TSMC

Based on Wikipedia: TSMC

The Company That Makes Everything Work

Here's a question that should keep geopolitical strategists up at night: What if a single company, located on a small island in the Pacific, manufactured the chips inside nearly every smartphone, laptop, server, and artificial intelligence system on Earth?

That's not a hypothetical. That's Taiwan Semiconductor Manufacturing Company, better known as TSMC.

The phone in your pocket almost certainly contains a chip made by TSMC. The same goes for the processor in your computer, the graphics card that renders your games, and increasingly, the accelerators that train the AI models reshaping our world. Apple, Nvidia, AMD, Qualcomm—these are household names, but none of them actually manufacture their own chips. They design them. TSMC builds them.

This distinction—between designing a chip and actually fabricating one—turns out to be one of the most consequential divisions of labor in the modern economy.

The Foundry Model: A Revolution in Specialization

To understand TSMC, you first need to understand what a semiconductor foundry is, and why the concept was revolutionary when it emerged in the late 1980s.

Before TSMC, semiconductor companies were what the industry calls "integrated device manufacturers"—they designed their own chips and operated their own fabrication plants, known as fabs. Intel, for instance, both designed and manufactured its processors. This made sense when the industry was young, but it created an enormous barrier to entry. Building a modern fab costs tens of billions of dollars and requires years of development. Few companies could afford to play.

Morris Chang, who founded TSMC in 1987, had a different vision. What if a company manufactured chips for other companies, taking on the staggering capital costs of fabs so that designers could focus purely on design? This "pure-play foundry" model would democratize the semiconductor industry, allowing small teams with clever ideas to compete against giants.

The idea was so unusual that when Chang sought investors, Texas Instruments and Intel both turned him down. Only the Dutch electronics giant Philips was willing to take a chance, contributing fifty-eight million dollars, technology transfers, and intellectual property in exchange for about twenty-seven percent of the new company. The Taiwanese government, recognizing the strategic importance of semiconductors, kicked in nearly half the startup capital. The rest came from Taiwan's wealthiest families, who were more or less required by the government to invest.

From day one, TSMC was never really a private company in the conventional sense. It was a project of the Taiwanese state.

Why Taiwan?

Taiwan's dominance in semiconductor manufacturing is no accident. The island had been deliberately building its electronics industry for decades, and by the 1980s, its government viewed semiconductors as essential to national security—not just economic prosperity, but literal survival.

Taiwan exists in a precarious geopolitical position. The People's Republic of China considers the island a breakaway province that must eventually be reunified with the mainland, by force if necessary. Taiwan's democratic government disagrees. For decades, the island has relied on American security guarantees and its own military to deter invasion. But there's another deterrent, often called the "silicon shield."

The logic is grimly simple: if Taiwan makes the chips that power the world's technology, then allowing it to be conquered or destroyed becomes unthinkable. Every major economy depends on TSMC's output. A war that disrupted its fabs would trigger a global economic catastrophe far exceeding the 2020 pandemic.

This isn't just theory. TSMC today accounts for roughly ninety percent of the world's most advanced chips—those manufactured at the cutting edge of what's physically possible. The company's technology is so far ahead of competitors that even Intel, long the industry leader, now outsources some of its most advanced work to TSMC.

The Physics of the Impossible

What makes advanced chip manufacturing so difficult? The answer lies in the almost incomprehensible scale of modern semiconductors.

A transistor is a tiny switch that can represent a one or a zero—the binary foundation of all computing. The more transistors you can pack onto a chip, the more powerful it becomes. But making transistors smaller requires manufacturing tolerances that push against the limits of physics.

TSMC's current leading technology produces chips at what the industry calls the three-nanometer node. A nanometer is one billionth of a meter. For comparison, a human hair is about eighty thousand nanometers wide. A strand of DNA is about two and a half nanometers across. TSMC is manufacturing structures not much larger than individual molecules, billions of times over, with near-perfect precision.

This requires machines that don't exist anywhere else on Earth. The most critical is the extreme ultraviolet lithography system, built by a Dutch company called ASML. Each of these machines costs roughly two hundred million dollars, weighs over a hundred tons, and uses a process that sounds like science fiction: a laser vaporizes tiny droplets of molten tin fifty thousand times per second, creating plasma that emits extreme ultraviolet light, which is then bounced off mirrors polished to atomic smoothness to etch patterns onto silicon wafers.

TSMC was the first company to commercialize this technology at scale. That head start has proven nearly impossible to replicate.

The Customer List That Runs the World

TSMC's client roster reads like a who's who of the technology industry. Apple is the company's largest customer, and every iPhone, iPad, and Mac sold in recent years contains TSMC silicon. Nvidia, whose graphics processors have become the essential infrastructure for artificial intelligence, manufactures exclusively at TSMC. AMD, which has staged a remarkable comeback against Intel in recent years, owes much of its success to TSMC's manufacturing prowess.

But the list extends far beyond consumer electronics. Qualcomm's chips power most Android smartphones. MediaTek supplies processors to budget devices across Asia. Broadcom makes the networking equipment that connects the internet. ARM, which designs the architecture used by nearly every mobile processor, partners closely with TSMC to push what's physically possible.

Even integrated device manufacturers—companies that theoretically compete with TSMC by running their own fabs—outsource work to Taiwan. Intel now manufactures some chips at TSMC. So do NXP, STMicroelectronics, and Texas Instruments. The gravitational pull of TSMC's technology is simply too strong to resist.

The Geopolitics of Chips

This concentration of manufacturing capability in Taiwan has begun to alarm governments around the world. The COVID-19 pandemic exposed the fragility of global supply chains, and the semiconductor shortage that followed—which idled auto plants, delayed consumer electronics, and contributed to inflation—drove the point home with painful clarity.

What would happen if those supply chains were disrupted not by a virus, but by a war?

This question has prompted a wave of government action. The United States passed the CHIPS Act in 2022, offering tens of billions of dollars in subsidies to encourage domestic semiconductor manufacturing. Japan has poured money into its own chip industry. The European Union has announced similar plans. Everyone, it seems, wants their own TSMC.

TSMC has responded by expanding beyond Taiwan—though notably, not by abandoning the island. The company is building a major facility in Arizona, with an initial investment of twelve billion dollars that may eventually reach thirty-five billion across six factories. In Japan, TSMC has partnered with Sony and Denso to build a fab in Kumamoto that will supply chips to Japan's electronics and automotive industries. There are plans for a facility in Germany as well.

But there's a catch. These overseas fabs won't manufacture TSMC's most advanced chips. The three-nanometer and upcoming two-nanometer processes will remain in Taiwan, where the company's most skilled engineers live and where decades of institutional knowledge are concentrated. Some critics argue this defeats the purpose of geographic diversification. Others suggest it's entirely intentional—Taiwan's silicon shield depends on remaining indispensable.

The China Problem

China presents TSMC with an impossible dilemma. The mainland is an enormous market and home to major customers like HiSilicon, the chip-design subsidiary of Huawei. But it's also the country that claims sovereignty over Taiwan and has threatened to take the island by force.

In 2017, over the objections of the Taiwanese government, TSMC invested three billion dollars in a manufacturing facility in Nanjing, China. The logic was commercial—serving Chinese customers required a local presence—but the decision was controversial. Critics worried about technology transfer and the strategic implications of building advanced manufacturing capacity in a potential adversary's territory.

Those concerns became acute when the United States, under both the Trump and Biden administrations, began imposing export controls designed to cut China off from advanced semiconductors. In 2020, TSMC announced it would stop supplying Huawei, complying with American restrictions. The decision was painful—Huawei had been a major customer—but TSMC depends on American technology and access to the American market far more than it depends on any single Chinese customer.

In late 2024, the situation grew more complicated when TSMC informed the United States Department of Commerce about a potential breach of export controls. One of its most advanced chips had apparently been sent to Huawei through an intermediary company with ties to the Chinese government. The incident highlighted the difficulty of enforcing export controls in an industry with complex, globe-spanning supply chains.

The Scale of the Operation

TSMC's numbers are staggering. The company has a global capacity of about thirteen million twelve-inch-equivalent wafers per year. Its consolidated revenue in 2020 reached forty-five billion dollars, with net income of nearly eighteen billion—margins that most industries can only dream of. By 2024, the company's market capitalization had reached historic highs, making it one of the most valuable companies on Earth.

This scale required decades of compounding growth. Since 1994, TSMC has achieved a compound annual growth rate of about seventeen percent in revenue and sixteen percent in earnings. That's thirty years of doubling roughly every four to five years, through multiple boom-and-bust cycles in the semiconductor industry.

The company reinvests aggressively. In 2011, TSMC increased research and development spending by nearly forty percent in a single year. Capital expenditures routinely run into the tens of billions of dollars annually. The company has to keep building new fabs and developing new processes just to stay ahead of the competition—and increasingly, there isn't much competition left at the leading edge.

The Succession Question

For three decades, TSMC was synonymous with Morris Chang, its founder and longtime leader. Chang had trained as an electrical engineer at MIT and Stanford, then spent twenty-five years at Texas Instruments before returning to Taiwan to build what would become the world's most important semiconductor company.

His retirement in 2018 marked the end of an era. Mark Liu became chairman and C. C. Wei became chief executive—a dual-leadership structure that reflected the difficulty of replacing a singular figure. In 2024, Liu retired and Wei assumed both roles, consolidating leadership once again.

Whether TSMC can maintain its technological edge without Chang remains one of the great questions hanging over the company. So far, the answer appears to be yes. The company has continued to advance its manufacturing processes, reaching three-nanometer production and announcing plans for two-nanometer mass production in 2025. But the margin for error is slim. A single misstep in technology development could allow competitors to catch up.

Energy, Water, and the Limits of Growth

Semiconductor manufacturing is extraordinarily resource-intensive. TSMC alone accounts for roughly five percent of Taiwan's entire energy consumption—more than the capital city of Taipei. The company uses enormous quantities of ultrapure water, and its fabs generate significant waste that must be carefully managed.

In 2020, TSMC became the first semiconductor company to join the RE100 initiative, pledging to run entirely on renewable energy by 2050. The commitment is genuine—the company has invested heavily in solar and wind power—but it also reflects the reality that Taiwan's power grid is a vulnerability. An island that depends on imports for fuel, and that faces the threat of blockade or bombardment, cannot take energy security for granted.

Water is perhaps an even more pressing concern. Taiwan experiences periodic droughts, and chip fabs require vast quantities of water purified to exceptional standards. During a 2021 drought, TSMC had to truck in water to keep its fabs running. The company has since invested in recycling and efficiency measures, but the fundamental constraint remains.

The Bigger Picture

TSMC's story illuminates something profound about the modern economy: the most critical industries are often the least visible. Few consumers know or care that their devices run on TSMC silicon. The company has no retail presence, no consumer brand, no advertising campaigns. It operates almost entirely in the background of daily life.

And yet its products are everywhere, embedded in the infrastructure of modern existence. The car you drive contains TSMC chips. So do the traffic lights you pass, the data centers that store your photos, the medical devices that monitor your health, the satellites that enable your GPS. Remove TSMC from the equation, and much of the technology we take for granted simply stops working.

This is what makes the company's concentration in Taiwan so consequential. A single point of failure for the entire global technology industry sits on a seismically active island, one hundred miles from a superpower that considers it sovereign territory, in an era of rising great-power competition.

TSMC's overseas expansion is a hedge against this risk, but only a partial one. The company's most advanced capabilities remain on the island, protected by what may be the most elaborate industrial defense in history—not just missiles and soldiers, but the sheer irreplaceability of what TSMC does. If the worst happened, no one else could pick up the slack.

That's the silicon shield. It may be Taiwan's best protection. It may also be a bet that the world simply can't afford to let anything happen to the company that makes everything work.