A Company You’ve Never Heard Of That Runs the Modern World
Somewhere in the Hsinchu Science Park in northwestern Taiwan, inside a building that looks unremarkable from the outside, workers in full-body cleanroom suits tend to machines that perform one of the most precise manufacturing operations in human history. They are etching circuits onto silicon wafers at scales measured in nanometers, a nanometer being roughly one hundred-thousandth the width of a human hair. The chips produced in that building will eventually end up inside smartphones, data centers, fighter jets, medical devices, and cars. The company doing this work is Taiwan Semiconductor Manufacturing Company, known almost universally as TSMC, and it may be the single most strategically important industrial enterprise on Earth.
That claim is not hyperbole. It is the conclusion that governments, economists, and military strategists have been arriving at, often with a note of alarm, for several years. TSMC’s peculiar position in the global technology supply chain is the product of decades of deliberate strategy, enormous capital investment, and a business model that most competitors dismissed and then failed to replicate. Understanding what that model is, how it came to dominate, and what hangs on its continuation is essential for anyone trying to make sense of the geopolitical and financial tensions that have defined the technology sector through the mid-2020s.
What TSMC Actually Does (and Why It’s Hard)
To understand TSMC, it helps to understand a division that defines the semiconductor industry: the difference between designing chips and manufacturing them. Companies like Apple, Nvidia, AMD, and Qualcomm design semiconductors. They employ engineers who spend years creating the blueprints for processors. But most of these companies own no fabrication facilities of their own. They are, in industry language, “fabless.” They hand their designs to a foundry, and the foundry builds the chips for them. TSMC is, by a significant margin, the world’s leading independent foundry.
This model sounds simple. In practice, it is extraordinarily difficult. Building a modern semiconductor fabrication plant, called a fab, costs somewhere in the range of tens of billions of dollars and takes years to construct and certify. The equipment inside a leading-edge fab includes extreme ultraviolet lithography machines, tools so complex that only one company in the world, the Dutch firm ASML, produces them. A single such machine can cost more than 150 million dollars. Operating a fab at the technological frontier requires not just capital but accumulated knowledge: the proprietary processes, the yield optimization techniques, the material science expertise that TSMC has built up over nearly four decades of continuous operation.
TSMC was founded in 1987 by Morris Chang, a Texas Instruments veteran who had spent decades in the American semiconductor industry. Chang’s insight was that the future belonged to specialization. If a company focused exclusively on manufacturing, with no competing chip designs of its own, it could serve any designer without being a commercial rival to them. That neutrality turned out to be as valuable as the technical capability. Intel, which kept manufacturing in-house for decades and competed directly with many of its customers in certain market segments, could never offer the same conflict-free relationship. TSMC could promise Apple that its chip designs would never be shared with Samsung. That promise was worth billions in business.
The Numbers That Explain the Dominance
By the early 2020s, TSMC’s market position had become remarkable by almost any measure. The company’s revenue grew substantially through the decade. In 2023, TSMC reported annual revenue of approximately 69 billion US dollars, a figure that represented a slight decline from a record year in 2022 as the broader semiconductor market cooled, but still reflected the company’s commanding position. Revenue has since surged with AI demand, reaching roughly 122 billion US dollars in 2025, and TrendForce put TSMC’s share of the global foundry market at a record 72.5 percent in the second quarter of 2026, far ahead of Samsung at 5.9 percent. Its gross margins consistently ran above 50 percent, extraordinary for a manufacturing business.
More telling than the revenue figures is the market share in the most advanced segment of the industry. When analysts and industry researchers have examined production of chips built at the most cutting-edge process nodes, the share attributable to TSMC has consistently been estimated at a large majority of global capacity. Its customers include some of the most valuable companies in the world: Apple has relied on TSMC for its custom silicon chips for years, and Nvidia’s AI accelerator chips, which became central to the artificial intelligence infrastructure buildout, are manufactured at TSMC’s facilities.
That last point matters enormously in the context of the AI investment surge that accelerated in 2023 and 2024. As technology companies poured capital into building the data center infrastructure needed to train and run large language models and other AI systems, demand for the most advanced chips climbed sharply. Nvidia’s H100 and subsequent accelerator chips, all manufactured by TSMC, became extraordinarily sought after. The backlog and constrained supply of these chips shaped which companies could move fastest on AI development. TSMC sat at the chokepoint of that entire buildout.
The Taiwan Problem and the Geopolitical Stakes
It is impossible to discuss TSMC without discussing its location. The company’s most advanced manufacturing is concentrated in Taiwan, a self-governing island democracy that the People’s Republic of China claims as its territory and has never renounced the use of force to bring under its control. This creates what defense analysts and economic security researchers have described as a profound structural vulnerability in the global economy.
The logic runs like this: if a military conflict were to disrupt Taiwan’s semiconductor production, the consequences for the global economy would be severe and would cascade across virtually every industry that depends on advanced electronics. A widely cited analysis by researchers at the Rhodium Group estimated that a hypothetical one-year disruption to Taiwan’s chip production could cost the global economy hundreds of billions of dollars annually. The automotive, consumer electronics, defense, and telecommunications sectors would all face production shutdowns. No alternative source of equivalent manufacturing capacity exists at the technological frontier.
TSMC itself has sometimes been described by Taiwanese security thinkers as a “silicon shield,” the argument being that Taiwan’s indispensability to the global technology supply chain makes any military action against it unthinkably costly for the aggressor and the global economy alike. Critics of this theory argue that it may be less deterrent than it appears, and that a determined adversary might calculate that controlling TSMC’s facilities would be valuable rather than simply destructive. The debate continues among security scholars without clear resolution.
What is undeniable is that this concentration of risk has prompted significant policy responses. The United States passed the CHIPS and Science Act in 2022, which allocated roughly 52 billion dollars to incentivize semiconductor manufacturing within the United States, with the explicit goal of reducing dependence on Asian production. TSMC itself announced plans to build fabrication plants in Arizona, with the first fab, built on an N4 process, entering volume production in late 2024. The company has since raised its planned US commitment to 265 billion dollars, including additional Arizona fabs aimed at more advanced nodes. The European Union launched its own chips act with similar supply chain diversification goals. Japan secured an agreement for TSMC to build a fab in Kumamoto, which began production in late 2024 on mature process nodes.
These efforts represent a genuine industrial policy shift of a kind that many Western economies had not attempted in decades. Whether they will succeed in meaningfully redistributing chip manufacturing capacity is a question that will take many years to answer. Building a new fab is one thing; replicating the deep ecosystem of suppliers, engineers, and process knowledge that surrounds TSMC’s operations in Taiwan is another. Industry analysts have been skeptical that geographic diversification alone can close the expertise gap within any near-term timeframe.
Why Competitors Have Struggled to Catch Up
Intel spent years attempting to reclaim manufacturing leadership after falling behind TSMC at the leading edge of process technology. Its foundry ambitions, which it pursued aggressively under CEO Pat Gelsinger after his return in 2021, reflected a recognition that the company needed to both modernize its own production and attract outside customers to its manufacturing services. Intel’s efforts involved enormous capital commitments and significant operational challenges, and progress was slower and more difficult than the company’s public timelines had suggested. Gelsinger departed in December 2024, and Lip-Bu Tan, who became CEO in 2025, has continued the turnaround. As of 2026, Intel says its 18A process is ramping and is courting outside customers for 18A and the later 14A node, though few outside customers had been publicly confirmed beyond Tesla, which Elon Musk has said plans to use 14A.
Samsung, the South Korean conglomerate, operates an advanced foundry business and has long been TSMC’s most direct competitor for leading-edge manufacturing. It has the capital, the technical talent, and the manufacturing infrastructure to compete at the frontier. Yet TSMC has consistently maintained a lead in yield rates (the percentage of chips on a wafer that function correctly) and in customer confidence. For a fabless chip designer, yield rates are financially decisive: a few percentage points of difference in yield translates directly into cost and profitability. TSMC’s reputation for reliability and execution has proven sticky.
The reasons for TSMC’s sustained advantage are partly structural and partly cultural. The company built its entire identity around serving customers rather than competing with them. It invested relentlessly in process engineering rather than diversifying into chip design or other businesses. And it accumulated what amounts to an institutional memory of how to solve manufacturing problems at scale, embedded in thousands of engineers who have spent careers at the company. This kind of knowledge is not easily transferred or replicated by policy decree.
The Financial Implications for Investors and Economies
For investors, TSMC represents a canonical case of a business with structural competitive advantages in a market with enormous and growing demand. The company’s stock, traded both on the Taiwan Stock Exchange and as American Depositary Receipts in the United States, became a significant holding for technology-focused funds and emerged as a component of major indexes. Warren Buffett’s Berkshire Hathaway disclosed a substantial position in TSMC in late 2022, then disclosed that it had largely sold the position in early 2023, with Buffett citing geopolitical concerns about Taiwan as a factor in the decision. That sequence, a major institutional investor recognizing the company’s quality and then retreating because of location risk, illustrates the analytical tension TSMC presents to any investor.
The capital intensity of the business is also worth emphasizing. TSMC regularly spends 30 to 40 percent of its revenue on capital expenditures, one of the highest reinvestment rates of any large company in any industry. In 2022, the company announced plans for capital expenditure of tens of billions of dollars for that year alone. This is both a barrier to competition and a constraint on free cash flow. The business requires constant, enormous investment just to stay at the frontier.
For national economies, the TSMC question has become inseparable from industrial strategy. Taiwan’s economy is deeply intertwined with TSMC’s fortunes: the company represents a significant portion of the island’s stock market capitalization and exports. For the United States, the CHIPS Act investment reflects a bipartisan consensus, unusual in contemporary American politics, that allowing a critical technology to be produced almost entirely outside the country created an unacceptable strategic dependency. For China, which has poured vast resources into its own domestic semiconductor industry through state-supported programs, the gap to TSMC’s leading edge remained wide as of the mid-2020s, with domestic Chinese firms still struggling to manufacture at process nodes that TSMC had mastered years earlier.
Looking Forward: Concentration, Diversification, and Irreversibility
The world is spending enormous sums trying to reduce its dependence on a single company in a single location. That effort is real, and it will produce results over time. TSMC began volume production of its 2nm process in Taiwan in late 2025. Its fabs in Arizona and Japan represent genuine steps toward geographic diversification, even if they are not yet producing chips at the same cutting-edge process nodes as TSMC’s facilities in Taiwan.
But there are reasons to think that TSMC’s centrality will persist for longer than policymakers hope. The company is not standing still. It continues to invest in the next generation of process technology. The workforce and institutional knowledge concentrated around its Hsinchu and Tainan facilities represent decades of accumulation that cannot be quickly replicated elsewhere. And the economics of semiconductor manufacturing favor scale and concentration: a fab that runs at high utilization with a large and diverse customer base is more efficient than a smaller, newer facility still building its customer relationships.
The deeper point is about the nature of modern supply chains. Advanced manufacturing has always tended toward concentration in clusters where knowledge, suppliers, and talent reinforce each other. Silicon Valley itself is an example of this dynamic in chip design. Taiwan became the equivalent cluster for chip fabrication, through a combination of deliberate policy, early investment, extraordinary execution, and the compounding advantages that come from being first. Undoing that concentration is not impossible, but it is slow and expensive and requires sustained political will that has historically been difficult for democracies to maintain across election cycles.
TSMC sits at an intersection of technology, finance, geopolitics, and industrial policy that is genuinely unusual in modern economic history. It is a private company whose operations have become a matter of national security for multiple great powers simultaneously. Its continued performance is embedded in the investment theses of sovereign wealth funds, pension managers, and retail investors on six continents. And the chips it produces are inside the devices you are almost certainly using to read this article. The world built a profound dependency on a remarkable company in a complicated location, and the work of figuring out what to do about that is now one of the defining economic and strategic challenges of the decade.