TSMC's past: Zhang Zhongmou's choice

Author: Yang Jiankai

"China's Chip Past" tells the story of China's first-hand chip industry spanning 58 years. By recreating the success or failure of the organization and the rise and fall of the industry, it comprehensively reviews the exciting eras and fragments in the history of Chinese chips, presents the laws of the industry, reflects on the ups and downs, and narrates the reform of the economic system, company operations and character behavior. The author of this book has interviewed dozens of people in Chongqing, Shenzhen and other places, and created new first-hand business cases based on hundreds of thousands of words of first-hand historical archives and up to 5 million words of reference materials.

Editor's note: This is an excerpt from "China's Chip Past", which describes the process of Zhang Zhongmou's return to Taiwan and the creation of TSMC.

Business plan for chip foundry

On August 20, 1985, Zhang Zhongmou was officially appointed as the president of the Institute of Industry and Research. The first task when he took office was to answer the proposition put forward by "Political Councilor" Li Guoding:

Three groups of Chinese returned from the United States, and successively established three chip companies in Hsinchu Science Park, Guoshan, Mao Silicon and Huazhi, all relying on factories in Japan and South Korea for manufacturing, and profits were earned, and as a result, the three seeded players were clamoring for money investment from the government.

Chiang Ching-kuo inspects Hsinchu Science and Technology Park

The chip industry on the island is getting farther and farther away from neighboring Japan and South Korea, and the crisis is gradually emerging.

Needless to say, Japan had surpassed the United States in the field of memory, which had the largest chip market share at the time. Under the fierce price war, Intel has been losing step by step, and the management has not yet found the direction of transformation.

South Korea's three major consortiums followed Japan and entered the chip manufacturing field together, with a shot of one billion US dollars, almost ten times the investment of the island's chip industry. Every weekend, flights from Tokyo to Seoul are filled with young engineers.

On the other hand, UMC was a unique sapling, and after its revenue exceeded NT$1 billion in 1983, it stagnated, and when it was listed in 1985, the stock price immediately broke, and the media announced that "UMC does not call", and the officials of the stock exchange called to let the maintenance of the plate, but were refused.

The island's chip industry has been given a "negative" outlook by shrewd public investors, and the authorities have made no move, and this emerging industry is feared to be stillborn.

How to solve this problem?

UMC came up with the idea of chip foundry early in the morning, and sent someone to the United States to lobby Zhang Zhongmou a year ago to ask him to come back and preside over the plan.

Zhang Zhongmou's ideas fit with Cao Xingcheng. He did not come to the island once or twice, and Deyi's factory in Taiwan allowed him to see the advantages of the island:

Heavy teaching, diligent learning, hard-working, from the operator on the production line, foreman, to the engineer in the laboratory, generally have a high level.

To be a professional chip foundry, there is a foundation.

In addition to Cao Xingcheng and Zhang Zhongmou, there should be many people on the island who have the idea of building a chip foundry. This is a natural and logical idea, and an easy solution to the authorities' immediate needs.

However, the authorities did not move until Zhang Zhongmou took over the public post of president of the Institute of Industry and Research, and Li Guoding came forward and asked Zhang Zhongmou to make a plan and take it to the new "chief executive".

TSMC is born

In late August 1985, a few days after arriving in Taiwan, Zhang Zhongmou took the materials that the two staff members had spent two days and two nights to prepare and report to the new "chief executive" and "cabinet" officials.

In order to avoid the embarrassment of the participants who did not know how to do it, he deliberately bypassed professional vocabulary, used a large number of charts and figures to describe the general trend of the chip industry in detail, and analyzed the response measures on the island.

After the briefing, the "Chief Executive" immediately instructed the establishment of a special group composed of the heads of four ministries, including the "Ministry of Finance" and the "Ministry of Economy," and Zhang Zhongmou, to be in charge of the "large-scale integrated circuit plan" of the Institute of Electronics of the Institute of Industry and Research.

In 1985, Zhang Zhongmou took a photo with technical officials on the island when he visited Tsinghua University in Taiwan

With unprecedented specifications, intensity and speed, the authorities have once again launched an assault on the chip industry.

In the last UMC attempt, the thirty-three-year-old Cao Xingcheng accidentally took on the task of heaven, rampaging through a barbaric world with little knowledge, and although he made a path, he could not satisfy the authorities, and the experienced technocrats may only regard him as a poor candidate.

This time, the fifty-four-year-old Zhang Zhongmou, with a road map that had already been planned, had gained the highest political trust on the island when he first arrived, and he seized this opportunity to eagerly push forward:

In August, the briefing was affirmed by the "Executive President", in September, the "Executive Yuan" reported, and in October, the organizational structure was released.

By 1986, the government-run "large-scale integrated circuit plan" was still on the march, and Zhang Zhongmou had begun to make preparations for privatization, looking for capital everywhere.

The "chief executive" believes that only by attracting the investment of an international manufacturer will the private capital on the island be interested in investing in a private company in the chip industry.

As early as Philips set up a factory on the island is interesting, Zhang Zhongmou thought that the factory was too powerful, so he wrote to more than a dozen American companies to sincerely invite shares, rarely replied, and finally only Intel and Deyi were willing to listen to the briefing, and after listening to it, they shook their heads and said no.

The chip industry in the United States does not have a soulmate! That had to accept Philips' stake and technical cooperation.

After getting a foreign businessman, Zhang Zhongmou began to lead a team to persuade private investors.

Most of the private enterprises on the island are conservative industrial capital, and they are particularly cautious about investing in emerging industries. He went to Formosa Plastics three times in two months, accompanied Wang Yongqing to run 5,000 meters at four o'clock in the morning, ate steak in the canteen, and answered the questions of the cannon one by one, and finally won the affirmation of this "god of management", but the investment amount was only a small amount of NT$200 million, which was only a fraction of the entire plan.

Fortunately, private capital is generally hindered by the feelings of the authorities, and under the urging, it has to be expressed. The scattered tens of millions of Taiwan dollars were gathered and became the starting capital of Zhang Zhongmou's chip foundry plan.

In February 1987, after half a year's sprint, "Taiwan Semiconductor Manufacturing Company" (later renamed as "Taiwan Semiconductor Manufacturing Co., Ltd.", hereinafter referred to as TSMC) was established.

Similar to UMC, TSMC also follows the route of "technology research and development of electronics institute ~ technology transfer of private enterprises ~ mass production of products to the market". The difference is that when TSMC was founded, it was a critical moment for the chip industry in East Asia, and Hong Kong journalists, South Korean officials and chaebol cadres were all trying to spy on the military.

However, the most uneasy, nervous, and even a little afraid of the establishment of TSMC is UMC, the leader of the island's chip industry.

Among all the subordinates of the company led by Zhang Zhongmou, Cao Xingcheng was the only one who dared to joke with him, and the two talked and laughed when they were together. But when he sensed the danger, he immediately reacted in a series of agile responses.

Double Hero Fight

Two days before the signing ceremony of TSMC shareholders, UMC held an interim press conference, announcing that it would increase its capital by NT$6 billion and build a new plant, and said that the "new company" should not enjoy any privileges. For a time, the newspapers were full of news that UMC challenged TSMC.

According to common sense, more advanced technology, higher production capacity, and updated equipment mean that TSMC can easily dig into UMC's corner from top to bottom, absorb its talents, and steal its market.

But when it comes to "fair competition," industrial policy is never fair, and once technocrats have more political trust in a businessman, it naturally means that that person has an advantage over others.

Of course, the original intention of all industrial policies is to enhance the overall strength and serve the overall welfare as the purpose. Before the establishment of TSMC, Zhang Zhongmou preached everywhere:

The new company is purely a factory that helps semiconductor companies, has no products of its own design, does not compete with anyone, and welcomes anyone to use large-scale production equipment.

Zhang Zhongmou said this, no one believes that for the battle between TSMC and UMC, observers are full of suspicion and have no special tendency.

On the one hand, they noticed that Philips accounted for 40% of the shares in TSMC, and they were instinctively wary of foreign businessmen, in contrast, UMC became a "representative of national industry".

On the other hand, they also believe that UMC, as the leader of the island's chip industry in the past five years, is a "vested interest", and publicly rebuking TSMC is nothing more than a normal marketing strategy to squeeze out opponents.

In fact, whether it is a "representative of the national industry" or a "vested interest", UMC's new factory is an arrow on the string and has to be launched.

The capacity of the factory built at the beginning was already full, and UMC originally planned to build another one in 1985, but the "Minister of Economy" who led the matter was impeached and removed from office.

In the past two years, the undercurrent has been surging, many new factories are about to move, and the chasers seem to have emerged from the ground overnight, and if they do not expand production and go up, UMC will be in a position where it is not up or down.

Without the support of the authorities, UMC decided to make a desperate bet and invest the profits earned in the past few years to increase capital and build a factory.

Zhang and Cao were on the same stage, hiding knives in laughter

In response, one observer commented:

Battle of the Trapped Beasts.

Stuck

The outside world overestimated the impact of TSMC's birth on UMC. In the three years since its establishment, TSMC has not had an impact on the island market. The reason for this is that UMC and the market structure on the island match.

Everyone stays in simple consumer electronic products, the equipment using UMC's 35 micron process is enough, if TSMC's 1.5 micron high-end factory is used, the cost will increase a lot.

The initial growth of the electronics industry on the island is mostly price-sensitive, and naturally they dare not use expensive ones. As a result, TSMC's existence on the island is independent, as if staying out of the matter, and practitioners just read news from newspapers, and few people really understand.

Zhang Zhongmou's explanation of the island is that the positioning is different. He said that the original intention of TSMC's establishment is to pull the level of the chip industry up, catch up with the United States, and the technical process has always been faster than UMC:

TSMC is ahead of UMC every step of the way.

It is precisely because of the step by step leadership that few people on the island can afford to use it, Zhang Zhongmou has set his eyes on market development overseas, especially the United States, matched by the rare foreign executive team on the island.

He set up an office in the United States and hired an old American who lives in Florida and commutes to Silicon Valley every week. He had frequent exchanges with his former colleagues and subordinates in the United States, and finally found an American who shared the idea to be stationed on the island. When these people came to Taiwan, they were surprised to see a group of young people coming out of the electronics institute rushing to the factory:

In the last decade or two, engineers of this rhythm have not been seen in the United States!

Especially in 1988. This year, the island's stock and real estate markets were surprisingly optimistic that few working people could ignore.

But TSMC's young engineers stayed up all night and quickly got through the technology of Philips, a major foreign shareholder. The sixty or seventy young talents cultivated by electronics have become the basic team of TSMC.

The VLSI plant was leased to TSMC by public tender in 1987, which was the initial starting point for TSMC's development

However, in the eyes of the outside world, the contradiction between American executives and local engineers cannot be reconciled, and "people" are the biggest challenges facing TSMC. The former receives a generous salary and moves in and out freely, while the latter comes out of the electronics institute to start a business, the salary is not high, and the risk is very large. In the long run, the situation is bound to be tense.

And that's exactly what happened. U.S. executives have been slow to develop the market, and local engineers are worried all the time.

A U.S. business representative worked for more than a year, did not bring in a single customer, and when he arrived in Taiwan, he just gave Zhang a thumbs up and said that he was strategically correct. The general manager was unable to adapt to life on the island because his family could not adapt to life on the island, and he handed in his resignation letter less than a year.

Lao Mei tossed around, and TSMC's only strong customer turned out to be only the Institute of Electronics. The three island customers that were originally envisaged to meet at the time of its establishment either had a sluggish business or their orders remained in Japan and South Korea.

Without a stable source of bulk orders, local engineers naturally feel insecure, and their vigilance against major shareholder Philips is also increasing, for fear of becoming a satellite factory under it, completely losing autonomy.

As a result, TSMC, the rising star of the island's chip industry, has fallen into a difficult dilemma:

The capital market did not trust, and Formosa Plastics took advantage of the big bubble in the stock market to dump its stocks in the second year of its establishment;

The general manager was difficult to give birth, and after the first general manager of Laomei left, there was no successor;

Sales channels cannot be opened, and if there is no capable person will run the US market, how to open sales channels?

On Christmas Day 1990, Zhang Zhongmou and two subordinates were sitting in the office, and the three looked at each other with heavy hearts. Towards the end of the year, he already knows that TSMC is another loss-making year this year, and the miracle will not happen in the next six days. How to find rice to put in the pot next year? There is currently little idea about this question.

Intel and Deyi are not optimistic about the foundry road, will there be a future?

Resurrection: The Establishment of American Culture

In 1988, Intel CEO Andy Grove visited Taiwan and temporarily decided to go to TSMC.

Zhang Zhongmou and the American professional managers he hired had a relationship with Intel before, but did not cooperate. TSMC seized the opportunity this time, and under the strong offensive of Zhang Zhongmou, Grove promised to transfer some of its business to TSMC, but it needed to pass the company's quality certification.

At that time, there was no unified ISO quality standard in the world, and Intel's certification was basically equivalent to an unimpeded international certificate.

There are more than 200 outsourced chip production links, and Intel sends people to check them one by one, and pass one before going to the next. More than two hundred questions such as Yang Xufei are irritating, and the young engineers are not convinced and raise objections, but the Americans will never let up. They had to work twenty-four-hour shifts, and finally won Intel's list, and found that they had to do business at a loss. Moreover, if the machine and process are changed in the future, it must be approved in advance by the United States.

It was certainly not a cost-effective business, but TSMC bought Intel's certification.

By 1991, the boom in the chip industry began to explode, and large chip manufacturers in the United States were demonstrated by Intel and handed over production orders to TSMC. In the next decade-long super chip boom cycle, TSMC has become the king of the chip foundry industry.

The great success of the 1990s contrasted sharply with the dismal operation of the first four years of its existence. It seems that overnight, TSMC gained the trust of almost all large American companies, and its performance soared like a rocket.

TSMC's success is the success of American-style corporate culture.

Zhang Zhongmou insists on recruiting professional managers in the United States as general managers, in addition to introducing foreign executives such as Intel CEO to visit, but also intends to establish a set of American-style corporate culture and management system:

The management system is American. There is no pretense in finance, the strictest standards are followed in accounting, the ranks and salaries are determined by the system in personnel, and even in language, full-time English consultants train the company, and English, e-mails, memoranda and high-level meetings, all speak English.

Technological innovation is American. In its first year, in order to save costs, TSMC invented the technology to pack wafers into a sealed box, so that only the dust-free environment of the box needs to be maintained, without the need to make the entire factory completely dust-free. Later, Hewlett-Packard, IBM and other American businessmen came to visit and learned this law.

Business principles are American. Integrity to colleagues, the schedule expires, it should be completed; There is absolutely no preferential treatment for suppliers, even if Zhang Zhongmou is found, he only acts according to price and quality; Strict commitment to customers, saying how many days of delivery, will not allow a one-day delay.

The service model is American. Most of TSMC's customers are located in the United States, which is across the ocean, and communication is inconvenient. But as soon as they sit in the office every day, just turn on the computer, you can see what stage the wafer processing is going on, make a call to the business representative, you can give modification instructions, get the next five years technology roadmap released by TSMC that year, and you can understand what technologies your products have for cooperation in the future.

TSMC has become a "virtual fab" for major manufacturers in the US chip industry. Zhang Zhongmou even imagined that since TSMC's factory is "the customer's own factory", it is natural to let them know how much it costs and how profitable it is.

The company's performance is American. TSMC is obsessed with the pursuit of advanced processes, the improvement of yield, the compression of delivery time, and the ultimate crushing, the company's profits are like sugar water on sugar cane, which surprises public investors.

The distribution of results is American. Whether it is a doctoral manager who has returned from the United States, an engineer who has a turtle on the island, or an operator who has graduated from high school, he has the right to subscribe to dividends.

Those operators hold wafers worth a BMW car in their hands, and their work value is not inferior to the engineers and managers around them, so they usually participate in business discussions on an equal footing.

But there is one thing that is not American, that is, the relationship between Zhang Zhongmou and TSMC.

TSMC is the product of Zhang Zhongmou's highly disciplined American culture. If it were not for his insistence on copying the form and implementing it in substance, and after 30 years of persistence, it is almost inevitable that a high-tech enterprise in East Asia will lag behind its fierce opponent.Chinese

Naturally, Zhang Zhongmou became TSMC's stern father.

For the partners who seek media, he turned them away, insisted on sole proprietorship, and developed independently.

For the management layer he took over in the future, he continued his practice in the Institute of Industry and Research, like a Shaolin strict master, describing the vision of pursuing Intel and Samsung, teaching the principles of daily operation, pointing out the errors of data and data, forcing all young managers to think and find their own answers.

For the grassroots employees who perform the daily work, he advocates that each performs its own duties, has a strict division of labor, and puts professionalism first.

Zhang Zhongmou

As a result, TSMC has become a highly centralized but rapidly expanding centralized organization, standing out among the island's enterprises.

To meet the serious challenges posed by this behemoth, UMC has made two preparations.

On the one hand, the three major criteria of "dividends and shareholding, prohibition of relatives, and retirement at the age of service" have been established, so that UMC has begun to take shape a modern management model.

On the other hand, UMC had no experience in the United States, and lacked enough talent, and the organization expanded like a cell division. UMC identified excellent candidates from its employees, funded them to open up land and established chip design companies such as Novatek and MediaTek as a barrier for UMC.

As a result, a decentralized but closely connected group has gradually formed, which the chip industry calls "United Family Army".

TSMC and the "United Family Army", centralized centralized organization and decentralized decentralized organization, two completely different forms of organizations, have become the protagonists of the business history of the electronics industry.

A magnificent drama of double heroes has just begun.