The City That Made the Microchip

A coastal city in northwestern Taiwan, smaller than Cambridge, Massachusetts, is the place where more than ten per cent of the world’s semiconductors are designed or manufactured. The story of how Hsinchu — a quiet, slightly sleepy university town — became, in less than half a century, the most important single location in the global technology supply chain is, in retrospect, one of the great success stories of late-twentieth-century industrial policy.

From rice paddies to clean rooms: the rise of Hsinchu

Hsinchu is a city of about 450,000 people on the northwest coast of Taiwan, about 80 km southwest of Taipei. For most of its history, it was a quiet provincial town: a fishing port, a university town (home to the prestigious National Tsing Hua University, originally founded in mainland China in 1911 and re-established in Taiwan in 1956), and a hub of Taiwan’s small but growing electronics industry. The city had, in the 1960s and 1970s, a handful of small electronics assembly plants, but no semiconductor industry and no significant industrial cluster. The change came in the 1970s, when the Taiwanese government, alarmed by the success of the Japanese and Korean semiconductor industries, decided to build a domestic chip industry. The instrument was the Industrial Technology Research Institute (ITRI), established in 1973 with a charter to acquire, adapt, and commercialise advanced foreign technology, and to spin out commercial ventures based on that technology.[1, 2, 11]

The choice of Hsinchu as the ITRI site was deliberate. The city was close enough to Taipei to be reachable but far enough to be distinct; it had two strong universities (Tsing Hua and the National Chiao Tung University, which had also been re-established from the mainland in 1958); it had a small electronics industry that could be expanded; and it had available land, including a former military base that the government could repurpose. The ITRI campus, established on the site of the former base, grew rapidly through the 1970s and 1980s, and the surrounding area was designated in 1980 as the Hsinchu Science Park, the first of Taiwan’s science parks and the template for the others. The park offered, to any company willing to set up there, cheap land, generous tax incentives, world-class infrastructure, and proximity to the ITRI and the two universities. The first tenants, in 1980, were a handful of small electronics companies; by 1985, the park was home to UMC, the first ITRI spin-out; by 1987, it was home to TSMC.[1, 2, 12]

The Hsinchu Science Park, in 2024, covers about 6.5 square kilometres in three districts (Hsinchu City, Hsinchu County, and a small extension in Taoyuan). The park hosts about 500 companies, of which the majority are in semiconductors and adjacent fields (chip design, photomask making, packaging, testing, equipment supply). The park’s annual revenue, in 2024, is about NT$1.5 trillion (about $50 billion), with the largest single contributor being TSMC. The park employs about 150,000 people, of whom a substantial fraction are engineers and scientists. The park is, by some distance, the most successful science park in the world: no other dedicated industrial cluster has produced so many billion-dollar companies, and no other has had comparable impact on the global technology supply chain.[1, 3, 13]

ITRI: the foundation

The Industrial Technology Research Institute (ITRI) was the seedbed of the Hsinchu semiconductor industry, and the Taiwanese chip industry is, in significant part, a story of ITRI’s spin-outs. The institute was established in 1973 with a budget of about $8 million (a significant sum for a country with a per-capita GDP of about $500) and a charter to acquire technology from abroad, adapt it to Taiwanese conditions, and spin out commercial ventures. The first major technology transfer was the 1976 RCA CMOS process — a 7-micrometre CMOS technology that ITRI engineers learned on the job at RCA’s research lab in the United States and then set up as a pilot line at the ITRI Hsinchu campus. The pilot line produced, in 1977, the first Taiwanese-made integrated circuit, in 1979 the first Taiwanese CMOS chip, and in 1982 the first Taiwanese ASIC. The Hsinchu pilot line was, in effect, the MIT of Taiwanese semiconductors: a place where a generation of engineers learned the trade, and from which they spun out to found the major Taiwanese chip companies.[2, 4, 14]

The first spin-out was United Microelectronics Corporation (UMC), established in 1980 by a group of ITRI engineers led by Robert Tsao, with backing from the Taiwanese government. UMC built its first fab in the Hsinchu Science Park (which had just been established) and was, for most of the 1980s, Taiwan’s only semiconductor company. The second spin-out, and the more important one, was Taiwan Semiconductor Manufacturing Company (TSMC), established in 1987 by Morris Chang, who had joined ITRI in 1985 as head of the Electronics Division and was given the charter to build a world-class semiconductor foundry. The TSMC business model — purely contract manufacturing, no own-brand chips — was a gamble, since the existing chip companies (Intel, IBM, NEC, Philips, Siemens) all made their own chips, and the assumption that the market would be willing to outsource production to a third party was unproven. The gamble paid off, spectacularly, and TSMC is now, in 2024, the largest and most advanced chipmaker in the world, with a market capitalisation of about $800 billion and customers including Apple, Nvidia, AMD, Qualcomm, and Broadcom.[2, 4, 15]

The ITRI-Hsinchu model has been studied and copied worldwide, but the replication has rarely been successful. The model’s distinctive elements — the combination of a strong research institute, a dedicated science park, generous government support, and a culture of long-term investment in process technology — are difficult to put in place quickly. The most successful replicas, in the form of the Hsinchu model, are in ITRI International‘s partnerships in India, the Philippines, and other countries, but none of these has, as of 2024, produced a chip company of TSMC’s scale. The reason is, in significant part, the culture: the Taiwanese semiconductor industry has a long memory, and the relationships between the engineers and the managers and the customers are deep and durable in a way that cannot be replicated by a one-off technology transfer. The Hsinchu ecosystem is, in this sense, a cultural artefact as much as a technological one.[4, 5, 16]

The Hsinchu Science Park ecosystem

The Hsinchu Science Park is, by any measure, the most successful industrial cluster in the world. The park hosts, in 6.5 square kilometres, about 500 companies, of which a majority are in semiconductors and adjacent fields. The largest single tenant is TSMC, whose Hsinchu fabs (Fab 12, Fab 14, Fab 15, Fab 18) produce the bulk of the company’s leading-edge chips. The second-largest is UMC, whose Hsinchu fabs (Fab 8A, Fab 8B, Fab 8C, Fab 8D, Fab 8F) produce mature-node chips for a wide range of customers. The third-largest is MediaTek, the world’s largest smartphone-chip designer, whose headquarters are in the park. Other major tenants include Realtek (a major fabless chip company), AU Optronics (a major display-panel maker), and a long list of smaller chip-design houses, equipment suppliers, and materials suppliers.[1, 3, 17]

The ecosystem is more than a collection of companies; it is a self-reinforcing network of relationships. The chip-design houses (MediaTek, Realtek, and a hundred smaller firms) need access to TSMC’s leading-edge fabs, and the fact that both are in the same park makes the collaboration close and efficient. The equipment suppliers (ASML’s local service team, Tokyo Electron’s local office, Applied Materials’ local service) need to be close to the fabs they support. The materials suppliers (Shin-Etsu Chemical, SUMCO, JSR, Tokyo Ohka Kogyo) need to be close to the fabs they supply. The university researchers at Tsing Hua and Chiao Tung need to be close to the industry that employs their graduates. The result is a network of relationships that is, in many ways, more valuable than any single piece of technology: a Chinese fab in Shanghai or a Korean fab in Pyeongtaek can buy the same equipment and the same materials, but cannot easily replicate the relationships.[3, 5, 17]

The ecosystem is, also, the product of forty years of cumulative investment. The Hsinchu engineers who, in the 1970s, learned the RCA CMOS process are now in their seventies, but their students, and their students’ students, are in their thirties and forties, and the cumulative expertise is in the heads and hands of three generations of engineers. The Taiwan semiconductor industry employs about 600,000 people, of whom perhaps 150,000 are in the Hsinchu Science Park. The pool of engineering talent is, by some measures, larger than the combined pools of the United States and the European Union, and the industry is able to attract the best graduates of the top Taiwanese universities. The cultural and demographic advantage is, in this sense, a real and durable advantage, and it is one that the new fabs in Arizona, Kumamoto, and Dresden will not easily replicate.[4, 5, 18]

Geopolitical stakes: the most concentrated vulnerability

The success of Hsinchu is, in 2024, also a strategic vulnerability. The leading edge of the chip industry (the 3 nm and 2 nm nodes) is concentrated in Hsinchu, and the most advanced lithography machines (EUV scanners from ASML, made in Veldhoven in the Netherlands) are used almost exclusively in Hsinchu fabs. A disruption to the supply of Taiwanese chips — whether from a natural disaster, a geopolitical event, or a strategic blockade — would have cascading effects on every sector of the global economy. The Hsinchu story is, in 2024, also the story of a single point of failure in the global technology supply chain.[7, 8, 19]

The natural-disaster risk is real. Taiwan sits on an active seismic zone, and the 1999 Chi-Chi earthquake (magnitude 7.7) caused widespread damage to the Taiwanese semiconductor industry, including the shutdown of several fabs for months. The Hsinchu fabs are designed to withstand earthquakes, and the industry has extensive business-continuity plans, but a sufficiently large earthquake (magnitude 8 or above) within 50 km of the park would, in the worst case, take several of the leading-edge fabs offline for six months or more. The chip shortage that would result would be measured in hundreds of billions of dollars.[7, 19]

The geopolitical risk is even larger. The Taiwan Strait is one of the world’s most contested flashpoints, and the Chinese government has not ruled out the use of force to compel unification. The leading-edge chip industry is now, in effect, a hostage to the cross-strait relationship: any military action against Taiwan would, in addition to its direct human and economic costs, take the world’s most advanced chipmaking capacity offline, with cascading effects on every sector of the global economy. The strategic planners in Washington, Brussels, Tokyo, and Beijing are, in 2024, well aware of this fact, and the United States, the European Union, and Japan are running coordinated industrial policies to reduce their dependence on Taiwanese semiconductors. The race is now on to build fabs outside Taiwan, and the Hsinchu ecosystem is, in effect, the model that the rest of the world is trying to replicate (or to escape from).[8, 9, 20]

The next chapter: the Hsinchu diaspora

TSMC, in response to the geopolitical risk, is now building a diaspora of fabs outside Taiwan. The first new fab, in Phoenix, Arizona, is being built with US government subsidies under the CHIPS and Science Act of 2022, and is expected to begin production in 2025. The second, in Kumamoto, Japan, is being built with Japanese government subsidies and is expected to begin production in 2024–2025. The third, in Dresden, Germany, is being built with German and EU subsidies and is expected to begin production in 2027. The total announced capacity, in the three new fabs, is about 200,000 wafers per month, or about 15 per cent of TSMC’s current capacity. The fabs will produce mature-node chips initially, with the most advanced nodes (3 nm and below) remaining in Hsinchu for the foreseeable future.[7, 8, 20]

The diaspora is, in significant part, a hedge against the geopolitical risk, and a real one. The new fabs will, in time, produce enough chips to meet a substantial fraction of the world’s demand, even in a worst-case scenario in which Hsinchu is offline. But the diaspora is not, and cannot be, a full replication of the Hsinchu ecosystem. The relationships between the fabs and the equipment suppliers and the materials suppliers and the design houses take decades to build, and the new fabs will, for the foreseeable future, be satellites of the Hsinchu mothership, not independent centres of gravity. The most advanced process technology will, for the next decade at least, continue to be developed and first produced in Hsinchu, and the new fabs will be licensed to use the latest Hsinchu-developed processes. The Hsinchu story is, in this sense, not ending; it is being globalised.[7, 8, 21]

The Hsinchu story is, in the end, a story about the power of place. The location of a technology cluster matters: the relationships, the culture, the accumulated expertise, the local universities, the local suppliers — all of these are tied to a specific geography, and the geography is, in turn, the product of decades of cumulative investment. Silicon Valley, the Route 128 corridor, Cambridge, Shenzhen, Bangalore, Dresden — all of these are, in different ways, products of the same kind of cumulative investment. Hsinchu is, by most measures, the most successful example of the model. The city that, in 1970, was a quiet Taiwanese university town has, in 54 years, become the most important single location in the global technology supply chain. The story is not finished. The next chapter is being written now, in the new fabs in Arizona, Japan, and Germany, and in the ongoing investments in Hsinchu itself. The Hsinchu model is being tested: can it be replicated elsewhere, and can it survive the geopolitical pressures of the next decade? The answer, in 2024, is “partly” — the new fabs are real, but the Hsinchu ecosystem is not easily transferable, and the most advanced chips will continue to be made in Hsinchu for the foreseeable future. The city that made the microchip will, for the next decade at least, continue to make the most advanced microchips in the world.[1, 7, 8]

Comparison: Hsinchu and the other great tech clusters (2024)

Cluster	Location	Founded	Anchor institutions	Headline products (2024)	Approx. annual revenue
Hsinchu Science Park	Hsinchu, Taiwan	1980	TSMC, UMC, MediaTek, ITRI, Tsing Hua, Chiao Tung	Leading-edge logic chips (3 nm, 2 nm), smartphone SoCs, display panels	~$50 billion
Silicon Valley	California, USA	1950s	Stanford, Berkeley, Intel, Nvidia, Apple, Google, Meta	Software, AI chips, fabless chip design	~$1 trillion+
Shenzhen	Guangdong, China	1980s	Huawei, Tencent (nearby), BYD, DJI	Telecom equipment, EVs, consumer electronics, mature-node chips	~$300 billion+
Hokkaido / Tohoku	Japan	1980s	TSMC (Kumamoto), Rapidus (Hokkaido)	Logic chips, image sensors, MEMS	~$30 billion
Brainport Eindhoven	Eindhoven, Netherlands	1990s	ASML, Philips, NXP, Signify	Lithography machines, automotive chips, lighting	~$80 billion

Key numbers

• ~450,000: population of Hsinchu city (2024).
• ~500: companies in the Hsinchu Science Park (2024).
• ~150,000: people employed in the Hsinchu Science Park (2024).
• ~NT$1.5 trillion (~$50 billion): annual revenue of the Hsinchu Science Park (2024).
• ~10 percent: share of the world’s semiconductors designed or manufactured in Hsinchu (2024).
• 1973: year of ITRI’s founding.
• 1976: year of the ITRI-RCA CMOS technology transfer.
• 1977: year of the first Taiwanese-made integrated circuit.
• 1980: year of the Hsinchu Science Park’s founding and of UMC’s incorporation.
• 1987: year of TSMC’s founding by Morris Chang.

References

1. Wikipedia, “Hsinchu Science Park.” (Standard reference on the Hsinchu Science Park, with sections on the founding and the anchor companies.) https://en.wikipedia.org/wiki/Hsinchu_Science_Park
2. Wikipedia, “TSMC.” (Standard reference on TSMC, with a section on the founding and the relationship with ITRI.) https://en.wikipedia.org/wiki/TSMC
3. Wikipedia, “Hsinchu.” (Standard reference on the city of Hsinchu.) https://en.wikipedia.org/wiki/Hsinchu
4. Wikipedia, “Industrial Technology Research Institute.” (Standard reference on ITRI, the Taiwanese research institute that founded the Hsinchu semiconductor industry.) https://en.wikipedia.org/wiki/Industrial_Technology_Research_Institute
5. Wikipedia, “Science park.” (Standard reference on the science-park model, with a section on the Hsinchu model.) https://en.wikipedia.org/wiki/Science_park
6. Wikipedia, “Morris Chang.” (Standard reference on Morris Chang, founder of TSMC, with a section on the ITRI years.) https://en.wikipedia.org/wiki/Morris_Chang
7. Wikipedia, “United States export controls on semiconductor equipment to China.” (Standard reference on the US export controls imposed in 2022 and tightened in 2023 and 2024, which frame the geopolitical stakes of Hsinchu.) https://en.wikipedia.org/wiki/United_States_export_controls_on_semiconductor_equipment_to_China
8. Wikipedia, “TSMC Arizona.” (Standard reference on the TSMC fab under construction in Phoenix, Arizona.) https://en.wikipedia.org/wiki/TSMC_Arizona
9. Wikipedia, “Cross-strait relations.” (Standard reference on the geopolitical relationship between Taiwan and China.) https://en.wikipedia.org/wiki/Cross-strait_relations
10. Wikipedia, “TSMC Kumamoto.” (Standard reference on the TSMC fab under construction in Kumamoto, Japan.) https://en.wikipedia.org/wiki/TSMC_Kumamoto
11. Wikipedia, “National Tsing Hua University.” (Standard reference on the National Tsing Hua University, one of the two anchor universities of the Hsinchu ecosystem.) https://en.wikipedia.org/wiki/National_Tsing_Hua_University
12. Wikipedia, “National Chiao Tung University.” (Standard reference on the National Chiao Tung University, the other anchor university of the Hsinchu ecosystem.) https://en.wikipedia.org/wiki/National_Chiao_Tung_University
13. Wikipedia, “MediaTek.” (Standard reference on MediaTek, the world’s largest smartphone-chip designer, headquartered in the Hsinchu Science Park.) https://en.wikipedia.org/wiki/MediaTek
14. Wikipedia, “Application-specific integrated circuit.” (Standard reference on the ASIC, the first Taiwanese-made chip, produced at ITRI in 1982.) https://en.wikipedia.org/wiki/Application-specific_integrated_circuit
15. Wikipedia, “Foundry (electronics).” (Standard reference on the foundry model, the business model that TSMC pioneered.) https://en.wikipedia.org/wiki/Foundry_(electronics)
16. Wikipedia, “United Microelectronics Corporation.” (Standard reference on UMC, the 1980 ITRI spin-out that preceded TSMC.) https://en.wikipedia.org/wiki/United_Microelectronics_Corporation
17. Wikipedia, “Realtek.” (Standard reference on Realtek, a major fabless chip company headquartered in the Hsinchu Science Park.) https://en.wikipedia.org/wiki/Realtek
18. Wikipedia, “AU Optronics.” (Standard reference on AU Optronics, a major display-panel maker headquartered in the Hsinchu Science Park.) https://en.wikipedia.org/wiki/AU_Optronics
19. Wikipedia, “Chi-Chi earthquake.” (Standard reference on the 1999 Chi-Chi earthquake, which caused widespread damage to the Taiwanese semiconductor industry.) https://en.wikipedia.org/wiki/Chi-Chi_earthquake
20. Wikipedia, “CHIPS and Science Act.” (Standard reference on the US CHIPS and Science Act, the legislative framework for the US effort to reduce dependence on Taiwanese semiconductors.) https://en.wikipedia.org/wiki/CHIPS_and_Science_Act
21. Wikipedia, “Taiwan Semiconductor Manufacturing Company.” (Standard reference on TSMC, with a section on the geographic distribution of fabs.) https://en.wikipedia.org/wiki/Taiwan_Semiconductor_Manufacturing_Company