The Semiconductor Supply Chain: How Three Companies Control the World's Chips

01 The Most Complex Supply Chain on Earth

The modern semiconductor supply chain is unlike any other manufacturing system humanity has built. A single microprocessor might pass through facilities in five countries before reaching a consumer, with each step requiring equipment that costs more than a commercial airliner. The chain begins with raw silicon — purified to 99.9999999 percent purity — and ends with a packaged chip containing tens of billions of transistors, each smaller than a wavelength of light.

What makes this industry structurally unique is the concentration of expertise at each stage. Where most manufacturing industries have dozens of competitive suppliers at every step, the semiconductor chain features near-monopolies at its most critical nodes. One company makes the lithography machines that print transistor patterns. One company manufactures most of the world's most advanced chips. A handful of companies design the architectures. This concentration is not an accident of markets but a consequence of physics, capital intensity, and decades of cumulative engineering knowledge that cannot be easily replicated.

The result is a supply chain where a single disruption — a factory fire, an export restriction, a power outage — can cascade through the entire global electronics industry. The 2021 chip shortage demonstrated this fragility when automakers idled factories worldwide because of delayed orders for chips costing a few dollars each. Understanding who controls these bottlenecks is essential to understanding the geopolitics and economics of the 21st century.

02 ASML and the EUV Monopoly

ASML Holding, headquartered in Veldhoven, Netherlands, is the only company on Earth capable of manufacturing extreme ultraviolet (EUV) lithography systems — the machines required to print the smallest transistor features on the most advanced chips. This monopoly is not a matter of patents alone; it reflects an engineering challenge so severe that no competitor has mounted a serious challenge despite two decades of effort.

EUV lithography works by generating plasma at 250,000 degrees Celsius — thirty times hotter than the surface of the Sun — by blasting microscopic droplets of molten tin with a high-power laser fifty thousand times per second. The plasma emits light at a wavelength of 13.5 nanometers, which is then collected by the most precisely polished mirrors ever manufactured and focused onto a silicon wafer to print circuit patterns. Each EUV machine contains over 100,000 parts and costs approximately $200 million. ASML's next-generation High-NA systems, which promise even finer resolution, cost roughly $400 million per unit.

ASML does not build these machines alone. Its supply chain includes Zeiss for optics, Trumpf for lasers, and Cymer for light sources. But ASML is the sole integrator, and the Dutch government's export controls — influenced heavily by United States pressure — determine which countries can receive these machines. China is excluded from EUV entirely, which constrains its most advanced domestic chipmakers to older deep ultraviolet (DUV) technology and limits the minimum feature size they can achieve.

03 TSMC's Foundry Dominance

Taiwan Semiconductor Manufacturing Company (TSMC) produces roughly 90 percent of the world's most advanced processor chips. Founded in 1987 by Morris Chang, TSMC pioneered the pure-play foundry model — manufacturing chips designed by other companies rather than designing its own. This decision removed the conflict of interest that plagued integrated device manufacturers like Intel and allowed TSMC to become the trusted manufacturing partner for Apple, Nvidia, AMD, Qualcomm, and dozens of others.

TSMC's advantage compounds with each generation. Each new process node — 5nm, 3nm, 2nm — requires capital expenditures in the tens of billions of dollars. TSMC spent approximately $30 billion on capital expenditures in 2023 alone. The company's gigafabs in Taiwan operate at a scale that drives down per-chip costs, and its engineering workforce has accumulated institutional knowledge about yield optimization that takes decades to build. A new entrant could spend tens of billions and still lack the process recipes, the defect-reduction experience, and the ecosystem of suppliers that TSMC has built over thirty-five years.

This geographic concentration carries extraordinary risk. Taiwan produces the majority of the world's advanced chips from facilities that sit within range of a potential military conflict. The United States, European Union, Japan, and South Korea have all launched subsidy programs to build domestic chip manufacturing capacity, but these efforts face a fundamental challenge: the knowledge required to run an advanced fab at competitive yields is concentrated in Hsinchu, not in Arizona or Dresden.

04 Intel's Foundry Pivot

Intel, once the undisputed leader in semiconductor manufacturing, lost its process advantage to TSMC during the 2010s. Under CEO Pat Gelsinger, Intel launched an ambitious turnaround strategy: to become a major foundry operator, manufacturing chips designed by other companies while also producing its own. This strategy, branded Intel Foundry Services, represents a bet that Intel can regain process leadership at the 18A (1.8nm) node while simultaneously offering capacity to fabless chip companies seeking geographic diversification away from Taiwan.

The pivot faces steep challenges. Intel must convince potential foundry customers — companies that are often its direct competitors in the CPU market — that their intellectual property will be protected. TSMC's neutral position as a pure foundry was a competitive advantage precisely because it had no competing chip designs. Intel also must execute its process roadmap flawlessly after years of delays at 10nm and 7nm nodes. The CHIPS and Science Act, which provides roughly $52 billion in subsidies for domestic semiconductor manufacturing, has supported Intel's expansion plans in Arizona, Ohio, and Oregon, but government funding covers only a fraction of the total cost.

05 The Geopolitical Dimension

Semiconductors have become the strategic resource of the 21st century, analogous to oil in the 20th. The United States has restricted China's access to advanced chipmaking equipment through export controls, citing national security concerns. These restrictions, tightened significantly in October 2022 and expanded in 2023, prevent Chinese foundries from acquiring EUV systems and restrict access to the most advanced DUV models as well.

China has responded with a massive domestic investment program, pouring over $100 billion into chip manufacturing subsidies and research. Companies like SMIC, Huawei, and YMTC have made progress in mature nodes and have surprised observers with products like the Huawei Mate 60 Pro's 7nm Kirin 9000S chip, manufactured despite the export controls. However, these chips are produced using older multi-patterning DUV techniques with lower yields and higher costs than TSMC's EUV-based processes. The gap between China's domestic capability and the global frontier remains significant.

The CHIPS Act in the United States, the European Chips Act, and Japan's semiconductor strategy all reflect a recognition that geographic concentration in Taiwan represents an unacceptable strategic risk. However, these programs primarily subsidize facility construction; they cannot easily replicate the dense ecosystems of equipment engineers, process integration specialists, and materials suppliers that have developed in Taiwan over decades.

06 The Bottleneck Economy

The semiconductor supply chain is best understood as a series of bottlenecks, each controlled by a small number of companies. Photoresist chemicals come primarily from Japanese firms JSR, Tokyo Ohka, and Shin-Etsu. Advanced packaging substrates are dominated by Ibiden and Unimicron. The ultra-pure silicon wafers are made by Shin-Etsu and SUMCO, both Japanese. Each of these niche monopolies represents decades of accumulated expertise that new entrants cannot quickly replicate.

This bottleneck structure means that the semiconductor industry is unusually vulnerable to targeted disruptions. A single earthquake in Taiwan, a diplomatic crisis between the Netherlands and China, or a factory fire at a specialist supplier can halt production lines worldwide. The automotive industry discovered this in 2021 when a fire at a Renesas fab in Japan, combined with winter storms in Texas and drought in Taiwan, contributed to a global chip shortage that cost automakers an estimated $210 billion in lost revenue.

07 The Path Forward

The semiconductor supply chain is unlikely to become less concentrated in the near term. EUV development took ASML over twenty years and cost tens of billions of euros; no competitor is positioned to replicate this within a decade. TSMC's process knowledge advantage at advanced nodes continues to compound with each generation. The most realistic scenario is a slow geographic diversification — additional fabs in Arizona, Japan, Germany, and possibly India — while the critical bottlenecks remain concentrated among the same few companies.

The deeper question is whether the current industry structure is sustainable. Chip complexity continues to increase, capital costs per fab are rising, and the number of companies that can afford to compete at the leading edge is shrinking. In 2000, there were over twenty companies manufacturing leading-edge chips. By 2025, only TSMC, Samsung, and Intel remain in that race. If Intel's foundry pivot fails, the world could face a future where one company in one country manufactures nearly all of the most advanced chips — a concentration of strategic capacity unprecedented in industrial history.

For policymakers, technologists, and investors, the lesson is clear: the semiconductor supply chain is not a market that self-corrects. The physics of lithography, the economics of fab construction, and the decades of accumulated engineering knowledge create natural monopolies that require deliberate intervention to manage. The next decade of chip manufacturing will be shaped as much by government policy and strategic competition as by technological progress.