Taiwan Semiconductor Industry–rise of global monopoly from a virtuous cycle

Focus on commercialisation of R&D outputs as product and process features in an emerging industry has led to the uprising of the Taiwan semiconductor industry and economy as a whole

Success of Taiwan semiconductor industry has been due to commercialisation of learning and R&D outputs, leading to forming a virtuous cycle of growth
Success of Taiwan semiconductor industry has been due to commercialisation of learning and R&D outputs, leading to forming a virtuous cycle of growth

The triggering of the chip war underscores the paramount importance of the rise of the Taiwan semiconductor industry as a global monopoly. Its crown jewel TSMC produces more than 90 percent world’s high-end microchips. Furthermore, while USA’s icon Intel reports a loss, TSMC reports above 30% net profit from exponentially growing revenue. But TSMC did not grow in isolation, making the Taiwan Semiconductor industry newsworthy.

In addition to foundry services, Taiwan has been gaining traction in fabless and chip packaging segments of the semiconductor value chain. MediaTek is another enviable success of the Taiwan semiconductor industry. This fabless semiconductor company provides chips for wireless communications, high-definition television, and handheld mobile devices. Due to a steady uprising, MediaTek has reached the 4th position in the global fabless segment of the industry. Furthermore, ASE is another star in the Taiwan Semiconductor industry, offering testing and packaging services. But what is the underpinning of the rise of the Taiwan Semiconductor industry?

Like China or India, Taiwan’s 20+ million people running the agricultural economy did not have a large domestic market. Besides, like USA’s Silicon Valley, Taiwan’s semiconductor industry did not get lucrative orders from the military to gain traction in the early days. On the other hand, Taiwan did not look for low-cost labor advantage to serve MNCs, like Malaysia or South Korea, to make an entry and craft a path of growth. Besides Taiwan Government did not fork out billions of dollars in subsidies to embark on a mega plan for building the Taiwan semiconductor industry. Although Taiwan had a sound schooling system, its strength in higher education and R&D in Science, Technology, Engineering, and Math (STEM) was very modest in the 1970s. Furthermore, its infrastructures in the 1970s were well developed. Besides, it has been under perpetual military and political threat.

A virtuous cycle has been fueling the Taiwan semiconductor industry

Conventional factors were not at all favorable for the development of the Taiwan semiconductor industry. To our surprise, it has happened. It’s due to creating a virtuous effect from the humble beginning. Like the nuclear chain reaction starting from a single splitting of a nucleus, Taiwan succeeded in nurturing an upward spiral of potential. With each success, Taiwan succeeded in garnering more resources which, in turn, allowed this island nation to achieve greater and greater success. Unlike a vicious cycle compounding the negative, a virtuous cycle keeps compounding the positive, making future progress easier and more manageable. Hence, the success of the Taiwan semiconductor industry has been the cumulative effect of incremental advancements creating a reinforcement effect. Instead of a mega plan and mega-budget administered by bureaucrats, the Taiwan semiconductor industry has been growing steadily, culminating in a big bang effect.     

Seeding the journey of Taiwan semiconductor industry through technology transfer

In the early 1960s, Taiwan’s economy was agricultural. To create higher-income jobs, Taiwan pursued an import substitution policy. But due to the realization of the limitation of import substitution, Taiwan moved to export-oriented manufacturing. The value addition of out-of-labor started showing signs of a ceiling of income growth. Hence, Taiwan’s leadership started looking for scalable income growth opportunities. In pursuit of it, in 1973, Taiwan established Industrial Technology Research Institute (ITRI), a technology research and development institution.  

Soon after its establishment, ITRI found semiconductors as an industry for developing a scalable growth path. But ITRI did not pursue the course of developing infrastructure and providing incentives for attracting multinationals to set up labor-centric testing, bonding, and packaging plant. ITRI did not recommend incentives to allure multinational corporations to set up commercial foundries in Taiwan. Instead, ITRI proceeded with the technology transfer agenda. Consequentially, as part of a four-year US$10 million integrated circuit development project, in 1976, ITRI gave a $4 million contract to America’s RCA to transfer CMOS process technology. In addition to setting up the fab, RCA also trained Taiwanese engineers. ITRI sent 19 young engineers to RCA’s facilities for IC design, process technology, IC testing, and semiconductor equipment training. Subsequently, more engineers went on the same pilgrimage to become Taiwan’s industry pioneers.

Former ITRI President Chintay Shih led Taiwan’s semiconductor technology transfer and assimilation endeavor. Just after graduating from Princeton University with a doctorate in electrical engineering and with brief experience working for a computer company in San Diego in 1975, Dr. Shih responded to his nation’s call. He started his mission with the ITRI semiconductor endeavor in 1976 as the leader of the process technology team.

RCA trained future leader of the Taiwan semiconductor industry—seeding the virtuous cycle

In conventional technology transfer projects, technology transferors provide training about how to use the transferred technology. RCA’s engagement was no exception. But Taiwanese engineers did not keep them limited within such conventional scope. Instead, they got into the mission of assimilating existing technologies to improve them further, leading to the development of Taiwan’s semiconductor industry. Notable trainees included Bob Tsao, emeritus chairman of United Microelectronics Corp.; Tsai Ming-kai, chairman of IC design house MediaTek Inc.; and F.C. Tseng, vice chairman of Taiwan Semiconductor Manufacturing (TSMC).

While young Taiwanese engineers were getting training in the USA, as part of the ITRI contract, RCA was building Taiwan’s first 3-inch wafer fabrication plant. It began its pilot run at the end of 1977, and young D. Shih became the first plant manager. This 3-inch wafer fabrication plant, more than a node behind the state-of-the-art, became “the origin of the country’s semiconductor industry.” Ironically, although starting from this RCA transferred technology Taiwan’s TSMC emerged as the global monopoly in high-end processes, RCA became extinct in the semiconductor business.

Instead of focusing on operating RCA’s plant, under the leadership of Dr. Shih, Taiwanese engineers focused on learning, generating ideas, and integrating them to improve yield. Hence, ITRI’s new plant started running at a higher yield within a few years than RCA’s plant operating in the USA. Through the exercise, ITRI seeded the process of generating knowledge and ideas and converting them into better process performance—a core competence of creating a virtuous cycle.  

Spinning off acquired competence for exploiting global commercial opportunities—for leveraging discontinuity  

Unlike India’s SCL, Taiwan focused on leveraging newly acquired semiconductor process technology competence to serve the commercial market. Besides, ITRI targeted the global market because the domestic demand for semiconductors in Taiwan was negligible. But in those days, the microchip business was in the hands of integrated device makers (IDMs) like NEC, Toshiba, IBM, Fairchild, and Intel. Hence, ITRI faced a high entry barrier in creating the market for its wafer processing capability. Therefore, ITRI was compelled to serve the untapped opportunity, which would likely grow.

In the late 1970s, a few small fabless companies started to grow. They used to design application-specific chips (ASICs) as off-the-shelf or customer-specific customized components. But they were facing a barrier to printing those designs on wafers. Hence, the demand for joint or shared fab facilities was growing. Although the overall demand was insignificant, ITRI targeted to serve this market. Consequentially, ITRI got into the venture of spinning off internal wafer processing capability as a commercial venture—UMC. On May 22, 1980, the United Microelectronic Company (UMC) was formally established as Taiwan’s first private integrated circuit company. Through the UMC’s foundry model, ITRI potentially opened the entry of commercializing the knowledge and ideas of wafer processing.

Birth of OSAT cluster in Taiwan to complement UMC’s foundry model

Upon getting the designs printed on the wafer from UMC’s foundry, fabless companies faced the challenge of getting them tested, bonded, and packaged. In those days, IDMs used to have such a capacity as captive facilities in Malaysia, the Philippines, and other low wages countries. Due to the proprietary nature of IDMs, there was a gap in supply and demand. Hence, Advanced Semiconductor Engineering (ASEC) came into existence in 1984 to offer assembling, testing, and packaging services to fabless companies. Consequentially, it gave birth to the outsourced semiconductor assembling and testing service (OSAT) segment of the semiconductor value chain.

OSAT segment generated more than $30 billion in revenue in 2022, the fastest-growing segment. Besides, due to the growing adoption of Chiplet, the demand for OSAT services will likely accelerate. Over the decade, Taiwan has developed a strong foothold in the OSAT segment of the industry. It’s not only that ASEC has the highest market share (30%); among the top 10 OSAT companies, five are from Taiwan. The remaining top four Taiwanese OSAT companies are PTI, KYEC, ChipMOS, and Chipbond, with 8.18%, 3.11%, 2.47%, and 2.39% market share in 2020, respectively. Collectively, these five Taiwanese companies had more than 46% of the global $30 billion OSAT industry.

PCB, ESD, and Assembling for complementing foundry and OSAT

The complementary role of ASEC’s OSAT services to UMC’s wafer processing for printing the microchip design of fabless companies started to form a cluster effect. As a result, it contributed to fueling the virtuous cycle effect. Besides, this dynamic also led to the demand for printed circuit board design, production, and populating. Taiwan also leveraged this segment. Down the value chain for assembling multiple PCBs and other components like displays, Foxconn came out from Taiwan. As we all know, Foxconn has reached a monopolistic situation in the market of contract assemblers of electronic products.

Return of Morris Chang to Taiwan and Joining ITRI

Upon getting a Bachelor’s and Masters’s Degree at MIT in Electrical Engineering, Morris Chang pursued Ph.D in the same institution. But upon failing the qualifying exam, he opted for a career in the industry. Although his career was at Sylvania Semiconductor, he joined Texas Instruments (TI) after three years. Due to his spectacular performance, TI offered him the chance to get Ph.D. with a full salary and tuition coverage. Within two and half years, he earned Ph.D. from Stanford University. Subsequently, he returned to TI; and by the end of 25 years, Dr. Chang rose to the tank of group vice president responsible for TI’s worldwide semiconductor business. After encountering insurmountable barriers to getting promoted to the rank of CEO and subsequently losing interest in what he was doing, he left TI in 1983. And he left TI without any job offer.

Upon screening a few jobs offers, Dr. Chang joined as president and the COO of General Instrument (GI) in New York in 1983. But as opposed to TI’s organic growth model, GI’s business model was around acquiring poor-performing firms, remaking them better, and selling them off at a higher price. Hence, General Instrument did not fit Dr. Chang’s background. Consequentially, after a little more than a year, he left GI without another job prospect.

Upon leaving GI in 1985, Dr. Chang responded to the call of the Premier of Taiwan, who personally appointed the president of ITRI–the largest and the only significant industrial technology research in Taiwan. He was appointed as the Chairman of ITRI to benefit from his ability to transfer technology from just research results to economic benefits for the Taiwan industry. Due to his exposure at TI and GI, Dr. Chang found Premier’s appeal to transfer research results into economic gains encouraging.

Driving the economic growth of Taiwan out of the commercialization of research results

Upon understanding the limitation of import substitution and export-led manufacturing for the commercialization of labor, Taiwan’s leadership realized the importance of trading knowledge and ideas. Yes, research produces knowledge and ideas; but their production alone is not sufficient to drive economic growth. The next challenge is to trade them as product or process features in the globally connected competitive market at a profit. Hence, due to his background in TI and GI to manage R&D and trade generated knowledge and ideas in the market for the economic gains for companies, Dr. Chang found premier’s appeal a good match.

Furthermore, instead of just for the benefit of the company, the job at ITRI asked to drive the economic growth of Taiwan’s industry. Hence, Dr. Chang’s professional interest got rejuvenated. He found the opportunity and challenge of making Taiwan a high-income country out of the production and trading of knowledge and ideas highly appealing. By that time, he was also financially secure to enjoy modest life. Hence, he decided to take the job of chairmanship of ITRI at a pretty low compensation. Therefore, he followed his interests, not where you think big money is.

He found semiconductors as a suitable ground for turning the appeal into reality. As Moore’s law states, transistors in those days were quite amenable to getting better and cheaper due to the flow of knowledge and ideas.

Although Dr. Chang did not chase money, money came to him. It’s not due to ITRI’s salaried job. Instead, his chance of starting up TSMC after coming to Taiwan gave him both money and the achievement of helping Taiwan become a high-income country out of investment in R&D and commercialization of R&D outputs for economic gain.

Birth of TSMC—creating the monopolization success story of the Taiwan Semiconductor Industry

The technology transfer from RCA in 1976 was insufficient to profit from commercialization. First of all, RCA transferred a generation of older technology. The next one is that RCA was not a top-tier semiconductor firm like Intel and TI. Although ITRI engineers kept improving the transferred technology, by the end of 11 years in 1987, they managed to get another generation and a half behind. Falling behind two and a half generations was quite understandable as ITRI started with older technology, and without a real commercial base, ITRI was limited in improving the technology. Hence, ITRI was consistently falling behind in keeping up with the pace of other entities that are commercial companies, like Intel and TI, and so on. Therefore, there was pressure on ITRI to vigorously pursue commercialization to monetize acquired competence and accelerate the advancement to keep pace.

Within a week of his joining as the president of ITRI, Taiwanese leadership asked him about the urgency of starting a new semiconductor company. After some thinking, he zoomed in Carver Mead’s writings about the semiconductor industry’s future, the pure—play foundry’s idea. In those days, Taiwan had no strength in R&D, intellectual property, and IC design, and in marketing and sales; he found pure-play foundry was the only option for Taiwan. But the pure-play foundry model faced a fatal flaw–  “Where is the market?”

Fabless future beaconed Dr. Chang to start a pure-play foundry—giving birth to Taiwan Semiconductor Manufacturing Company (TSMC) in 1987

The virtual nonexistence of fabless industry posed the threat of dismissal of the idea of pursuing pure-play foundry. On the other hand, catering IDMs’ capacity overflow was not an option for a stable market. But Dr. Chang’s observation of the desperation of designers working at TI and other IDMs gave him hope for the future of fabless companies.

Despite the desperation, experienced chip designers could not leave and start their microchip venture due to the barrier of fab—the most capital-intensive segment of the semiconductor industry. For this apparent reason, Dr. Change felt that a pure-play foundry might succeed in offering the future fabless companies fab services. He envisioned that with the availability of pure-play 3rd party foundry services, those designers would successfully form their own fabless companies. Consequentially, they would become customers and constitute a stable and growing market for pure-play foundry.

But in the early days, many fabless companies were not around; they were tiny. Hence, Dr. Chang’s option to start a pure-play foundry was to serve the leftovers of IDMs. Besides, the demand was unstable.

As head of ITRI’s IC development center, Dr. Shih joined then-ITRI President Morris Chang in negotiating a partnership with the Netherlands’ Royal Philip Electronics. It culminated in the establishment of TSMC in 1987. As mentioned, TSMC started the journey with ITRI’s two-and-a-half generation older technology. Between 1987 to 1991, TSMC depended mainly on the big IDMs giving only leftovers and a few small fabless companies, first in Taiwan and a little later in the United States. Therefore, the growth and profitability of TSMC in those early years were meager. But the situation started changing in favor of TSMC due to the exponential growth of fabless semiconductor companies.

Ramping up of TSMC

The early 1990s witnessed mushrooming growth of fabless companies. Most were in the USA, and a few were in Taiwan and Europe. By the late 1990s, fabless companies also started to grow in China. To leverage fabless companies, TSMC had to deal with the trust, technology, and cost of production. There was a severe trust issue in sharing the design with foundry service providers.

On the other hand, TSMC had to meet process node maturity requirements to ensure competitive chip density, following Moore’s law. Furthermore, the cost of low-volume production was also a challenge. With success in dealing with these challenges, TSMC succeeded in exiting loss by 1990 and pursuing profit-making growth.

To deal with challenges, TSMC focused on process IPs and customer trust. Besides, with prudent decisions of technology upgrading, IP and yield management, and Economies of Scale advantage, TSMC kept growing along with the growth of fabless companies. The success of ITRI and TSMC in developing thriving fabless companies in Taiwan also contributed to the steady development of demand for foundry services. Among the top fabless Taiwanese companies are (1) MediaTek, (2) Novatek, (3) Realtek, (4) Phison, (5) Himax, (6) Silicon Motion, (7) ESMT, (8) Stronix, (9) Fitopower. Among them, MediaTek is the leading Taiwanese fabless company with almost $16 billion in revenue in 2021.

Monopolization of high-end market by TSMC

The growth of the mobile handset market gave a big push to TSMC. Notably, the decision of Intel to reject Apple’s request to produce an A1 chip for the iPhone was a blessing. Upon rejection, although Apple went to Samsung, subsequently, Apple has become the largest customer of TSMC. Furthermore, except Samsung, all other mobile handset makers preferred to source processors from fabless companies. Hence, along with the growth of fabless companies, TSMC also kept growing due to the development of mobile processor complexity and the diffusion of smartphones. In 2021, Apple generated as high as 25% of TSMC’s total revenue of $57.22 billion. Among the top 10 in 2021, other TSMC customers are (1) MediaTek (5.8%), (2) AMD (4.36%), (3) Qualcomm (3.9%), (4) Broadcom (3.77%), (5) Nvidia (2.83%), (6) Sony (2.54%), (7) Marvell (1.39%), (8) STMicroelectronics (1.38%), (9) Analog Devices (1.09%), and (10) Intel (0.84%).

In 1987, TSMC started the journey with a 3-micron process node—a couple of generations behind that top-tier chip makers like Intel. By the 1990s, it was still behind with a 180 nm process node. But along with the growth of fabless companies, primarily due to the development of processors for mobile handsets and other handheld devices, TSMC process node maturity started ramping up, reaching 10nm by 2015—ahead of Intel.

By 2020, with the adoption of EUV lithography, TSMC and Samsung emerged as the duopoly in the 5 nm process node. And in 2022, TSMC became the only semiconductor company offering production-level foundry services at 4nm. And Apple became the lead customer for getting its bionic A16 chip, containing 16 billion transistors, processed with TSMC’s 4nm process node. Besides, while Intel has struggled with the yield and technology complexity to move to 7nm, both Apple and Intel plan to use TSMC’s 2nm node in 2025.

Unbeatable market share, revenue, and profitability growth

In 2022, it was estimated that as high as 90% of high-end microchips were produced by TSMC’s foundry. Although Intel reported loss and revenue decline in 2022, TSMC announced as high as 44% revenue growth during the first 09 months of 2022. Besides, in November 2022, TSMC’s revenue jumped as high as 50%, primarily due to Apple’s A16 chip. Furthermore, the net income of TSMC for the twelve months ending September 30, 2022, was $30.498B, a 45.45% increase year-over-year.

Learning, retaining, and reinforcing process yield management competence–intangible capability

Business success in wafer processing depends on yield management through learning and process development. It demands knowing how to prepare chip designs for production, build them, and, most importantly, fix the inevitable bugs that appear. TSMC has done a great job in institutionalizing, capturing, and retaining this knowledge, which often generates in the form of art through experience. Its practice at immense scale, repeatedly, has created immense capability of offering higher at less cost than competitors can deliver. This factor is self-reinforcing, creating a virtuous effect.

Consequentially, TSMC gets to advanced nodes before everyone else and thus gets better yields faster and succeeds in monetizing the premium market. Subsequently, it affords to advance again and again. Along with this migration, it keeps getting its high-paying customers like Apple to the next node. As a result, TSMC keeps leaving followers like Intel with less profitable older process technologies.

Rise of the Taiwan semiconductor industry as a monopoly due to a virtuous cycle

It has been a typical textbook example of keeping generating knowledge and converting it to revenue, opening endless frontier of growth. Unlike the USA or Europe, Taiwan did not begin the journey with a strong foundation in infrastructure, STEM education, and R&D capacity. It started the journey with technology transfer. But instead of operating the imported technology in serving the existing market, often leftovers, the TSMC-led Taiwan semiconductor industry opted for an unconventional path. It deliberately chose the option of pursuing an unproven, risky path in technology, business model, and market.

In the 1970s, Taiwan could not conduct R&D for systematically generate knowledge and ideas—let alone commercialize them. But both Taiwan’s political leadership and ITRI embarked on it. Furthermore, there is no pure-play foundry market. But Dr. Chang envisioned the future around fabless companies and embarked on the venture. Along with developing IPs for improving yield and upgrading process nodes, the Taiwan semiconductor industry also focused on developing cluster effects. Hence, ITRI, Government, and TSMC focused on the growth of OSAT, fabless, PCB design, and electronic system design companies. Furthermore, TSMC also got the benefit from the wrong decision Intel took in rejecting Apple’s request.     

The rise of the Taiwan semiconductor industry is also linked with the uplifting of Taiwan as a high-income country. Its infrastructure, education, and R&D expenditure kept improving along with the advancement of commercialization of semiconductor R&D outputs. As a result, Taiwan avoided the middle-income trap and colossal upfront investment for infrastructure. Therefore, the uprising of the Taiwan semiconductor industry could be a good development lesson for opening up the endless frontier of growth out of STEM—as envisioned by Dr. Vannevar Bush. It’s also a remarkable example of migration from idea importer to exporter—without making a massive upfront investment for conventional indicators.    

Enlighten others

Subscribe

Subscribe now to our newsletter

You May Also Like

Leave a Reply

Your email address will not be published. Required fields are marked *