TSMC: The Most Important Company You Have Never Thought About

When Nvidia became, in July 2025, the most valuable company in the history of the world, crossing four trillion dollars on the back of the artificial-intelligence boom, the achievement was rightly celebrated as Nvidia’s. But Nvidia does not make a single one of the chips that made it the most valuable company on Earth. It designs them. The actual manufacturing, the almost unimaginably precise act of carving billions of transistors into slivers of silicon, happens somewhere else, in the fabrication plants of a company headquartered in the Taiwanese city of Hsinchu, a company most consumers have never knowingly bought anything from in their lives.

That company is TSMC, and it occupies the strangest and arguably the most powerful position in all of modern industry. It is the invisible foundation beneath the visible giants. Apple’s chips, Nvidia’s chips, AMD’s chips, Qualcomm’s chips, the silicon brains of the entire digital age, are overwhelmingly manufactured by TSMC. And the deepest part of the story is that TSMC reached this position not by inventing a famous product, but by inventing a famous business model: the decision to make nothing of its own, and to become, instead, the trusted factory for everyone else.

The World Before TSMC

To understand why TSMC was revolutionary, you have to understand the brutal economics of making chips before it existed.

For the first decades of the semiconductor industry, the rule was absolute: if you designed chips, you also had to manufacture them. The two were inseparable, and the model had a name, the integrated device manufacturer, or IDM, practiced by giants like Intel. The problem was that building and running a leading-edge fabrication plant, a “fab,” is one of the most capital-intensive and technically punishing endeavors in all of industry. A single advanced fab can cost many billions of dollars and must be rebuilt, essentially from scratch, every few years as the technology advances. This meant that to compete in chips at all, a company needed not only brilliant designers but also the colossal capital and manufacturing expertise to run fabs at the bleeding edge.

The consequence was suffocating. A brilliant engineer with a revolutionary chip design but no fab and no billions could not enter the industry. The cost of the factory was a wall that kept out almost everyone. The world was full of potential chip designers who could never become chip companies, because they could not afford to also become manufacturers. That wall is the thing TSMC tore down.

The Founder

TSMC’s founder is Morris Chang, and his story is unusual among great company builders in almost every way.

Chang was not a young upstart. He was fifty-six years old in 1987, a veteran of decades at the American semiconductor pioneer Texas Instruments, when the government of Taiwan recruited him to help build a semiconductor industry for the island. He had spent his career inside the IDM model and understood, from the inside, both its power and its crushing cost. And he saw something almost no one else did: that the rising complexity and expense of manufacturing was about to make it impossible for most chip designers to also be manufacturers, and that this created room for a company that would do nothing but manufacture, brilliantly, for everyone.

The idea was radical to the point of seeming absurd. A “pure-play foundry” would build chips to other companies’ designs and would itself design nothing, sell no branded product, and crucially, never compete with the very customers it served. At the time, the notion that a serious company would hand its most precious manufacturing secrets to an outside party was close to unthinkable; manufacturing was considered core, something you never outsourced. Chang spent the early years visiting potential customers one by one, slowly building trust in a model the industry did not believe in. A pivotal validation came in 1988, when TSMC passed the rigorous certification of Intel, then led by Andy Grove, proving a foundry could meet the standards of the most demanding chipmaker in the world.

There is a detail about Chang that captures the character of the enterprise. As founder he took essentially no founding equity; the company was funded largely by the Taiwanese government and the Dutch firm Philips. The fortune he eventually built came from buying TSMC shares over the years with his own salary. He was less a wealth-seeking entrepreneur than an architect building a national and industrial institution, and that institutional, mission-driven character runs through TSMC to this day.

The Inflection: The Fabless Revolution

Once a few brave customers proved the foundry model worked, it set off one of the most consequential chain reactions in business history: the birth of the “fabless” semiconductor company.

If a trusted foundry like TSMC would handle the manufacturing, then a chip company no longer needed to own a fab at all. It could be “fabless,” design only, and simply send its designs to TSMC to be built. Suddenly the wall that had kept out everyone but the giants collapsed. A company with brilliant designers and no factory could now compete with Intel, because it could rent the world’s best factory by the wafer. This unleashed an explosion of innovation: Nvidia, Qualcomm, Broadcom, and eventually a huge share of the most important chip designers on Earth were built on the foundation that TSMC made possible. They could exist because TSMC existed.

And this created a flywheel of extraordinary power. The more fabless companies relied on TSMC, the more volume flowed through its fabs. The more volume, the more money TSMC could pour into building the next generation of even more advanced fabs. The more advanced its fabs, the more indispensable it became to customers who needed the cutting edge, which brought still more volume, which funded still more advanced fabs. Manufacturing leadership, once achieved, fed on itself. TSMC crossed fifty percent of the global foundry market around the turn of the millennium and never looked back, marching down the relentless ladder of miniaturization, from microns to nanometers, eventually becoming the first foundry to manufacture chips at the 7-nanometer, 5-nanometer, and 3-nanometer scales in high volume.

The Moat

TSMC’s moat is one of the deepest illustrations anywhere of an advantage built from scale, capital, and accumulated expertise compounding over decades into something a rival simply cannot buy or build quickly.

Consider what a competitor would need to challenge TSMC at the leading edge. It would need to spend tens of billions of dollars on fabs. It would need to acquire the almost unimaginably complex manufacturing know-how, the thousands of finely tuned process steps, the accumulated learning from trillions of transistors, that TSMC has built over more than thirty years and that exists nowhere in complete written form. It would need the specialized equipment (above all the extreme-ultraviolet lithography machines from a single supplier, ASML), the ecosystem of materials suppliers, the trained engineers, and the institutional culture of relentless manufacturing excellence. And even with unlimited money, it would need years, because much of the knowledge is learned, not bought. This is why even nations spending vast sums to build domestic chip capacity find that catching TSMC is, as Chang himself bluntly put it, extraordinarily difficult, a thirty-year compounding advantage of talent, culture, and ecosystem.

On top of the manufacturing lead sits the trust moat, the original one. Because TSMC designs nothing and competes with no customer, every fabless company can hand it their crown-jewel designs without fear that the foundry will steal them and compete. A would-be rival that also designs chips can never offer that pure neutrality. TSMC’s promise to make nothing of its own is not a limitation; it is the foundation of the trust that makes it the universal manufacturer.

And there is a final, geopolitical layer, the so-called “silicon shield.” Because so much of the world’s most advanced chip manufacturing sits in Taiwan, TSMC has become an object of global strategic significance, its fabs arguably among the most important physical assets on the planet. This is both a source of its mystique and the central risk to its future.

The Wealth Created

TSMC’s investor story is a compounding story rather than a crash-and-recovery one, but with its own distinct lesson: the immense, durable value of being the indispensable supplier rather than the famous brand.

An investor who recognized, over the past few decades, that TSMC was quietly becoming the irreplaceable factory beneath the entire digital economy, and who held through the chip cycles and the geopolitical scares, would have owned one of the great compounding industrial franchises of the modern era, its revenue and market value climbing with every wave, PCs, then mobile, then artificial intelligence, as the demand for ever-more-advanced chips grew without limit. By the mid-2020s TSMC’s market value reached well over a trillion dollars, with the overwhelming majority of its revenue coming from the most advanced nodes that it alone could produce at scale.

The lesson is one of the most valuable, and most overlooked, in all of investing: the famous company that captures the public imagination is not always the best place to be. Often the most durable, defensible position is one layer down, in the indispensable supplier that every famous company depends on, that benefits from all of their competition with one another, and that cannot be easily replaced. TSMC wins no matter which chip designer is ascendant, because they all need its fabs. Owning the factory beneath the gold rush can be better than owning any single miner.

What Everyone Got Wrong

Mistake #1: Believing manufacturing was a low-value commodity.
Reality: for years the glory and the perceived value in chips went to the designers, while manufacturing was seen as grunt work to be commoditized. TSMC proved the opposite, that at the leading edge, manufacturing is the scarcest, hardest, and most defensible capability of all.

Mistake #2: Assuming the foundry model would never be trusted.
Reality: the industry initially could not imagine handing manufacturing to an outsider. TSMC’s neutrality, never competing with customers, turned that supposed weakness into the deepest trust moat in technology.

Mistake #3: Thinking TSMC’s lead could be bought or rebuilt quickly.
Reality: governments and rivals have repeatedly assumed that enough money could replicate TSMC. The accumulated, three-decade compounding of process knowledge, ecosystem, and culture has proven far harder to copy than capital alone can solve.

The Alternative Timeline

A counterfactual, clearly hypothetical.

Picture the world where Morris Chang never returns to Taiwan, or where the pure-play foundry model fails to win trust and dies in its first decade.

In that timeline, the wall stays up. Making chips remains the exclusive province of those who can afford to own fabs, the integrated giants like Intel. The fabless revolution never happens, or happens far more slowly and partially. Nvidia, founded by designers with no fab, struggles to exist; Qualcomm and a thousand other fabless innovators are stillborn or stunted. The explosion of specialized chip design that powered the smartphone and then the AI era is throttled, because the manufacturing wall keeps out the very designers who would have built the future. The pace of the entire digital age slows, and the AI boom of the 2020s, which runs on chips that fabless companies designed and TSMC built, arrives late, or in some diminished form.

It did not happen that way, because a fifty-six-year-old veteran saw that making nothing of your own could be the most powerful position of all, and spent years convincing a disbelieving industry to trust him. The lesson is that some of the most world-changing businesses are not the ones with the famous product, but the ones that quietly remove a barrier and let a thousand others build. TSMC’s genius was to become the road on which the entire fabless industry could travel.

Why This Matters to Investors

TSMC fits the thesis cleanly. The improbable idea: a company that designs nothing and manufactures everything, trusted by competitors who hand it their most precious secrets. The stretch where failure looked inevitable: the early years when the industry did not believe a neutral foundry could work, and Chang went door to door building trust in a model no one wanted. And the point where success looks obvious: the indispensable keystone of the entire digital economy, the maker of nearly every advanced chip on Earth. The opportunity, as always, lived in the long middle, when the foundry model was an unproven gamble rather than the foundation of the modern world.

The reason to study TSMC is that it teaches one of the most powerful and least appreciated ideas in investing: indispensability beats fame. The asymmetry that makes legendary companies legendary is, in TSMC, the asymmetry of being the one supplier that every competitor in a booming industry must use, and that none can replace. The greatest opportunities are often hidden one layer beneath the obvious winners, in the unglamorous, irreplaceable infrastructure they all depend on. TSMC, for decades, was exactly that hidden layer, the most important company almost no one was thinking about.

Lessons in Order of Depth

On the surface — the move

Own the irreplaceable layer, not the famous one. TSMC does not bet on which chip designer wins; it manufactures for all of them and profits from their competition. The trader’s analogue is positioning in the indispensable infrastructure of a trend rather than picking which competitor will capture it.

Below the surface — the Money

The willingness to make the massive, concentrated bet at the moment of maximum opportunity. Chang bet nearly half the company’s cash on the Apple line because the asymmetry favored it. The investor’s version is the discipline to size up dramatically when a rare, high-conviction opportunity appears, rather than spreading thin out of timidity.

Below that — the Mind

Give up the glory to own the foundation. Chang’s deepest insight was that making nothing of your own, refusing to compete with your customers, was not a sacrifice but the source of unbreakable trust and universal demand. The trader’s parallel is the ego discipline to take the unglamorous, structurally superior position rather than the flashy one that feels more like winning.

At the deepest level — the question

TSMC’s power is that everyone depends on it and no one can replace it, an advantage built from decades of compounding expertise that money alone cannot buy. So the deepest question TSMC poses is about the nature of an advantage that compounds in knowledge rather than capital: what can you build, slowly, that becomes more valuable and more irreplaceable the longer you do it, in a way a competitor cannot shortcut with money? TSMC’s moat is thirty years of learning that exists in no manual and can be acquired only by living it. The deepest edge is not a product or a price but accumulated mastery, the kind of compounding competence a rival cannot buy, cannot copy quickly, and cannot catch, because by the time they reach where you were, you have moved further on. That is the rarest moat of all: a lead made of time and learning, widening faster than anyone can close it.

The Legendary Scorecard

Category	Score	Notes
Founder Vision	10 / 10	Saw that making nothing of your own could be the most powerful position
Innovation	10 / 10	Invented the foundry model; first to volume 7nm, 5nm, 3nm, EUV
Execution	10 / 10	Decades of leading the hardest manufacturing on Earth
Moat	10 / 10	Compounding expertise + neutrality + scale; among the widest anywhere
Capital Allocation	9 / 10	Relentless, disciplined reinvestment into ever-advancing fabs
Wealth Creation	9 / 10	A great compounding franchise; >$1T market value
Durability	8 / 10	Technically near-unassailable; the live risk is geopolitical
Historical Importance	10 / 10	Enabled the fabless industry and the entire modern chip economy
Overall Legendary	9.6 / 10	The indispensable keystone of the digital age

Scores are an editorial verdict on the standard eight-category scale used across the Greatest Companies series. The overall is a judgment, not a weighted average.

Company Timeline

• 1985 — Taiwan’s government recruits Morris Chang to build a semiconductor industry
• 1987 — TSMC founded (Feb 21) on the pure-play foundry model; government 48.3%, Philips 27.5%
• 1988 — Passes Intel’s certification under Andy Grove
• 2000 — Surpasses 50% of the global foundry market
• 2008 — Launches the Open Innovation Platform
• 2010s — Node leadership: 28nm, 16nm, 10nm, 7nm
• ~2014 — The all-in Apple bet: a ~$9B crash line built in record time
• 2018–2019 — First to high-volume EUV at 7nm
• 2022 — 3nm mass production begins (late year)
• 2024 — Arizona Fab 21 begins production
• 2025 — 2nm mass production; ~$165B global expansion; market cap ~$1.5T

Key Numbers

Founded	1987 (Hsinchu, Taiwan)
Founder	Morris Chang (ex-Texas Instruments)
The model	Pure-play foundry; designs nothing, never competes with customers
Node leadership	First to volume 7nm, 5nm, 3nm; 2nm in 2025
Advanced-node revenue	~74% of wafer revenue from sub-7nm
TTM revenue (2025)	~$88B
Market value (2025)	~$1.5 trillion

Related Reading

More Greatest Companies

• ASML: The One Machine the Whole World Runs On (the EUV monopoly that makes TSMC’s leading edge possible)
• Nvidia: The 90% Drawdown That Built the AI Age (the fabless designer TSMC manufactures for)
• Apple: Ninety Days From Death to the World’s Most Valuable Company (the customer behind the company-betting Apple line)

Lesson Hubs

• Competitive Moats (the moat of compounding expertise and scale)
• Innovation Cycles (how the foundry model unleashed the fabless industry)

Across the Library

• The AI Boom (Market Mayhem — the demand wave TSMC and ASML sit beneath)

This article is part of the Greatest Companies series, adapted from the book Greatest Companies, now available on Kindle. Explore the wider framework in The Complete Trader’s Edge.

Frequently Asked Questions

What does TSMC actually do?

TSMC is a semiconductor foundry: it manufactures chips designed by other companies. It does not design or sell its own branded chips. Apple, Nvidia, AMD, Qualcomm, and many others send their designs to TSMC, which builds them in its fabrication plants, mostly in Taiwan.

Who founded TSMC and when?

TSMC was founded in 1987 by Morris Chang, a veteran of Texas Instruments recruited by Taiwan’s government. He pioneered the “pure-play foundry” model, a company that only manufactures and never competes with its customers’ designs.

Why is the foundry model such a big deal?

Before TSMC, a chip company had to own hugely expensive factories to compete, which kept out almost everyone. By offering trusted manufacturing for hire, TSMC enabled the “fabless” industry, letting design-only companies like Nvidia and Qualcomm exist and compete without building their own fabs.

What is TSMC’s competitive moat?

Three layers: decades of accumulated manufacturing expertise that money cannot quickly buy; neutrality, since it designs nothing and so can be trusted with customers’ secrets; and the scale to keep funding the most advanced fabs on Earth. Together these have made it extremely difficult for any rival or nation to catch up.

What is the biggest risk to TSMC?

Geopolitics. Because so much of the world’s most advanced chip manufacturing is concentrated in Taiwan, TSMC sits at the center of global strategic tension, sometimes called the “silicon shield.” This concentration is both a source of its importance and the principal risk to its future.