Zhang Zhongmou is a famous name in the world semiconductor industry.

The Taiwan Integrated Circuit Manufacturing Co., Ltd. (TSMC), which he founded, almost changed the pattern of the world semiconductor industry on its own, allowing wafer foundry to be spun off independently.

Through the unremitting efforts of 120000 SMC people in several generations, they have constantly conquered the chip manufacturing of the most advanced process, and the chip, as the core element of cutting-edge electronic equipment, has a status of "sticking neck" to the scientific and technological progress of all countries. it also makes TSMC a must-compete place for geo-strategists today.

In his own words, the business model of wafer foundry services is a destructive innovation.

While wafer manufacturing and process breakthrough is a technology-and capital-intensive field in the semiconductor industry chain, the advantages established by TSMC through 30 years of efforts have been unmatched by competitors.

In 2020, TSMC's market capitalization reached US $600 billion, far surpassing its former hegemon Intel, and Intel, which was previously dismissive of TSMC's model, also began to announce the entry of wafer foundry, which is ironic to Zhang Zhongmou.

Speaking of TSMC's competitors, Zhang Zhongmou believes that at present, there is a gap between the mainland and the United States, and Samsung Electronics needs to be the most vigilant.

The following is the full text of the one-hour speech organized by Congtou to share with you--

The title of my speech today is "cherish the advantages of semiconductor wafer manufacturing in Taiwan". This speech is my appeal to Taiwan and TSMC. I am a retired TSMC person and have no rights in the company. Therefore, TSMC is also an important object of my appeal this time.

First of all, I would like to talk a little bit about the history of semiconductors, otherwise we may not quite understand the advantages of semiconductor wafer manufacturing in Taiwan.

The PPT I prepared this time, the first chapter is a brief history of semiconductors, and the second chapter is about the importance of semiconductors and why they have now become a must for geopolitical strategists.

This has to be reduced to the division of labour in the semiconductor industry. If it were not for the division of labour, we would not have this problem today.

This division of labor began decades ago.

The fourth part is about to get to the point-- (China) Taiwan's advantages in wafer manufacturing.

Fifth, we should talk about the semiconductor industrial park with the fab as the center.

Because in terms of wafer manufacturing, TSMC is not the only one in Taiwan, and wafer manufacturing has also led to a lot of upstream and downstream and mid-stream industries.

Sixth, let's talk about the founding of the professional wafer manufacturing industry, and it is also the founding of TSMC. This is actually a rare gathering, and I don't think such an opportunity can occur sometimes in an era.

Seventh, the success of TSMC.

Eighth, the status of TSMC today

The ninth to the eleventh is about competitors, the United States, Chinese mainland and South Korea, in wafer manufacturing, these three are the most important competitors.

Finally, the 12th point is also in response to what is being discussed in society. After TSMC, I would also like to talk about the possibility of the next "protecting G Shenshan".

Bell Labs invents transistors, ushering in a new era in the semiconductor industry

The conductivity of semiconductors is between conductors (such as metals) and insulators (such as wood). The conductivity of semiconductors can be controlled by the intervention of the purity of their materials, so they are called semiconductors.

But until 1948, only scientists, especially physicists, knew about semiconductors, which people had not heard of.

In 1948, a major event in the semiconductor industry happened. The AT&T of the United States was AT & T, the largest telecommunications company in the United States at that time. There was a Bell Lab under it. Bell Labs has been a world-class research institution for many years.

Three scientists at Bell Labs, led by Shockley, invented transistors (transistors).

This invention is very important because it is very small. In the past, only vacuum tubes could do what transistors could do, but vacuum tubes were very large. During World War II, a computer made of vacuum tubes in the United States was the size of a house.

When I came to the United States in 1949, in my first year in the United States, I didn't see a computer. I went to Harvard University at that time, and I didn't see a computer at all.

It was only when I transferred to MIT in 1950 that I saw the computer.

And at that time, I also began to learn some programming. at that time, the computers were vacuum tube computers, and the whole room was as big as the computer. At that time, the computing power of a computer as big as a room was far less than the mobile phone that everyone is using today.

In short, in 1948, the invention of the transistor, AT&T also saw the importance of this invention, the three physicists won the Nobel Prize in 1956 for the invention of the transistor.

Shockley was the same age as my mother. I was his student when I was at Stanford. I often listened to him. He gave a good lecture, but the most unforgettable thing was his arrogance.

There are many students, including me, who are afraid to ask him questions.

Because we see other students asking him questions, he taunts them before answering them, and asks them a few simple questions, mocking that their questions are ridiculous, so that few students ask him questions again.

In 1952, AT&T realized that it could not own the transistor because the invention was so important that it granted patents to other companies, including GE, IBM, and Texas Instruments, and began to invest in the production of transistors.

Texas Instruments, where I later worked, was a very small company at the time, only an authorized manufacturer, but over the next few decades, Texas Instruments was the most successful.

After that, computers and semiconductors began to develop in parallel, but as I said just now, computers need transistors and semiconductors most.

Joined the semiconductor industry in 1955 and witnessed the birth of integrated circuits

After I got my master's degree in MIT in 1955, I joined the semiconductor industry. It can also be said that the development of semiconductors has been integrated with my life since 1955.

In 1958, I joined Texas Instruments, and a colleague of mine named Jack Kilby (Jack Kilby, the inventor of the integrated circuit), who was eight years older than me.

But we were all regarded as contemporaries, and he was studying integrated circuits (that is, integrated circuits).

Kilby is a very innovative person, his education is not high, there is no PhD degree, if you talk to him about theoretical physics he does not understand, but he is very innovative, his own identity is an engineer.

Occasionally, when someone said that Kilby was a scientist, he would correct it as soon as he heard it, saying, "I'm not a scientist, I'm an engineer."

He is such a man that it can be said that Kilby invented the integrated circuit under my nose.

At that time, there was a man named Robert Noyce (Robert Noyce, one of the founders of Intel and another inventor of integrated circuits) who worked for Fairchild Semiconductor.

I had just met at IEDM (International Electron Devices Meeting International Electronic Devices Conference) in Washington, when the semiconductor industry was very small and fewer than 100 people attended the conference.

At that time, Noyce and I will talk about Gordon Moore (one of the founders of Intel, Moore's Law), both at Fairchild Semiconductor. I had just joined Deyi to attend the meeting on behalf of Deyi. At that time, the competitors were still very friendly, unlike now, they did not feel murderous, and each other was very gentlemanly.

After the meeting, I went for a beer with Noyce and Moore.

I was only 27 years old, Noyce was 31, Moore was 29, and everyone was young.

We are all very excited, consider ourselves the favorites of heaven, and are lucky to join the promising field of semiconductors.

After we had had enough to drink and eat, we came out of the restaurant, snowflakes were falling in the sky, and we went back to the hotel singing songs.

Back to the point, Kilby and Noyce invented the integrated circuit almost at the same time, and Kilby was about a month or two early. Later, the two men also had some disputes over the right to invent, and finally the court ruled that the two sides had jointly invented the integrated circuit.

Later, in 2000, Kilby won the Nobel Prize for inventing the integrated circuit. Unfortunately, Noyce died in 1990 at the age of 63, but he also had a wonderful life. He had a lot of girlfriends and flew his own plane when he went out. I often go scuba diving and swimming.

The greatest significance of Moore's Law lies in its reflexivity, forcing semiconductor companies to double the density of transistors every two years.

The next big event for semiconductors, I think, was in 1965, when Moore predicted that the transistor density of electronics would double every one and a half to two years. This prediction, later known as Moore's Law, was quite valid until recently.

Although Moore only made a prediction at the time, it was actually an indicator, forcing every semiconductor company to double the transistor density every year and a half to two years.

Later, the semiconductor industry was no longer gentlemanly, and competitors began to kill each other, so every company was thinking that if I didn't double it, I would be overtaken by my competitors, so everyone tried their best to do it.

In this process, the invention and implementation of MOS (Field effect Transistor) also made Moore's Law possible. Later, IEDM asked me to give a speech and asked me to list the most significant innovations in the semiconductor industry. I included transistors and integrated circuits, as well as Moore's Law, because it was equivalent to a pressure.

In addition, I put MOS into it, Moore's Law would not be realized without MOS, and I also regard Foundry as an important innovation in the history of the semiconductor industry.

Since the 1980s, semiconductor applications have expanded rapidly, mainly from PC (personal computer) to later mobile phones and other electronic products. Why do I say since 1980, because IBM launched PC products in 1980 to popularize them.

Because IBM was a technology giant at that time, we saw that IBM began to promote PC, which shows that PC is commercially available, not just a small number of people's toys.

Before IBM, some small companies were making PC, but people thought it was just a toy, just like a game console, but after IBM began to do PC, it quickly became popular, especially in the 1990s, PC has entered the home and become a real personal computer.

The 1990s was also the era of Andy Grove Intel's then CEO, who was regarded as Intel's greatest CEO, which is a bit of a hero of the times, because the 1990s was also the era of the rapid popularity of PC, and Intel was almost the only company that made processors.

Then in 1987, Zhang Zhongmou set up TSMC, a brand new business model in Taiwan, specializing in wafer manufacturing services, a subversive innovation coined by Clayton Christensen, a professor at Harvard Business School.

It can be said that the establishment of TSMC has subverted the entire semiconductor industry. Why? I'll explain later.

In 2020, TSMC became the most valuable semiconductor company, and former hegemon Intel also began to announce the entry of wafer foundry.

In 2020, TSMC became the world's largest semiconductor company by market capitalization.

Previously, Texas Instruments dominated the semiconductor industry from the 1960s to the mid-1980s, and Intel dominated the semiconductor industry from late 1980 to recently.

Last year, TSMC's market capitalization reached 600 billion US dollars, while Intel only had more than 200 billion, less than half of TSMC's. Of course, it is not certain what will happen in the future.

In 2021, the former overlord Intel announced that they would also do wafer foundry services, which was quite ironic to me. In their heyday, they did not think much of wafer foundry.

When TSMC was first established, I went to Intel to invest, but they refused at that time.

However, the timing was not very good at that time, because the economy was not very good in 1985, but in any case, they were somewhat contemptuous at that time.

Later, TSMC was founded, and several founders of Intel helped us a lot in my personal face.

But they never thought that the business model of wafer foundry would become so important, and they never thought that one day they would also enter the local wafer manufacturing.

After talking about history for such a long time, we are now speeding up the progress a little bit.

The importance of semiconductors needless to say, national defense missiles, navigation GPS, now everyone is very familiar, there are all in the car.

There is another short story about GPS. I heard GPS in Deyi for the first time. I have top secret authority in Deyi. I am in charge of the semiconductor department.

At that time, Deyi also had a defense system department, and they needed integrated circuit boards to do GPS to navigate missiles. When I heard them talk about GPS technology, I thought it was amazing that they could hit targets with a diameter of 5 feet from thousands of miles away. Now almost every car has a built-in GPS navigation system.

A brief history of semiconductor division of labor, from IDM to closed testing, IC design, and wafer fabrication

Industry and commerce, such as computers, daily life, such as mobile phones and semiconductors, are everywhere. Of course, the COVID-19 epidemic has accelerated the global social transformation. Next we will talk about the division of labor of semiconductors.

When I first joined the semiconductor industry, that is, in 1955, no one was doing IC design (integrated circuit design). It was still in the era of transistors. A few years later, I began to do IC design.

IC design is very technology-intensive and high value-added, but it is not capital-intensive and does not require a lot of capital investment.

On the other hand, process R & D and wafer manufacturing are both technology-intensive and capital-intensive, and the added value is also very high.

Encapsulation and testing are both less technology-intensive and capital-intensive. It requires more capital investment than IC design, but not as much technology investment as IC design.

In short, each semiconductor company started out on its own, and the division of labor began in the 1960s.

What was separated by our ancestors at that time was a closed test. at that time, the wages in Hong Kong, Taiwan and Singapore were much lower than those in the United States, and the wages in Japan were even only 5% of those in the United States.

At that time, I suggested that Deyi go to (China) Taiwan for development, but they prefer to go to Japan, because Japan also has only 5% of the wages of the United States, and (China) Taiwan has only 1%, but there is not much difference.

But I said that wages in Japan will rise soon, and Taiwan will not be so fast. Later, Deyi came to Taiwan. Of course, this is another small episode.

In short, packaging and testing have been divided since the 1960s, although they have not yet been outsourced to specialized companies, but to set up branches in places with lower wages.

But in this way, local entrepreneurs found that both Deyi and Motorola had come to Kaifeng to test the plant, and then they would follow suit, because the technology of closed testing was not difficult, and the indirect costs of local manufacturers were even lower than those of international manufacturers. And then it was really separated.

In 1985, when I was president of General Instruments (General Instrument) in New York, I met a man named Gordon Campbell, who had run a semiconductor company before and was very successful. He had just sold the company in 1984, and he came to me and said that he wanted to set up another company and needed an investment of 50 million US dollars.

I asked him if he had a business plan, and he said it was all in his head. I said that even if I supported you, I would report to the board of directors of General Instruments, so I must have a business plan. He said give him two weeks.

But three weeks later, he didn't send me a business plan. I was actually quite interested in his plan, so I called him. He said, "I don't need your 50 million. Just 5 million is enough. I'll put it together myself."

I said, why?

He said that I would not do wafer manufacturing because it was too capital-intensive and I was going to start a special design company-- that was the first time I had heard of a company that specializes in IC design.

I thought at that time that since there could be a company specializing in IC design, there could be a company specializing in wafer manufacturing.

Of course, I didn't come up with the specific idea until a year later, and the establishment of TSMC is the realization of this idea.

Why is wafer manufacturing a destructive innovation? because process development and wafer manufacturing used to be the core of IDM, it is generally believed that closed testing can be separated, but as the core of wafer manufacturing, no one wants to separate it from IDM.

TSMC's disruptive and innovative business model has turned semiconductor companies into customers, resulting in the loss of jobs in the former manufacturing department of semiconductor companies.

According to TSMC's business model, semiconductor companies have become our customers and friends, which is the greatest significance of the business model, that is, how to define your customer base and how you make money from them.

But semiconductor companies used to make their own wafers, and if we want to make wafers specifically for them, then our competitors are the wafer manufacturing departments of these companies, which is the destructive part of this business model.

If you do a good job in wafer manufacturing, half of your customers, half of the people in the semiconductor company will be very happy, but half of them will be very unhappy because their jobs have been robbed, so this actually involves the internal politics of the company.

Americans are not as dedicated as Chinese. Taiwan has a sound semiconductor industry chain.

Next we will talk about the advantages of Taiwan in wafer manufacturing.

First of all, it is very important that there are a large number of excellent and dedicated engineers, technicians and operators who are willing to join the manufacturing industry.

Americans are definitely not as dedicated as Taiwanese, at least engineers, but the semiconductor industry needs good and dedicated engineers, technicians and workers, and they need to be willing to work in manufacturing.

In the United States, the manufacturing industry lost its popularity decades ago, and many people have gone to invest in the financial industry, which is more popular, so this is the first advantage of Taiwan.

Another point is that high-speed rail and highways are suitable for the transfer of large-scale manufacturing personnel.

At TSMC's three manufacturing centers, Hsinchu, Tainan and Taichung, thousands of engineers can be transferred without having to move, they can complete an one-day commute via high-speed rail, and TSMC also provides staff dormitories.

And (China) there is not only TSMC in Taiwan, but also many design companies upstream, such as MediaTek.

In the middle reaches, there are many suppliers, gas suppliers and so on, as well as important semiconductor equipment manufacturers around the world, such as ASML, Applied Material and so on.

Downstream, there are assembly and testing companies, which is a quite complete industrial chain.

TSMC was founded in 1985, which can be said to be a gathering of people at the right time.

At that time, the ITRI had been doing semiconductor business for ten years, and it was at the end of the mountain. It needed a lot of money every year, and it was said that it was not good for the technology to compete with the people. At that time, they wanted to find a way out, so I took the opportunity to enter, but it could not be said to be a hitchhiker, because there was no precedent for the first wafer foundry service, and the advantage was that there were no competitors and the disadvantages were that there were no customers, so I would not talk about this part.

What are the reasons for TSMC's success?

A very important part of it is the advantages of Taiwan in wafer manufacturing that I just mentioned. Another point is that TSMC is led by professional managers from the chairman down to the board of directors. Of course, this also has disadvantages. But I think professional manager model is still a very good model to set up a world-class enterprise.

In addition, we have persisted in R & D investment for a long time. We have had 120000 employees before and after, including those who have already left. Our turnover rate is actually very low, only about 3 to 4 percent. We now have 50, 000 employees, and the total number of employees in the past 30 years is only 120000.

Of course, the support of the community is also very important, and this is also an important part of my appeal today. I hope that you will continue to support TSMC and cherish TSMC.

TSMC manufactures 1 stroke 4 semiconductor components, and almost everyone uses TSMC's products.

TSMC is the leader of semiconductor wafer foundry.

The volume of the semiconductor market last year was US $476 billion, of which memory (that is, memory) was 117 billion and logic components were 359 billion, of which about 1MB 4m-in fact, only 1MB 4-was made by TSMC.

Almost everyone uses semiconductor products made by TSMC in their daily life and work.

I need to wear a hearing aid now. I recently found out that the integrated circuit in the hearing aid is also made by TSMC.

How did I find out?

This hearing aid company comes back to sell new products to me every three years, and I ask him what are the benefits of the new products?

He talked about a lot of changes, and said that the integrated circuits in it had been upgraded, but I also listened to him and changed three or four, but to be honest, the chip upgrade did not seem to improve the function of hearing aids.

Wafer manufacturing in the mainland lags behind TSMC by more than five years, and the United States should not worry about it.

What needs to worry most is Samsung of South Korea, whose advantages are similar to those of TSMC.

Next, let's take a look at the competitors.

The United States is the most powerful. Compared with Taiwan, land has an absolute advantage in the manufacturing conditions of American wafers, but I have just mentioned that Taiwan has an advantage that the United States does not have.

Talents in the United States are not as good as those sent in the past by engineers, technicians, foremen, operators, and even (China) Taiwan.

The result is that the unit cost of wafer manufacturing in the United States is significantly higher than that in Taiwan, so it relies on subsidies, but the subsidies are only short-term and cannot make up for the long-term competitive disadvantage.

The second is the competition with the mainland.

After 20 years of subsidies totaling tens of billions of dollars, the mainland's semiconductor manufacturing still lags behind TSMC by more than five years, and logical semiconductor design lags behind the United States and Taiwan by one or two years.

Therefore, the mainland is not yet a rival, especially in wafer manufacturing.

Then there is the competition with South Korea.

In the field of wafer manufacturing, Samsung Electronics is a strong competitor to TSMC. Why?

Because South Korea's advantages in wafer manufacturing are similar to those in Taiwan, Samsung has both talent and personnel transfer, and their managers can also play a first-class role in South Korea.

(China) it will be difficult for Taiwan to have another "sacred mountain". I hope you will cherish TSMC.

Now some people talk about looking for the next sacred mountain. If Shenshan is defined as an industry that is important to the world and has a high market share in Taiwan.

(China) there are only 27 million people in Taiwan, and the place is small. You are looking for an industry that has potential advantages and is important to the world. I found one, that is, wafer manufacturing.

But over the next few decades, I didn't find a second such industry, and even if you find such a new business model, it will take many years to operate, so my answer to this question is--

难！

Semiconductor wafer manufacturing is an important industry for people's livelihood, economy, and national defense. it is also the first industry in Taiwan to gain a competitive advantage. this advantage is not easy to come by, and it is not easy to maintain it. I hope that TSMC will work hard to keep it. This is my call today. Thank you.