Taiwan's transformation into global semiconductor leadership and future challenges

Jackson Hu, Op-Ed contributed to DIGITIMES Asia 0

Credit: AFP

In university, I studied electrical engineering, and at that time, semiconductors were an emerging field. The electronics courses we took still taught vacuum tubes. It wasn't until near graduation that a visiting professor from abroad started teaching courses on semiconductors. After completing my military service, I went to the United States to study computer engineering. Computer engineering is closely related to semiconductors, as semiconductor chips gradually replaced vacuum tubes as components of computers. In 1978, after graduating from graduate school, I found a job as a design engineer in Silicon Valley, California, and began to delve into advanced microprocessor design.

In 1976, Taiwan sent over thirty engineers to RCA Corporation in the United States to learn about CMOS design and manufacturing. You could say that Taiwan's semiconductor industry was in its infancy and learning phase at that time. Let's also revisit who initially suggested to the government to promote this. According to Dr. Hu Ting-hua, the first director of the Institute of Electronics at the Industrial Technology Research Institute (ITRI), in the later stages of Taiwan's 10 Major Construction Projects, then-Premier Chiang Ching-kuo instructed his secretary-general, Fei Hua, to plan and develop one or two industries that could fully utilize the ongoing infrastructure projects. Fei Hua, a graduate of the Civil Engineering Department at Shanghai Chiao-Tung University, had a classmate named Pan Wen-yuan who graduated from the Electrical Engineering Department and was working as the head of semiconductor technology research and development at RCA Corporation in the United States at the time.

Through this connection, Fei learned about the usefulness of semiconductors, their potential, and how they could help Taiwan's economic development. Then a small group of people suggested to relevant government departments to introduce and develop semiconductor technologies. At that time, RCA Corporation focused on developing consumer products such as color televisions. Power consumption for consumer products must be very low, which is why they suggested learning CMOS technology instead of the mainstream, fast NMOS technology. NMOS consumed too much power and could not integrate many transistors.

So, the first decision to choose CMOS was the right one. According to records, the government submitted a technology license proposal to 14 American semiconductor companies, and RCA's proposed technology licensing conditions were the most favorable. Therefore, it was decided to cooperate with RCA, and the licensing fee paid was US $3.5 million. Most of the engineers selected to go to the US for study were chosen from the Industrial Technology Research Institute. At that time, there were no semiconductor companies in Taiwan yet.

After those engineers returned from RCA, they were assigned to different units to start designing and manufacturing semiconductor chips. A small 4-inch production line was established in the Industrial Technology Research Institute, which later spun off and became United Microelectronics Corporation (UMC) in 1980. TSMC was established in 1987.

Who created the business model of pure-play foundry?

UMC was founded as an IDM (Integrated Device Manufacturer) company, meaning it had both design and manufacturing capabilities. TSMC was founded as a pure-play foundry company specializing in manufacturing for customers. In the United States at that time, all semiconductor companies were IDM. Later, with the evolution of Moore's Law, the cost of process technology research and development and building new factories became increasingly high, leading to the rise of the foundry model.

The idea of this model was proposed by Professor Carver Mead of the California Institute of Technology. Global IDM companies gradually realized that the cost of building fabs was too high, and if their product demand weren't large enough, the cost of producing chips would be too high and not competitive. Therefore, the entire semiconductor industry gradually split into two areas: design and foundry. Foundry companies focused on process technology research and improving yield.

Following Moore's Law curve, they introduced a new generation of processes every two years that were smaller, faster, and more power efficient. Design companies focused on designing competitive products. Because they didn't need to invest in fabs, they significantly reduced financial risks. Design companies came to foundry companies for manufacturing. With the increase in foundry scale, cost sharing, and competitiveness of customers' products, this became a mutually beneficial business model. Both design and foundry companies profited, and soon, the entire world participated in this economic model. Consequently, the semiconductor industry flourished. The rapid growth of the personal computer and smartphone industries also benefited from this business model.

Earning industry status through hard work and great leadership

Former US President Trump recently said, "Taiwan stole American jobs because the US used to produce everything themselves." This statement is entirely political rhetoric from an outsider and does not stand up to scrutiny. Today, Taiwan's contract manufacturing semiconductor output exceeds 70% of the global total. TSMC alone accounts for over 50%, making it the world's largest pure-play foundry, with over 90% of the market share in advanced process technologies. This process has been the result of many years of diligent work and incremental steps.

Back when Taiwan's semiconductor industry began, it didn't know what products to produce, and its experience in quality control and yield was insufficient. It was far from the standards required by leading semiconductor companies in the world such as Intel and Texas Instruments. How did Taiwan manage to catch up actively? And even surpass international giants like Intel?

Two individuals played significant roles in this: K.T. Li and Morris Chang.

Morris Chang was one of the earliest engineers with Chinese background to participate in the semiconductor industry. Texas Instruments was a leading player in the global semiconductor industry, almost inventing integrated circuits (ICs) simultaneously with Intel. In this environment, after years of training, Morris Chang naturally accumulated considerable experience in technology and management. According to Morris Chang himself, Texas Instruments had internal rules for ranking managers. Morris Chang ranked third and was in charge of the entire semiconductor industry. It was impossible to rise to this position in Texas Instrument without being able to convince others in terms of technology and management. Additionally, Texas Instruments had very high requirements for the character, integrity, and honesty of senior executives. Therefore, he was one of the top semiconductor engineers and senior executives in the Chinese community.

K.T. Li's achievement lay in his ability to recognize talent. He found a highly trained senior semiconductor management talent from Texas Instruments for Taiwan. He repeatedly persuaded Morris Chang to return to Taiwan to work over several years. Li also served as Minister of Economic Affairs and Minister of Finance and made significant contributions to Taiwan's economic and industrial development.

In my opinion, Morris Chang's main contributions to TSMC were twofold:

First, he continuously recruited experienced foreign engineers strengthening the company's research and production capabilities in process technology. From initially falling behind companies like Intel and TI to eventually surpassing them and becoming the world's number one, this process also evolved into the company's recruitment culture, one of the main reasons other foundries couldn't catch up.

Secondly, Morris Chang established a culture of integrity and honesty within the company. From a small story told by his wife: he wanted his wife to return company notebooks because he insisted, they shouldn't be given as gifts to friends. This drew a clear line between public and private matters, eliminating illegal activities such as insider trading, soliciting kickbacks from suppliers, and embezzlement. In Asian cultures, relationships and private gains from public resources are common. If supervisors aren't honest, they can't manage subordinates. This is the principle of leading by example.

Why China lags so far behind in semiconductors?

Looking back at the history of semiconductors, Taiwan was incredibly lucky. At that time, there were broad-minded leaders like Chiang Ching-kuo who entrusted capable officials like Sun Yun-Hsuan, K. T. Li, and Chao Yao-tung to plan and execute boldly. Another point that many people may not have noticed is that mainland China was during the Cultural Revolution, destroying its culture and talent.

During the ten years of turmoil from 1966 to 1976, technological development came to a complete standstill. It wasn't until after 1978 when Deng Xiaoping's "reform and opening up" began that students started graduating from high schools, universities, and graduate schools after another ten years. This "reversal of fortunes" is why mainland China fell behind Taiwan in semiconductors.

SMIC (Semiconductor Manufacturing International Corporation) in Shanghai was mainland China's first 8-inch wafer fab, established in 2000 by Richard Chang, who was educated in Taiwan and trained in the US.

In 2019, the South China Morning Post held a series of three seminars exploring two topics: 1. Why is mainland China so far behind Taiwan in semiconductors? 2. Can mainland China catch up in design and production? Participants included me and senior practitioners from the mainland China Semiconductor Industry Association. In fact, in the 1950s, China sent students to the US to study semiconductors and computer technologies. In the early 1960s, scholars who had completed their studies began returning to mainland China to "serve the motherland." However, with the Cultural Revolution, these scholars with American backgrounds suffered persecution, and many couldn't bear the pressure and committed suicide. This dealt a significant blow to the development of semiconductor and computer technologies in mainland China.

The seminars concluded that it's difficult for mainland China to catch up with Taiwan in manufacturing because of the high complexity of manufacturing technology. In terms of design, there is an opportunity if there are design tools available. Later, the US imposed commercial sanctions, not allowing the sale of advanced lithography machines like EUV to China, which had an even greater impact.

Looking back at Taiwan, it was under the leadership of Chiang Ching-kuo that a group of engineers-turned-government officials like Sun, Li, and Chao were appointed. These officials were loyal to the country and brave in bearing their duties. Under their efforts, Taiwan established the Hsinchu Science Park, the semiconductor industry, China Steel Corporation, and more. These projects all succeeded, laying a solid foundation for Taiwan's economy, and their influence continues to this day.

Taiwan's future challenges

So, I would say Taiwan was extremely fortunate! Unfortunately, that inspiring era may never return. Today, Taiwan is democratic, and the government's focus is only on winning the next election. Officials appointed at all levels are those who have contributed to elections, and whether they are suitable for their positions or have professional backgrounds doesn't seem to be a consideration. In the long run, there's a lack of individuals with responsibility and foresight to lead the country. They only continue to enjoy the semiconductor dividend, which won't take Taiwan very far.

Today, if we examine Taiwan's semiconductor achievements over the past 44 years, it is the world's No.1 in wafer foundry, No.2 in IC design, and No.1 in packaging and testing. The achievements are impressive and are what other countries around the world hope to learn from and emulate. The development of semiconductor process technologies is approaching physical limits, and Moore's Law is beginning to slow down. When will it hit a ceiling and be unable to continue? That is something everyone worries about. If that day comes, how should Taiwan respond? What new technologies should Taiwan develop to maintain its world-leading position? These are things that the government and the industry should seriously consider. If we don't plan ahead and wait until that day comes to worry, it might be too late.

Finally, I want to point out that Taiwan's semiconductor companies received government subsidies back then, some did well, and some did not. There is a significant difference over the past few decades. For example, if you compare TSMC and UMC's performance, you can see it. Today, TSMC's market cap is more than 30 times that of UMC's, according to reports. The average salary at TSMC is twice that of UMC's, and TSMC was founded seven years later. Why is there such a substantial difference? This is a question worth exploring, and I encourage doctoral and master's degree students in the School of Management to do their dissertations on this topic to find out the main reasons and the crux of the problem. This will serve as a reference for the future development of emerging industries in Taiwan.

About the author:

Jackson Hu retired as chairman and CEO of United Microelectronics Corp. (UMC) in 2008. He is now the director and advisor of one American firm and one European start-up company.

Hu graduated from the Department of Electrical Engineering at National Taiwan University and earned his Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign. He also studied for an MBA while working. He spent 25 years of his career in Silicon Valley and accumulated rich experiences in IC design. Before the availability of EDA design tools, he designed a 16-bit microprocessor manually; Hu also designed and developed multiple graphics accelerator chips at S3, which significantly improved the graphics capabilities of personal computers running under Microsoft Windows; He designed and developed GPS navigation chips at SiRF to eventually bring the navigation function into the cell phone. After joining UMC, a major foundry in Taiwan, Hu worked with customers to make ICs smaller, faster, and more power-efficient.

Disclaimer: The views expressed within the Op-Ed article are solely the author's and do not reflect the opinions and position of DIGITIMES Asia and its affiliates.