip shortage that began in 2020 was not caused by a single event. It emerged from several structural features interacting: lean inventory practices that had removed buffers, a sudden demand shift toward consumer electronics during lockdowns, and a fabrication system that could not add capacity faster than its three-to-five-year build cycle.

The shortage made visible what had been invisible: the degree to which diverse industries—automotive, medical devices, consumer electronics, industrial equipment—all depended on the same constrained fabrication capacity. Automakers discovered they were competing with smartphone manufacturers for the same production slots, and that their purchasing volume gave them less priority than they had assumed.

Recovery was slow because it had to be. Capacity additions announced in 2021 began producing chips in 2024 and 2025. The system's response time is structural, not a management failure.

Risks and Pressures

Geopolitical pressure is reshaping the chain. Advanced fabrication concentrated in Taiwan creates strategic risk governments are actively trying to mitigate. Efforts to build capacity in the United States, Europe, and Japan are underway, but replicating the ecosystem—supplier networks, workforce expertise, operational culture—takes longer than building the physical structures.

The tension between efficiency and resilience is structural. Just-in-time practices minimize inventory costs but maximize vulnerability. How much redundancy to build into the system is now an active debate—redundancy has ongoing costs while disruptions are intermittent.

Technology transitions add pressure. Each new chip generation requires more precise equipment, cleaner materials, and tighter coordination. The chain becomes more capable but also more fragile—fewer firms can participate at each new level of precision.

What This Reveals About Industrial Structure

• Concentration is a structural outcome — When capital requirements are extreme and expertise accumulates slowly, industries naturally concentrate. This is not a market failure but a feature of the underlying economics.
• Buffers have value that efficiency metrics do not capture — Inventory, redundant suppliers, and excess capacity appear as costs in normal conditions and as essential infrastructure during disruptions.
• Response time is set by the slowest constraint — A supply chain can only adapt as fast as its slowest-to-change component allows. For semiconductors, that component is fabrication capacity.
• Visibility decreases with distance from disruption — Firms far from a bottleneck may not recognize their dependency until the constraint binds.
• Government intervention changes flows but not physics — Subsidies can redirect capital, but they cannot compress the time required to build expertise, supply networks, and operational capability.

Inside CompanyGraph

View this perspective inside CompanyGraph.