Morris Chang's latest interview: He bluntly said that chip manufacturing in the United States is wasteful and futile

Latest update time：2022-04-21

Reads：

Source: Content from Business Weekly , thank you.

In its latest podcast, the Brookings Institution in the United States specially invited Morris Chang, the founder of TSMC, the world's largest wafer foundry, to talk about his life from Shanghai to the United States, the opportunity to establish TSMC, and the bottlenecks of the US semiconductor manufacturing industry.

U.S. President Biden pointed out: "Although the United States is a leader in chip design and research, we produce less than 10% of our own chips." The Chips Act passed by the Biden administration a few days ago will subsidize chip manufacturers with more than $50 billion in order to strengthen the U.S. semiconductor manufacturing industry.

However, Morris Chang believes that increasing semiconductor manufacturing in the United States is futile, wasteful and expensive; even if a conflict occurs in the Taiwan Strait, the United States will not have to worry about just "chip manufacturing."

The origin of the semiconductor godfather

Hong Kong war refugees, studying and working in the United States

Morris Chang was born in Ningbo in 1931 and spent his childhood in Hong Kong, Chongqing, and Shanghai. He fled Shanghai before the Communist takeover in 1949 and applied to American universities in Hong Kong, where he described himself as "a bit like a refugee in Hong Kong." After a year at Harvard, Morris Chang transferred to the Massachusetts Institute of Technology (MIT) to study mechanical engineering.

Soon after graduation, Morris Chang successfully joined Texas Instruments, laying a solid foundation for him in the semiconductor industry. However, the road was not all smooth sailing. Although he called his experience of being "forced to retire" at Texas Instruments and serving as president of General Instruments for only one year, it paved the way for him to return to Taiwan and establish TSMC.

The host of the Brookings Institution asked (hereinafter referred to as Q): After graduating from MIT, how much impact did your many years of experience working at Texas Instruments have on your subsequent career development and the establishment of TSMC?

Morris Chang's Answer (hereinafter referred to as Answer): Without the years at TI, I would not have founded TSMC. I did rise to a very high position at TI, and for many years I was the third highest paid person at TI, responsible for the global semiconductor business. Later, the company missed the opportunity to enter the consumer products business, and the semiconductor business also lost to the competition, which put my career at TI in jeopardy.

Later, I was transferred to another department, which was equivalent to being "put out to pasture" (Editor's note: forced to retire early) for me who had been in the semiconductor industry for 20 years. Five years later, I decided to leave, and I was only 52 years old at that time.

After leaving TI, I was offered many job opportunities. In 1983, I became the president of General Instrument, and realized my dream of moving to the United States when I was young: living in a nice apartment in New York and being able to walk a few blocks to the office. At the time, I lived in Trump Tower, and in fact, I ran into Trump in the elevator several times.

General Instrument was a very different company from Texas Instruments. At Texas Instruments, I learned how to build a company from scratch. But General Instruments did it differently. They bought small businesses, improved them, and then sold them. After about a year, I decided that this was not what I wanted to do, so I left in 1985.

TSMC's key advantage is "people"

Manufacturing costs in the U.S. are 50% higher than in Taiwan

From Trump Tower to Taiwan, the key person who brought Morris Chang back to Taiwan was the godfather of Taiwan technology, Kuo-Ting Lee. After Morris Chang left General Instrument, Kuo-Ting Lee recruited him to return to Taiwan to take over as the president of ITRI, turning the IC integrated circuit project from an ITRI project into a living business, and TSMC was born.

Q: How did TSMC become the world's most advanced semiconductor foundry?

A: In the early days of TSMC's development, its technology node lagged behind (Editor's note: 3 nanometers and 5 nanometers are the common technology nodes of semiconductor processes), but its yield rate was very high. Therefore, by becoming a wafer foundry, we avoided our weaknesses and strengthened our advantages, because foundries do not need to design, but only manufacture.

Taiwan's competitive advantage in semiconductor manufacturing is almost entirely related to people. The United States also had these manufacturing talents in the 50s, 60s, and early 70s, but since the 70s, manufacturing talents have migrated to higher-paying industries, and I don't think this is a bad thing.

Q: As top talent moves to other industries, what are the bottlenecks for the U.S. to increase domestic chip production?

A: TSMC has had a manufacturing plant in Oregon for 25 years. In the early days of operation, we naively thought that our costs could be comparable to those in Taiwan. However, the gap between the two places has always been large: the same product, produced in Oregon, is about 50% higher than in Taiwan. After several years of hard work, we have to accept this gap.

Regardless, I believe that the United States' efforts to increase domestic semiconductor manufacturing would be a very expensive endeavor. The United States would face significant cost increases and would not be competitive in the global market, especially against companies like TSMC.

Finally, Morris Chang admitted that if a conflict breaks out in the Taiwan Strait, the United States will worry about more than just chips. He further pointed out that in the absence of war, "increasing U.S. semiconductor manufacturing is futile, wasteful, and expensive." Regarding the complex situation between the United States, China, and Taiwan, Morris Chang hopes that everyone can maintain friendship. He pointed out that "being able to be everyone's wafer supplier is the reason for TSMC's success."

Earlier interview with Morris Chang

There is no real IDM now

How does the head coach who fought in world-class wars and led the rise of TSMC look back on the development of Taiwan's semiconductor industry? Morris Chang accepted two 4-hour interviews with Caixun, talking about the rise and fall of global semiconductor giants and the key to the rise of Taiwan's semiconductor industry from scratch, as well as some unknown secrets.

"You are the first reporter among all media outlets to put Taiwan in the context of global semiconductor industry development," TSMC founder Morris Chang said in an exclusive interview with Caixun in late January.

Opening the rise and fall of the top 10 semiconductor giants in the world over the past 50 years compiled by this magazine, Morris Chang pointed to Sylvania, which was ranked fourth in 1955, and said, "This company was my first job. At that time, it was still a transistor company." He went on to say, "Even Hughes, which was ranked first, was very small at that time and could not be counted."

In 1972, Morris Chang was promoted to general manager of the semiconductor business unit of Texas Instruments. That year, Texas Instruments was the world's largest semiconductor company with the highest revenue, and Morris Chang was the general manager of the Texas Instruments Semiconductor Group with 40,000 employees worldwide.

That year, Intel was only four years old, but it was already regarded as the "strongest rival" by Morris Chang. His competition with Intel was later forced to be interrupted due to a fatal strategic misjudgment by the CEO of Texas Instruments. Intel thus stood at the pinnacle of the global semiconductor industry.

After leaving Texas Instruments and taking over as president of the Industrial Technology Research Institute (ITRI) in Taiwan, Morris Chang saw the advantages and disadvantages of Taiwan's semiconductor development and founded the world's first professional wafer foundry, Taiwan Semiconductor Manufacturing Company. Thirty years later, under his leadership, TSMC has grown from a startup to a semiconductor giant with market value and process technology comparable to Intel. Thanks to this innovative business model, Taiwan has developed a unique and dense foundry supply chain, starting from scratch and gradually becoming an important global semiconductor cluster.

Time goes back 50 years. How does the head coach who fought in world-class wars and led the rise of TSMC look back on this journey? At the end of January, Morris Chang accepted two exclusive interviews with Caixun for a total of 4 hours, talking about the rise and fall of global semiconductor giants in his eyes and the key to the rise of Taiwan's semiconductor industry from scratch, as well as some unknown secrets.

"Texas Instruments made a fatal mistake in 1972" Revealing the inside story of Intel's overtaking 20 years later

Reporter: We have compiled the revenue rankings of the top 10 semiconductor companies in the world over the past 50 years and found that the market has changed dramatically. How do you view such changes?

Morris Chang's answer (hereinafter referred to as the answer): I joined the semiconductor industry in 1955, and then joined Texas Instruments in 1958, and did not leave until 1983. (Editor's note: From 1965 to 1985, Texas Instruments was one of the top two companies in the world in terms of semiconductor sales).

The development of semiconductors can be traced back to 1948 when Shockley, Bardeen, and Brattain invented the transistor at AT&T's Bell Labs. However, there was no semiconductor industry before 1952. After the invention of this technology, it was first manufactured by AT&T's manufacturing division, Western Electric Company, on a very small scale.

In 1952, many people already knew that transistor technology would be very important in the future, so AT&T, shouldering its social responsibility, licensed this technology to other companies. Many large companies bought the license, and Texas Instruments also bought the technology, but it was still a very small company at the time. RCA, GE, Sylvania, and Motorola were all larger than Texas Instruments.

▲ When Morris Chang (second from right) was the general manager of the Germanium Transistor Division of Texas Instruments, he received a certificate of merit from a customer. (Photo/Reproduced from Morris Chang's autobiography)

In 1964, he was promoted to General Manager at Texas Instruments.

But by 1965, TI was already quite powerful, with about 10,000 employees in the semiconductor business. At that time, I had already received my Ph.D. from Stanford and returned to TI. In 1964, I became the general manager of TI's Germanium Transistors unit. When I was the general manager, that unit had the largest revenue scale, but it was not the most promising business. In 1965, I became the general manager of Silicon Transistors, and in 1966, I became the general manager of Integrate Circuits (IC integrated circuits, which combine multiple transistors into more complex circuits). I was really promoted every year.

In 1966, the IC process was still bipolar, and MOS (metal oxide semiconductor) was just a research project. In 1972, I became the vice president of the company group and the general manager of the semiconductor group. MOS (Editor's note: MOS is the mainstream process technology in the semiconductor industry later) also changed from a research project to a business. All semiconductors of Texas Instruments were under my management. I was only in my 30s at that time, and I was promoted quickly and at a young age. It should be the most memorable period of my life.

Q: When you were at Texas Instruments, the company had the largest semiconductor revenue in the world, but later Intel caught up. What was the turning point?

Answer: Texas Instruments has two battlefields. The first is memory, which is my biggest challenge. The other battlefield is where we have an absolute advantage, which is computer chips.

But in 1972, TI made a fatal mistake. That mistake was that TI thought the semiconductor industry was already quite mature and it was not easy to grow significantly. So it changed its business focus to consumer products as its future development. This dragged down the entire company, including the semiconductor division that once dominated the world. This was the key to Intel's overtaking of TI in 1995.

50 years ago, Intel was considered as a rival, and the key to overtaking Texas Instruments was dismantled

Reporter: Apart from TI's own mistakes, what other reasons do you think were behind Intel's success?

Morris Chang: Intel was founded in 1968. When I took over Texas Instruments' entire semiconductor business in 1972, I regarded Intel as my strongest competitor because of its technology. I was also very familiar with their founders, Robert Noyce and Gordon Moore.

At that time, Intel had the advantage of being a startup company. They could issue stock options to attract good talents. At that time, Intel's best-selling product was its memory. I still remember its number 1103. It was a very famous product and was popular for several years.

The founders of Intel are Noyce and Moore, both of whom came from Fairchild Semiconductor, which was a strong competitor of Texas Instruments as early as 1965.

It all started with Shockley. After inventing the transistor, he left Bell Labs and found a wealthy businessman named Beckman who admired him very much, and established the Shockley Semiconductor Laboratory.

Shockley was a very ostentatious man. He called Noyce and Moore directly and asked, "Are you Noyce? I am Shockley." He did not introduce himself. Of course, the two of them knew Shockley's name, so they went to him.

The 8-member team issued an ultimatum to the big boss

But Noyce and Moore only worked in Shockley Semiconductor Laboratory for one year. First, they felt that Shockley was too contemptuous and could not get along with him; second, they felt that Shockley wanted to make a PNPN2 (Shockley 2-pole), not a transistor, but a diode. Shockley thought this would be very successful, but the 8-member team led by Noyce and Moore did not agree.

For these two reasons, these eight people issued an ultimatum to the real big boss Baker: "If Shockley doesn't leave, then we will leave." This is the story of the Traitorous Eight that has been circulating in Silicon Valley for a long time. The person supporting them was Arthur Rock, who can be said to be the earliest venture capitalist in history.

Later, Arthur Rock approached Sherman Fairchild, the owner of Fairchild Camera and Instrument, and told him that there was a group of outstanding people. Fairchild agreed and set up a semiconductor department. Noyce was the head of this department, and Moore was responsible for research and development. Fairchild Semiconductor was established in 1957.

Unexpectedly, in 1968, they repeated the drama in Shockley Laboratory, but at this time Arthur Rock had enough financial resources to fund an independent company, so Noyce and Moore founded Intel.

Intel's success is mainly due to stock options and technical talent. In addition, California is indeed a reason. Talents like to go to California. Moore is originally from California. Although Noyce is from the East Coast, he also likes California after arriving. It can be said that everyone likes California, especially in the 1960s and 1970s.

The starting point of the competition between Samsung and TSMC: "Korea has Lee Kun-hee, Taiwan has me"

The reporter asked: Why did Japan enter the top 10 semiconductor countries in the world in the 1980s, but then declined?

Morris Chang: Toshiba and NEC were very strong in the 1980s. But in fact, Japan has never been strong in logic chips. They are strong in memory, so strong that even Intel surrendered and stopped making memory.

The biggest reason for the decline of Japan's semiconductor industry was the so-called Plaza Accord. In September 1985, the month after I came to Taiwan, the main purpose of the United States was to reach an agreement with five countries, namely France, Germany, Britain, and Japan. Because the United States had a huge trade deficit with Japan and Germany, the United States wanted the currencies of Germany and Japan to appreciate.

I didn't pay attention to Germany, but the yen, before the Plaza Accord in September 1985, was 240 yen to the dollar. At the end of 1987, it became 120 yen. The yen kept rising and later rose to 80 yen. The rise from 240 to 120 alone doubled the cost of Japanese companies.

In addition to the appreciation of the yen, the most important reason for the decline of Japanese companies is Lee Kun-hee of South Korea's Samsung, a very powerful man.

Lee Kun-hee arranged for Morris Chang and Stan Shih to "visit the factory"

Question: Japan was so strong in memory at the time that even Intel couldn’t beat it. Why was Samsung able to rise in such an environment?

A: In 1989, Lee Kun-hee, the former chairman of Samsung, came to Taiwan and invited me (Editor's note: ITRI chairman at the time) and Stan Shih to have breakfast. He knew that Stan Shih wanted to invest in Acer and hoped that Taiwan would not make memory, so he invited us to visit Samsung's factory. He said, "After you see our factory, you will know how much capital is needed to run a memory factory? How many talents are needed? After you know this, maybe you will decide not to invest and cooperate with us."

I knew that Samsung had started making memory in 1983, and I thought it would be a good idea to go and see it, so I flew to Seoul (Editor's note: now Seoul) with Chin-tai Shih, Stan Shih, and the other two. As soon as we got off the plane, we were all treated by Samsung and we stayed in their hotel.

The second day was a full day of visits. The Samsung receptionist would answer any questions. It was indeed a large-scale factory. I had visited many factories before. I only needed to spend half an hour to see them and I could tell their potential. To be honest, I had been away from TI for 6 years at that time. Compared with the best Japanese memory factory of TI in my impression, Samsung was definitely not inferior.

Before we returned to Taiwan on the third day, we met with Lee Kun-hee, who said, "You've seen it, and now you should know that this will require a lot of capital and a lot of talent." I was actually still a little confused at that time, but Stan Shih didn't say anything. A few months later, Stan Shih announced the establishment of Techtronic.

Lee Kun-hee is definitely the key to the rise of Samsung in South Korea, because he is the boss. The semiconductor general managers of Japan's powerful semiconductor companies Toshiba, Hitachi, and NEC are just professional managers, not bosses.

It is very difficult to find a good boss, only one in a thousand; however, Lee Kun-hee happens to be one of those thousand. Lee Kun-hee himself is not a semiconductor expert, but he knows and understands the potential of semiconductors and mobile phones. He is a key figure and a hero who creates the times.

Japan lost to South Korea in memory, but Taiwan did not benefit from it. This drama has nothing to do with Taiwan. Japan does not have logic chips or foundry. South Korea has Lee Kun-hee, and Taiwan has me. Although I am not the boss, I am the one who dares to do it, take risks, and create new business models.

Fundraising was difficult at the beginning. "If I had invested in TSMC, the return would have been a thousand times higher by now."

Reporter: Can you review the process of founding TSMC?

Morris Chang: It was Li Kuo-ting's idea to found TSMC and start raising funds, but Premier Yu Kuo-hua also supported it. Li Kuo-ting was actually solving another problem. Everyone was very proud of the integrated circuit project of the Institute of Electronics in 1965, but where was the future?

The Institute of Electronics built a demonstration factory with good yield, but it was competing with UMC. Both the Institute of Electronics and UMC only looked at Taiwan, and the market was not that big. UMC would also think that the government-supported government agencies were competing with it, which was competing with the people. At the same time, Kuo-Shan, Mosel, and Hua-Chi hoped that the government would support them in building semiconductor factories. Therefore, Li Kuo-Ting and then-Minister of Economic Affairs Li Da-hai had been planning for the Institute of Electronics for 10 years and believed that establishing another company was the best way.

Then I was asked, "Mr. Morris Chang, you are an expert, what kind of company should we set up?" In fact, they were thinking of setting up a company similar to UMC, which I had no interest in, and I thought it would not be very successful, so they proposed the idea of professional wafer manufacturing.

In the past, Taiwan had never placed its semiconductor industry in the context of the world's semiconductor industry. At that time, UMC had already been established, and the Institute of Electronics had started making semiconductors in 1975, but they wanted to serve Taiwan's electronics industry. But how much of the world's electronics industry does Taiwan account for? I'm afraid it's less than 1%. What I brought was the world's semiconductor industry. I was familiar with the world's semiconductor industry, and I put TSMC in the global semiconductor industry and cooperated with them.

Q: Why did you come up with the business model of "professional wafer manufacturing"?

A: Actually, the term business model only appeared in the 1990s. I asked the dean and professors of the business school of MIT, and they all told me that there is no formal definition of business model. The main focus is on what customers you have, not what products you have.

Wafer foundry is not a new idea. In 1980, Carver Mead and Lynn Conway co-wrote a book about chip design. They can be said to be the forefathers of chip design. The book mentioned that chip design must be linked to wafer manufacturing process, so they found this link and predicted that there would be companies that could provide manufacturing services in the future. Moore saw this and named the company that provides wafer manufacturing services to others as Foundry. The name of Foundry was given by Moore.

But Moore did not expect that he could build a dedicated foundry. This was an idea I proposed and I was the one who implemented it.

In 1998, it was the 50th anniversary of the invention of the transistor. An American investment bank held a celebration and invited all those who had contributed to the semiconductor industry in the past 50 years. I was invited to attend, as was Carver Mead. That was the first time I met him. Carver Mead said to me, "Ah! I really admire your professional wafer manufacturing. I have always believed that wafer foundry can be done, but professional wafer foundry was the idea you proposed."

Q: The professional wafer foundry model is a brand new concept. What difficulties did you encounter in the early stages?

A: There were many difficulties, but luckily I was only in my 50s at that time. The first was the difficulty in raising funds. In fact, if I had invested in TSMC, the return would have been a thousand times higher by now.

When writing a fundraising letter for TSMC, one company simply replied, "We are not interested in investing"

At that time, I proposed the idea of professional wafer manufacturing and used this model to raise funds. I wrote more than 10 letters, including to Intel, Texas Instruments, Motorola, Wang An, etc. Although Sun Yunxuan, the Advisor, had suffered a stroke at that time, he was still very helpful. However, most companies did not respond, and some companies simply replied: We are not interested in investing.

Yu Guohua hoped to find a multinational company, and a technologically advanced semiconductor company. Later, we cooperated with Philips. After TSMC was established, Philips quickly became a fairly large customer.

Philips invested in TSMC because of its cheap labor

Why is Philips interested in investing in TSMC? This is a very interesting story. In 1985, the strategy of Philips Netherlands Headquarters was to use Taiwan as a production base with cheap labor, when mainland China was just starting to emerge.

Simply put, first, Philips has its own IDM semiconductor business and will not compete with TSMC. Second, Philips wants to establish a good relationship with the Taiwanese government so that it will be easier to invest in Taiwan in the future. In fact, Philips once had about 40,000 employees in Taiwan in the 1990s, but later they all moved to the mainland.

Another difficulty is that you have locked in your business model, but who are your customers? The customers were actually the semiconductor industry at the time, as well as the fabless design companies that sprang up like mushrooms after a rain. TSMC's business model has indeed subverted the semiconductor industry.

Yield is TSMC's winning card, which makes Philips give up its leading position

The reporter asked: Why did you think the professional wafer foundry model could be successful in Taiwan at that time? What were the conditions in Taiwan?

Morris Chang: In 1985, I came to Taiwan to take over as the president of ITRI. The process capability of ITRI's Electronics Institute was able to produce two microns. At that time, the world's leading companies, including Intel, Texas Instruments, Motorola and even Philips, were already producing one micron. The technological gap between Taiwan and the leading companies was about 2.5 to 3 generations behind. At that time, the only thing we could do to compete with them was the high yield rate of backward technology.

Taiwan's human resource quality determines yield rate

The yield rate of the demonstration factory of the Institute of Electronics is very good, which is significantly different from that of Texas Instruments and General Motors. The key lies in the quality of Taiwan's human resources, including operators, technicians, foremen, and engineers. When I was at Texas Instruments, I found that the yield rate of the Japanese factory was much higher than that of the Texas factory. The reason for this is the talent factor.

In Japan, workers have at least graduated from high school, or even college, and most of them are women. The turnover rate is very low. The Japanese factory manager told me that workers would only quit if they wanted to get married. But the Texas factory could not find employees who graduated from high school. In 1980, it was even difficult to find people who knew English. The turnover rate was as high as 30% when the economy was good, and it was more than 10% when the economy was bad.

As the process improves, the equipment and maintenance technicians must also improve. But in Texas, you can only find people who repair motorcycles and cars, and no maintenance personnel for fine equipment. As a result, the new machines can only operate smoothly 60% to 75% of the time on the production line, while in Japan it can be 90%, which is a big difference. Because semiconductor process equipment is very expensive, there is only one on a production line. If the smooth operation time is reduced, all the work-in-progress will be piled up in front, and it will be difficult for you to improve the yield.

Japanese process engineers are quite good, while the United States had already begun to give up manufacturing at that time; more than 30 years later, the United States now wants to turn back time and resume manufacturing, and I think this is also one of the problems.

The reason why Philips gave up more than half of TSMC's shares was revealed

Q: Why was Philips willing to give up its right to acquire 51% of TSMC and allow TSMC to go public smoothly?

A: Our cooperation with Philips actually took quite a long time. After TSMC was established, Philips quickly became a large customer.

Philips found that although we have relatively backward technology, the price he paid us for those backward technologies was actually lower than his own cost; the fact is that our price is lower than the cost of vertically integrated manufacturing, which is a very important discovery. Of course, TSMC's price competitiveness also comes from good yield.

Why was Philips willing to give up 51% of its shares? When it started raising funds, Philips's shares were about 28%, but it could be increased to 51% with options. Originally, Philips wanted to control the company, but the head office later found that we could run the company better than they did. At that time, Philips' semiconductor department was struggling, and even the general manager was dismissed.

Therefore, Philips adjusted its strategy and switched to pure investment. Whether the pure investment can be recovered depends on whether the company can go public. I told Philips at the time that one of the conditions for the company to go public in Taiwan is that no shareholder can hold more than 50% of the shares. Only then was TSMC able to go public.

Most IC design talents are in UMC, so Taiwan's chances of DRAM success are actually very small

Reporter: What do you think of the development of Taiwan's IC design industry?

Morris Chang: Most of Taiwan's IC design talents are in UMC. In the late 1980s and early 1990s, I think most of them were in UMC. The electronics department did not have many design talents. Around 1995 or 1996, UMC (spun off) MediaTek, Uni-President, and Novatek. Basically, it derived design talents to MediaTek, Uni-President, and Novatek, which was almost 10 years later than the electronics department of ITRI derived TSMC.

Q: What do you think of Taiwan’s development in the memory sector? Are there still opportunities in the future?

Answer: I feel very sorry about Taiwan's past performance in the memory sector. Don't forget, Texas Instruments is a major memory manufacturer. I have the most experience there. Originally, World Advanced was preparing to make DRAM (dynamic random access memory).

In short, Taiwan has had some success in memory. For example, Macronix makes ROM (read-only memory), and they mainly make NOR (non-volatile flash memory). I see that they are also making NAND (storage flash memory), but the vast majority of memory is DRAM. I have tried DRAM many times, but I gave up after losing a lot of money. So it should be said that Taiwan is not very successful in memory, and has only succeeded in relatively small areas.

In fact, Taiwan’s investment in DRAM is not that much. I believe in the Experience curve, so the chance of success is actually very small.

TSMC has a high market value, but I don’t think much of it.

Q: The market value gap between TSMC and Intel is getting wider and wider. What do you think about this?

Answer: Regarding market value, I have always taken it quite lightly. I have seen so many companies with large market value that were considered to be top-notch at the time, but they all faded away later.

Do you know what IBM's market value is now? It was a god in the past! Now its market value is more than $100 billion, and Intel still has more than $200 billion. In 2000, I went to see John Chambers, the CEO of Cisco. At that time, Cisco's market value was about $500 to $600 billion, ranking first in the world. I said, "John, aren't you worried? Aren't you afraid?" He replied, "I think it will definitely be higher in the future." Now, Cisco's market value is more than $200 billion, and it is probably ranked 50th to 100th.

So, not just you, many people think that TSMC's high market value is amazing. I think you should not think like that. I don't take it that seriously.

HPC capabilities unlock AMD and other companies

Q: What changes will occur in the market when TSMC releases its ability to manufacture CPUs?

Answer: This is a very meaningful question. We have released the capabilities of advanced processes in high-performance computing (HPC) to many companies, including AMD, Intel, NVIDIA, Apple, MediaTek, Qualcomm, Broadcom, etc. Of course, they need to have design capabilities to cooperate.

Now, there are Amazon, Meta, Google, etc. These companies are the future Apple. By releasing advanced process technology through TSMC, the market will not be dominated by a single company like Intel.

Q: There are still several large IDM companies in Europe. Will they still exist when new demand for electric vehicles emerges?

Answer: First of all, I do not agree that there are several large IDM companies in Europe. I do not quite agree with the word large. There is no real IDM in the world now. IDMs have all become FAB Lite (light wafer factory).

Europe's STMicroelectronics (STM), Infineon (Infinion), and NXP (NXP) can also be said to be an IDM because they have old FABs, but they are not very large.

Q: Isn’t Intel a real IDM?

A: No, Intel is a big customer of TSMC.

TSMC's two strategic values determine the company's future

Reporter: Geopolitics has driven a global semiconductor investment boom. Can Europe, the United States, and Japan see a semiconductor manufacturing revival again?

Morris Chang replied: It depends on what happens in Taiwan, South Korea, and mainland China. If they collapse, they will also revive.

Last year, I gave a speech on the theme of "Cherishing Taiwan's Advantages in Semiconductor Wafer Manufacturing." Of course, Taiwan lacks everything, but I think that with determination, these negative factors can be resolved; my speech was not only an appeal to the Taiwanese government, but also to society and TSMC, hoping that everyone will cherish it. Our advantages in semiconductor wafer manufacturing in Taiwan are not easy to come by, and if we don't cherish them, it will be easy to lose them.

To answer your question, if South Korea and Taiwan work hard and mainland China can continue to progress on its own, it is unlikely that Europe, the United States, and Japan will see a revival in semiconductor manufacturing within 20 years.

Taiwan Semiconductor from scratch

Q: Looking at this period of Taiwan's semiconductor history, which has gradually turned from weak to strong, how would you comment on it?

A: Taiwan has grown from nothing to something! To say that it has grown from weak to strong is not an accurate description of the process. It should be said that it has grown from nothing to something, and then from a little bit to a little bit to a strong one.

This period was indeed the golden age of Taiwan's semiconductor industry. Taiwan saw its advantage in wafer manufacturing, created a new business model, and implemented many strategies to realize this advantage. For example, Taiwan's semiconductor development advantage was good yield, which has always been the case, but no one noticed it at the beginning.

Q: How did you build TSMC from scratch and then make it strong?

A: TSMC's innovative business model also requires good strategies to be implemented. The only time I wrote down TSMC's strategy was on September 7, 1998. I wrote it in English at that time. There was no need to translate it into Chinese at that time. Anyway, I only showed it to about 10 core members. Basically, it was about how we could make money with high yield. The most important strategies include the following two:

The first strategy is to meet customer needs. I listed 11 values that customers need, such as providing technology that can help customers compete, low prices, short time from finalization to mass production, protecting customers' intellectual property rights, and providing design services that help customers compete.

TSMC's strategy is to be significantly better than its competitors in all aspects except low price. Our pricing strategy is to earn premium without losing customers. To successfully achieve this strategy, we must have excellent customer relationships, and the benchmark for measuring this relationship is whether the customer looks at us "first and last look".

Another strategy is to make TSMC a market- and service-oriented enterprise. In TSMC's organizational culture, every employee should be a business person of the company and see serving customers as an opportunity. In addition, TSMC must establish a world-class marketing planning department. This department is not only capable of determining the technology blueprint with the R&D organization, but also capable of developing new customers, recommending new services that can be launched, and formulating different marketing strategies for different application products, fields, regions, and customers.

As for what indicators to use to measure the success of the strategy? Compared with the three ideal quantitative results I proposed in 1998, they are: first, maintain a return on equity of more than 20%; second, become the world's largest integrated circuit manufacturing company, with revenue at least twice that of the second largest competitor; third, achieve revenue of US$10 billion before 2010. Later, it was proved that our strategy was executed very successfully and the results exceeded expectations.

Q: Is today's success due to good strategies and good leaders? Are there big differences between different companies that are developing in Taiwan?

A: Yes, and another one is values. I think for a company, values, vision, and strategy are three essential elements, and values are very important. If we have to rank them, values are ranked first, and they will determine the future of a company.

*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.

Today is the 3017th content shared by "Semiconductor Industry Observer" for you, welcome to follow.

Latest articles about

■TSMC's 2nm is too powerful, UMC is too miserable

■The United States has heavily funded this semiconductor technology

■Tesla is also snapping up HBM 4

■Intel's next-generation AI chip is exposed for the first time

■Nvidia releases its largest chip yet

■The Danish robot giant invites you to "do something" together

■Self-developed DPU released: Microsoft chip, full of firepower

■In the post-Moore era, optical computing chips have become the key to breakthrough, and domestic manufacturers have great potential!

■Open source software in crisis

■The global semiconductor equipment giants are all in trouble