Taiwan's semiconductor talents have moved to the mainland on a large scale. Taiwanese media analyzes the comparison of semiconductor strengths on both sides of the Taiwan Strait!

18 years, from a lone soldier in a foreign land to a "high-level mercenary"

In the two great eras, the fates of Taiwan’s semiconductor people who went west to the Chinese mainland were very different. In the early 2000s, 250 elites from the UMC system, led by Xu Jianhua, the “Prince of UMC” who was the most likely successor at the time, went to Suzhou to build a factory. However, they were investigated and prosecuted by Taiwan, and their dream of listing and allotment was shattered.

This group of people is like the lone foreign army left behind on the Thai-Myanmar border after the Kuomintang-Communist Civil War. Mainland China does not pay attention to them, and companies in Taiwan have no place to go. More than a decade later, under the Sino-US trade war, China's "comprehensive chip refining" struggle in the semiconductor industry has changed everything. Not only has the Hejian, which has been silent for many years, turned around and planned to be listed on the A-share market in mainland China, but a new generation of high-level mercenaries in Taiwan has also appeared. They are the core part of the "Chinese chip" furnace, but they symbolize the departure of the last batch of key talents with the highest gold content in Taiwan.

June 2018 is of special significance to the Taiwan region of China.

The retirement of TSMC founder Morris Chang, who created the legend of wafer foundry, marks the end of an era.

At the same time, China and the United States are caught in a new Cold War, and the semiconductor industry has become the focus of the two powers' game. Talents in Taiwan are being poached by mainland China in large numbers. Even UMC's mainland China subsidiary Hejian announced plans to go public on the A-share market on June 29 to prevent talent from being lost to other new wafer fabs in mainland China and to replenish capital.

At this critical moment, Business Weekly visited the most watched semiconductor clusters in mainland China: Shanghai, Hefei, Wuhan and Nanjing. As soon as the interview began, news kept coming in...

Liang Mengsong, a research and development expert from TSMC and Samsung, may help SMIC catch up with UMC in terms of process technology next year, or even surpass it! After being poached by SMIC, the largest wafer foundry in mainland China, he raised the trial production yield of its 14nm process from single digits to over 90% in just three quarters.

People in Taiwan achieved what SMIC could not do in the past 18 years in just three quarters; and they built the memory industry in mainland China, which had been stagnant for nearly 20 years, from zero to one in just two and a half years.

At first glance, this group of people from Taiwan, China, seem to be in a good position, but many of them describe themselves as high-level mercenaries panning for gold under the guns of the United States, China, and Taiwan. They are caught in the hail of tariffs and sales bans imposed by the United States, China, and Israel, and behind them, their former employers in Taiwan, China, are hunting them down by laws such as leaking business secrets.

Why are these former "upstarts of Hsinchu Science Park" willing to leave for mainland China despite the infamy? What changes will their westward migration bring to both sides of the Taiwan Strait? The following is our tracking record.

Going west not only brought plane tickets, but also letters of proof, "All 60 people here received it"

Our first stop was Shanghai, the most mature and highest-value semiconductor cluster in mainland China. SMIC, Hongli Semiconductor, which was invested and founded by Formosa Plastics' second-generation founder Wang Wen-yang in 2000, and MediaTek's mainland China rival Spreadtrum all have their headquarters here.

In 2016, Shanghai's semiconductor output value exceeded RMB 100 billion, accounting for about 23% of mainland China's semiconductor output value, making it the largest semiconductor base in mainland China.

In terms of physical distance, Shanghai is only about an hour and a half away from Taipei by plane. However, the semiconductor people who have moved to Shanghai in the past two years have a psychological distance from Taiwan, and they are prepared to never return to Taiwan to work again.

Their brains affect the destination of tens of billions of profits.

A person from Taiwan who has worked in a wafer foundry in mainland China for nearly ten years told me that now when people go west to mainland China, in addition to the plane ticket, they also have another document to accompany them: a registered letter sent by the former employer.

Because the intellectual property (IP) in everyone's mind can easily involve tens of billions of dollars in interests and may overturn the fate of a company or even an industry.

"All 60 people here have received (the letter of evidence), all of them have received it!" In a cafe in Zhangjiang Science Park, an interviewee who has been working for nearly ten years deliberately lowered his voice. He spoke to me with a voice recorder that was almost inaudible. After working, this group of people were warned not to disclose business secrets. "Their families called to ask, how could this happen? Why did they receive the letter of evidence?"

The 60 people he mentioned are the team that was reported at the beginning of last year when Shanghai-based Huali Microelectronics poached members of UMC's 28nm process development team. Although UMC denied it, it has become an unspoken secret in the Shanghai semiconductor circle.

How much benefit can a breakthrough in the 28nm process bring?

Take UMC as an example. Its 28nm process orders contributed at least NT$22 billion in revenue last year. Even though TSMC's 28nm has been in mass production for seven years, it still accounted for 23% of its receivables last year, equivalent to approximately NT$225 billion in revenue.

Another recent case is that memory manufacturer Nanya Technology announced on June 22 that it had sued a former engineer for allegedly stealing trade secrets to work in mainland China. Nanya believes that if a competitor uses the technology, it could gain more than NT$3.8 billion, equivalent to nearly 50% of Nanya's net profit in the first quarter of this year.

Precisely because these mercenaries involve billions or even tens of billions of dollars in interests and allow mainland China to accelerate its threat to Taiwan's semiconductor industry, they have to bear the unsympathetic eyes of people from all walks of life in Taiwan.

Fighting for a high-paying stage, facing compatriots "seems like treason"

A semiconductor worker who has left mainland China was charged with four crimes by his former employer. In court, "the judge's attitude was just short of pointing at me and saying: You are sorry for Taiwan." Another Taiwanese who works at the center also recalled that when he returned to his alma mater to share his work experience in mainland China with his juniors, "many of the juniors asked questions that made them feel like they were traitors."

The people who can fully empathize with the feelings of these Taiwanese mercenaries are the semiconductor people from South Korea. South Koreans are more nationalistic and cannot tolerate "betrayal". Almost all the South Korean semiconductor talents who have come to mainland China have made up their minds not to return home.

According to the Zhili Management Consulting Company, which handles hundreds of mid- to high-level semiconductor headhunting cases every year, and estimates from semiconductor industry insiders, there are currently about 2,000 semiconductor people from Taiwan working for mainland Chinese companies, and another 1,000 working at mainland Chinese companies' bases in Taiwan.

Most of these 3,000 people are senior managers or R&D personnel, accounting for about 6%-7% of the total 43,300 R&D personnel in the semiconductor industry in Taiwan.

High salary and stage are one of the key factors that attract this group of people to mainland China.

Mainland China needs them so badly because it is the world's largest chip consumer market, importing more chips than crude oil each year. However, over the past decade, the capital-light, quick-money mentality of mainland China's Internet companies has made them view the semiconductor industry as "something for fools" because it requires tens of billions of dollars in investment and a technology takes three to five years to come out.

Now that the US-China trade war has begun, the talent and technology gap has been more clearly highlighted.

At around 5pm, we saw a large number of people leaving work at the entrance of SMIC. I was curious, this time the Chinese mainland invested 100 billion yuan and introduced talents from Taiwan and other countries around the world, will it really be as Morris Chang said: it will progress very quickly?

Are you still going there even if your technology has been learned? Peripheral contractors: "Chinese mainland semiconductors are like Tang Monk's meat, the whole world wants a bite."

At eight o'clock in the morning, we drove from the suburbs southwest of Hefei to Hefei Ruili (Changxin), a memory factory 40 minutes away, which had an initial investment of nearly NT$250 billion.

This factory, funded by the Hefei municipal government, was the main destination for 500 engineers poached by Inotera and Nanya Technology from mainland China last year. Its CEO, Wang Ningguo, was born in Nanjing but grew up in Taiwan, and Inotera's retired senior vice president, David Liu, also worked here.

Every wafer fab design drawing must be submitted

This wafer factory covers an area of ​​60,000 square meters and is larger than TSMC's factory. Its core projects, such as the pressure-free chamber, electromechanical, wastewater and gas recovery, etc., are all undertaken by contractors in Taiwan.

Zhou Zhiming, general manager of China Mainland Region of Megalink Industrial, which specializes in wastewater treatment for semiconductor factories, pointed to the construction buildings built next to the wafer factory and said, "This is just a small Taiwanese village. All the well-known semiconductor engineering companies in Taiwan, such as Fanxuan, Asia Space, and Han Tang, have come here."

Mainland China is now "making chips for everyone". Even Jilin, where the weather is cold and not suitable for wafer fab operation, and inland cities such as Zigong City in Sichuan Province, which have no industrial cluster effect, all want to build wafer fabs. In fact, it is not just people from Taiwan. Samsung, Micron, Intel and other manufacturers have also set up shop. It is estimated that there are at least 40 wafer fabs in operation in mainland China.

Two interviewees told me: With policy support, mainland China's semiconductors are like a piece of Tang Monk's meat. Not only Taiwan, but also manufacturers around the world want to have a bite.

When companies and mercenaries from all over the world gather on the stage of mainland China, it is indeed catalyzing the rapid development of China's semiconductors. Currently, in the fields of IC design and memory, China has almost reached a tie with Taiwan.

Companies that come to mainland China to participate in the "National Chip Refining" event are actually aware of the price they will pay if their technology is learned away.

A local Taiwanese businessman revealed that engineers working in mainland China have to hand in a design drawing to the authorities in the name of project safety review. "The drawings must be stamped by the authorities, and all drawings will be seen. Many people will hide them, but even if you do, they (mainland China) can still learn something from them."

Manufacturers come not only for profit, but also for defense. Because China may subsidize the Chinese supply chain to promote its own chips and reduce foreign purchases as much as possible. Hong Binghong, a partner of Huanyu Wealth who serves as a consultant for a semiconductor engineering company, said, "If you don't gamble, you have no chance of winning..."

Fortune and wealth are sought in danger. Knowing "measure" has become a necessary survival skill for mercenaries.

Mercenary survival skills: Be disposable chopsticks and give fewer comments

A senior executive from mainland China who has worked at SMIC for more than ten years said that his principle is not to participate in national projects or things that require a stamp of approval within the company. "You are a mercenary here, so don't be too opinionated... People from Taiwan are more sensitive about administrative responsibilities."

More than one interviewee said: "You should treat yourself as disposable chopsticks and don't think about working until retirement (in mainland China)."

This group of semiconductor people are very clear that the technology they have in their minds has a shelf life. Many of them set their careers in mainland China to five or seven years. Because if they cannot continue to make progress in mainland China and contribute what they have learned in Taiwan, after the contract expires, they will have to accept a pay cut, layoffs, or move down the stairs to a lower-level mainland Chinese company, continue to sell the technology they originally knew, and become semiconductor nomads.

As Sino-US relations become more sensitive, mercenaries in Taiwan are walking on a tightrope. A Taiwanese businessman gave an example, saying that after Hefei Ruili poached a large number of engineers from Inotera in the past two years, some engineers defected and were suspected of leaking trade secrets. Last year, Micron sued, and the Chinese mainland has suspended employment and salary, and the accused engineers will be reinstated after resolving the lawsuit.

Hefei showed me that when a large amount of capital and talent gathered in mainland China, it was adding fuel to the "national chip refining". However, when mainland China accelerated forward, did this mean that the loss of talent in Taiwan's semiconductor industry was just the beginning...?

As soon as I got off the plane, the company called me and said, "From this moment on, you are the property of the Party."

When we entered Wuhan, we found that the work pace here was significantly faster than in Shanghai. An interviewee compared it to "the Hsinchu Science Park in the 1990s."

This place looks like a foreign company, 10% of the employees are foreigners

Wuhan is the base of Yangtze Memory Technologies, the national memory team of mainland China against the United States, Japan and South Korea. At the airport, you can see many Japanese people, most of whom are here to support the company's partners.

This emerging memory base in mainland China currently has about 3,000 employees, 10% of whom come from all over the world, not only Taiwan, but also mercenaries from Japan, South Korea, Singapore and the United States.

A senior manager from Taiwan who once worked here said that this place is more like a foreign company than a Chinese company. The regular weekly manager meetings are held. Among the 20 people attending, 80% of them are not native Chinese speakers. "All meetings are held in English, and here, everyone can be seen, even Yukio Sakamoto (former president of the bankrupt Japanese memory manufacturer Elpida) has been here."

As a national team, the working pattern is also different. In recent years, TSMC's capital expenditure has averaged about US$11 billion to US$12 billion per year, which is rare in the world, but Yangtze Memory's plan is US$88 billion in five years.

With capital as the backing and vision as the fuel, the middle and senior executive gave an example. He once asked for a global supply chain map. In Taiwan, this might take a week to compile. But after he handed it over in mainland China, the map was hung in the office the next day. "Young people here now have a strong passion for Chinese chips. They feel that they are doing something very proud and completing a great cause for the nation."

The situation and pressure go hand in hand. The middle and senior executive recalled that when he first landed, he received a call from the company's human resources department, who said to him: "From this moment on, you are the party's property. You can't talk nonsense or go to places you shouldn't go."

Every day they are racing against time. "If we can't do it, we mercenaries will take the money and leave, but people like Zhao Weiguo (Chairman of Tsinghua Unigroup) might be arrested and imprisoned!" Therefore, unlike the mature settlements in Shanghai, working overtime until midnight is the norm here, just like Hsinchu Science Park before the implementation of dividend expense.

Taiwan is stagnant, and people in their 40s and 50s are confused

In fact, the managers we interviewed came to mainland China not only for money, but also because the development of Taiwan has stagnated.

"In Taiwan's semiconductor industry, people in their 45s and 50s are very confused. Most companies are not expanding... You may become a department manager at the age of 50 and stay in that position for the rest of your life!" said a Taiwanese manager.

From the manufacturing perspective, the last time UMC started construction of a new plant in Taiwan was six years ago, and Powerchip and Macronix have been eight to ten years away. On the design side, the top three IC design companies in Taiwan, MediaTek, Ruimin, and Lianyong, have all seen a significant decline in the average profit per employee over the past decade. But MediaTek has seen a 92% decline over the past decade.

One side is declining, while the other side is booming. The memory industry employees in Taiwan, who have always relied on others, feel it more deeply. This time when China starts the fight, everyone hopes to have a chance to compete with the big brothers Samsung and Micron! "Give me ten years, and I have a good chance to bring my technology and products to world-class standards," said the manager.

We asked, "Do you think the influx of talent into mainland China will continue?" Everyone answered: In the next five years, the situation will only intensify, but it will not continue indefinitely, because in five or seven years, when China has enough local talent, it will no longer rely on mercenaries.

When the talent base is weakened, what should Taiwan, China do next?

Two years ago, this place was covered with yellow earth... This year, Taiwan's semiconductor industry may be surpassed.

Our last stop was Nanjing, the most anticipated new semiconductor cluster in mainland China. This is TSMC’s first 12-inch wafer factory in mainland China.

Two or three years ago, the Pukou Special Economic Zone where TSMC's Nanjing plant is located was still a vast expanse of loess. Now, not only are there many high-rise buildings, but the circular building of TSMC's new plant is like a shining sun, attracting about 200 upstream and downstream manufacturers to settle in, ranging from IC design, equipment suppliers to chemical raw materials, and so on.

Two hundred upstream and downstream factories followed, like planting seeds

Hou Mingxiao, semiconductor industry analyst at Lyon Securities, estimates that within four years, the Chinese mainland market will likely account for more than a certain percentage of TSMC's revenue and will be one of TSMC's important growth engines.

TSMC's Nanjing plant is like a seed for China's semiconductor industry. Currently, the 16nm process produced by the Nanjing plant is ahead of dozens of other wafer plants in China. The Chinese government hopes that talents and technology can overflow from TSMC to accelerate the upgrading of Chinese companies, from one plant to a hundred suppliers, and then to a thousand or ten thousand talents.

After TSMC built its factory, many global suppliers also moved to Nanjing, including ASML, a semiconductor lithography equipment manufacturer from the Netherlands.

Among the 24 members of the ASML office, 6 are from Taiwan. Huang Qun'en, the regional manager in charge of the office, was sent from Taiwan. He does not consider himself a mercenary, but his important task during the two years of his assignment, in addition to serving TSMC's Nanjing plant, is to lead his local Chinese colleagues so that they can take charge of the Nanjing office in the future and serve TSMC alone, while he will retire and return to Taiwan.

To this end, they also sent their mainland Chinese colleagues to Tainan for training, and some even stayed there for half a year. "People from Taiwan are really dedicated to their work," said a mainland Chinese employee who had trained in Tainan for half a year.

This is our second visit to the Pukou Special Zone. When we interviewed last September, TSMC's plant was still in the final stages of construction. Trucks were parked in front of the gate, bringing in the machines and equipment needed for the plant. Now the plant has begun mass production. It took only 20 months from plant construction to mass production, which is at least 50% faster than other companies.

In the new Cold War era, the more fiercely the United States clamps down on mainland China's technology, the stronger mainland China's efforts to recruit semiconductor mercenaries to move westward will become, and no one will be able to stop it.

Second- and third-tier Taiwanese factories will be eliminated at an accelerated pace

How should the Taiwan region of China view this matter?

We must realize that although this will not shake the national foundation, it will accelerate the polarization of the semiconductor industry.

If an industry is short of talent, people will flow to top companies like a funnel, leaving the middle and back end of the industry facing a talent crisis. When a large number of semiconductor mercenaries move westward, the first to bear the brunt will be the middle and back end companies.

The likely scenario in the future is that top talent will continue to stay in first-tier companies such as TSMC and MediaTek, but second- and third-tier semiconductors in Taiwan will be eliminated faster due to a lack of talent.

Looking to the future, we can be more open to mainland China.

More than one semiconductor industry executive has urged the government not to interfere with Chinese mainland capital investing in Taiwanese companies. “If you don’t sell what you have now and get money, how can you invest in new areas?”

However, we can also choose not to rely solely on mainland China, but to expand globally.

Huang Rican, a partner at Jones Day, an international law firm specializing in corporate mergers and acquisitions, has recently repeatedly urged that industries in Taiwan must "go beyond China," that is, actively diversify their layout and explore markets outside of mainland China. "We can no longer be wishful thinking in mainland China. It's difficult for you to stay and not leave. This is a wake-up call."

The future of Taiwanese manufacturers: Go beyond mainland China and expand globally

He gave an example, such as Taiwan-based GlobalWafers, which has diversified its layout and revenue sources through continuous acquisitions of factories in Europe and Japan in the past decade. In recent years, domestic sales, combined with sales in the Americas and other regions, have exceeded 50%. "GlobalWafers is no longer dependent on the mainland, it is completely different." Huang Rican said.

In the past, the semiconductor industry has created a "silicon barrier" for Taiwan, meaning that in order to prevent the global IC industry from being disrupted, countries will work hard to maintain the security of the Taiwan Strait. Today, this silicon barrier has gradually weakened due to the rise of semiconductors in mainland China, and the semiconductor output value of Taiwan is likely to be surpassed by mainland China this year.

A new Cold War between China and the United States has indirectly accelerated the evolution of cross-strait semiconductor competition. Every change in current affairs is always accompanied by opportunities and threats.

Although we cannot stop talented people from leaving, we can create a fertile soil that can attract new talents and find new ways out for the industry.

For example, Zhaolian has doubled its popularity in mainland China in the past two years because it undertook the TSMC Nanjing plant project. Now it is "looking for people" and has customers coming to its door all the time. While Zhou Zhiming is making money in mainland China, he is thinking: Can I take the money I made here back to Taiwan to develop new technologies so that I can always run faster than my competitors? Can Taiwan focus more on niche technologies?

We even see the rise of new semiconductor companies, which have been using mixed talent combinations and tactics to compete in the global market and bring surprises from the beginning.

More stringent competition is not a bad thing for Taiwan, China, nor for its talents. It is forcing us to face up to our own values, take action and make changes.

Developing the semiconductor industry is a long road. However, the Internet of Things and AI have become the driving force for mainland China to overtake other countries in the field of IC design.

Unlike mobile phones or computer processors that emphasize high performance, IoT devices are small, have low battery capacity, and emphasize low power consumption, so they do not require the most advanced military semiconductor processes. But the key is to have a service provider who can pass on all hardware to the ecosystem, and then use big data to enable these devices to meet consumer needs.

Mainland China is very accustomed to this kind of ecosystem development. There are Internet giants such as Tencent and Alibaba that have taken control of people's communications, shopping, payments, games and other aspects of life. They dominate all kinds of data and can roll out AI and the Internet of Things ecosystem to create more applications such as unmanned stores and smart homes.

The most obvious example of the new play of mainland China's semiconductor companies is Bitmain, a Bitcoin mining machine manufacturer. They make their own chips, install them in mining machines for sale, and combine Bitcoin cloud mining platforms, managed mining machines and other services to form an ecosystem. The company positioned itself as an IC design company in small steps, but it was one of TSMC's top ten customers last year. Two mainland Chinese youths under the age of 40 founded the company about five years ago, and its revenue last year exceeded NT$75 billion.

Foxconn is isolated from China's major home appliance manufacturers, and will use its advantages in industrial Internet of Things and smart homes to reverse its development from the application end and return to the development of semiconductors.

As of the end of last year, there were 1,380 IC design companies in mainland China, more than five times the number in Taiwan. Among them, there are "unicorns" such as Cambrian and DeePhi Technology with a valuation of more than US$1 billion.

Mainland China is replicating the logic of the Internet to attract users and nurture the ecosystem to the semiconductor industry. However, except for Bitmain, no significant results have been achieved so far. Will it succeed? The next three years will be the key indicator for observation.

The Chinese mainland ecosystem is catching up, a full analysis of the semiconductor strengths on both sides of the Taiwan Strait

The Sino-US Cold War has accelerated the development of semiconductors on the other side of the strait and also brought about a new situation in the competition between the semiconductor industries on both sides of the strait.

Taiwan has always been the most proud of its semiconductor industry, which has provided us with a "silicon barrier". What is the current situation compared with mainland China?

Comparing the top technologies on both sides of the Taiwan Strait in various sub-fields, the answer is: On average, Taiwan is still three to five years ahead. But except for foundry and GEM, the gap in other fields is narrowing rapidly, and there is almost no gap. The semiconductor output value of both sides of the Taiwan Strait is likely to cross this year.

Business Weekly interviewed Chen Huiming, a former foreign analyst who was voted the best semiconductor analyst in the Asia-Pacific region by Asiamoney, Hou Mingxiao, a semiconductor industry analyst at CLSA, Lin Jianhong, a researcher at TrendForce, and other people in the semiconductor industry, to analyze the gap in semiconductor strength between the two sides of the Taiwan Strait.

First, the gap between the two sides is most significant in wafer foundry. TSMC has already mass-produced the industry's most advanced 7nm process this year, a process that currently only TSMC and Samsung are capable of producing. In comparison, SMIC, the most advanced wafer foundry in mainland China, can currently only mass-produce the 28nm process, which is three generations behind TSMC.

However, this does not mean that SMIC can be underestimated, because it is very likely to catch up with the former wafer giant UMC next year and start to lead the latter.

After SMIC poached Liang Mong-song, a former R&D general from Samsung and TSMC, as co-CEO last October, the company's 14nm trial production yield has rapidly increased from 3% to 95%.

Hou Mingxiao said that although a high trial production yield does not mean a high mass production yield in the future, after UMC has announced that it will temporarily suspend the research and development of more advanced processes below 14nm, if SMIC successfully mass-produces the 14nm process next year, it means that the company's technology has officially caught up with UMC.

It is worth thinking about that SMIC relied on poaching Liang Mengsong and his team to quickly break through the new process in less than three quarters. Whether it can replicate this model in the future and poach R&D teams from technology leaders such as TSMC, Samsung, and Intel to continue to narrow the gap with its competitors will be worth observing.

Another area where Taiwan is far ahead is packaging and testing. TSMC is at least five years ahead of mainland China in terms of its "CoWoS" and "InFO" technologies, which it uses to serve a small number of customers such as Apple. However, TSMC is not a professional packaging and testing company and only provides this service to its foundry customers.

Although Taiwan has a solid position in advanced packaging technology, Lin Jianhong said that in the packaging and testing industry, advanced technology only accounts for about 30% of the business, and the other 70% is about production capacity and yield. Currently, among the top ten global packaging and testing foundries, the third, sixth, and seventh are all Chinese mainland manufacturers, which puts pressure on Taiwanese manufacturers. This is also one of the reasons why ASE is eager to merge with SPIL.

At present, the strength of most of the packaging and testing factories in Taiwan is almost the same as that in China. Therefore, in the past two years, SPIL and Chip Micro have successively acquired shares in their Chinese subsidiaries with Tsinghua Unigroup, using equity to consolidate the cooperative relationship and exchange the market. In the future, most of the packaging and testing factories in Taiwan can only compete with mainland China in terms of production capacity, yield rate and customer relations.

In the field of semiconductor materials and equipment, Taiwan's most significant lead is in silicon wafers, whose stock prices have soared in the past two years due to shortages and price increases.

Hou Mingxiao believes that GlobalWafers, the only Taiwanese company that can supply 12-inch silicon wafers, is at least three years ahead of mainland China and two years ahead of mainland China in 8-inch silicon wafers. However, he is not so optimistic in other areas. "Since Hanwei Microelectronics was sold, basically no one in Taiwan is making semiconductor front-end process equipment. Mainland China has deliberately supported local equipment manufacturers, but Taiwan has not. In fact, if we have to say, mainland China is ahead of Taiwan."

Lin Jianhong also believes that Taiwan's semiconductor equipment industry has almost no R&D and production of complete equipment. "Although mainland China has just started, it may emerge in ten or twenty years (only catching up to the level of European and American manufacturers), but Taiwan will not."

Since the government-led merger of Taiwan Memory Corporation in 2009, Taiwan's memory industry has become more conservative and focused on profit in its pursuit of advanced processes. Inotera, which originally had the most advanced process, was acquired by Micron at the end of 2016 and became its subsidiary.

Lin Jianhong said that if Yangtze Memory Technologies, led by former Inotera chairman Gao Qiquan, or Hefei Ruili (Changxin), invested by the Hefei municipal government, goes into mass production by the end of this year or next year, there will be no gap between the two sides of the Taiwan Strait in the memory field. Mainland China will even be ahead of Taiwan by producing flash memory that has never been produced in Taiwan.

Although China's memory will most likely only be sold in China in the future due to suspected infringement of the intellectual property rights of companies such as Micron, and will have limited impact on the operations of Taiwanese manufacturers, Taiwan has already lost in terms of technology.

Finally, this is the IC design that Taiwan is likely to be overtaken in. Compared with MediaTek, the most technologically advanced company on both sides of the Taiwan Strait, and HiSilicon, a subsidiary of Huawei, Hou Mingxiao believes that there is no gap between the two.

At the same time, IC design manufacturers in Taiwan are facing two major challenges. The first is the import substitution policy of mainland China, which does not care about quality but only cares about "mass production".

It was reported a few days ago that Chinese mainland officials required panel manufacturer BOE to use at least 50% domestic panel driver ICs. Even Taiwanese companies such as Lianyong, whose technical strength is still superior to that of mainland China, cannot resist the policy banner of nationalism.

The second challenge is that mainland China’s IC design companies have risen in the fields of AI, the Internet of Things, etc., using new strategies such as “building an ecosystem together” and “selling software more than hardware”, and have already created unicorns.

"The trend is that you must move towards the ecosystem, not the supply chain. The supply chain (era) ends with Apple." Chen Minghui, who has frequently traveled between the two sides of the Taiwan Strait in the past two years, made a comparison. For example, in the PC era, the OEM and ODM factories in Taiwan that were responsible for IC design, and then the customers were connected to brand factories such as HP and Dell, did not really understand the terminal market and customer needs.

Although most of the newly emerging IC design companies in mainland China do not have the same solid IC design capabilities as those in Taiwan, they are close to the ecosystem built by giants such as Alibaba, Tencent, Baidu, and Huawei, and are close to customers. They can better understand which gaps in the ecosystem are not being met, and therefore have the opportunity to rise. In short, IC design companies in mainland China are now more "down-to-earth."

In the next three to five years, in addition to wafer foundry and cutting-edge packaging and testing, we will see mainland China gradually catch up in various semiconductor sub-fields, but it is not time for Taiwan to admit defeat. In this month, we also visited Nangang and Hsinchu and saw the new tactics of Taiwan's semiconductor industry. Taiwan's semiconductor industry is not dead in the road ahead, but we are sandwiched between the two superpowers of China and the United States, and we have to fight a more flexible and difficult battle than before.

Future Strategy: How to seize business opportunities amid US-China confrontation and cross-strait tensions?

IC design team "mixed blood guerrillas"

Currently, it is a great era for the semiconductor industry. Trends including AI, Internet of Things, and 5G will bring tens of billions of dollars in business opportunities to the semiconductor industry.

But now, we see that many semiconductor companies in Taiwan are in trouble. In the first quarter of this year, MediaTek's earnings per share fell out of the top ten Taiwanese manufacturers: VeriSilicon faces the threat of China's call to purchase more than 50% of domestic panel driver ICs; Ruimin's average profit contribution per employee has fallen by 41% compared with the past decade.

Corporate positioning should be “borderless”!

Network communications are registered in three places and managed according to local conditions

The success of AI chips requires cooperation with network giants, big data, and close proximity to the terminal application market. For Taiwan, which missed the development of mobile networks, the times seem to be against us. Fortunately, we also see examples of Taiwanese companies breaking through in Nangang and Hsinchu. The following are the breakthrough strategies of these companies:

AI startup Naineng is a rare IC startup in Taiwan. The company just completed its latest round of financing of US$18 million in May this year, led by Hong Kong tycoon Li Ka-shing's Horizons Ventures. Investors include the "Silicon Valley Taiwan Gang" reported by Business Weekly in the past, such as Lin Shibin, founder of the live broadcast platform Twich, and Chen Xinsheng, partner of Presence Capital; previous investors also include international first-tier companies such as Alibaba and Qualcomm.

From the first day, Kneron has been using US R&D and Taiwan's semiconductor experience to serve Chinese customers, deploying talents in all three places with a balanced ratio. "The US has creativity, Taiwan is most familiar with semiconductor manufacturing, but mainland China has the market," said founder and chairman Liu Juncheng. In this case, why not take advantage of each other's strengths.

"Borderless" is particularly important for companies with mainland China as their main market at a time when the new Cold War between China and the United States and the cross-strait situation are tense.

Similar to Kneron, NetLink Communications, a new IoT chip startup that has combined resources from the U.S. and Taiwan to target the Chinese mainland market, has a CEO named Lin Mingyou who studied in the U.S. since childhood and worked for semiconductor companies such as Qualcomm and Broadcom after graduating with a master's degree. He admitted that he was confused about how the company should "identify" and "brand" itself.

Lin Mingyou said that when he was participating in exhibitions, Chinese mainland manufacturers often told him: "Your technology is so strong, you should call yourself an American company, so that you can have a premium and have room for imagination." However, if he positioned himself as an American company, his dilemma is that it is difficult to be acquired in the Chinese mainland market, and even if the trade war worsens, the products may not be sold to mainland China. But if he is a Chinese company, although he can enjoy tax, procurement, film investment and other assistance, the challenge is that the team may not be willing to move all to mainland China, or after landing, the talent will be poached faster.

Finally, he decided to position his business as a "borderless" international enterprise, registered in three places.

Starting a business with mixed races is a real challenge for management. Lin Mingyou said that employees in different regions actually want different things, and incentives must be tailored to local conditions. For example, people in Taiwan want a work-life balance, employees in mainland China want stock options, and American employees use joining a startup as a springboard to apply for a residence permit, because most of the company's employees in the United States are non-Americans.

Lin Mingyou said that when the situation is turbulent, the smaller the company, the more it should act as a flexible guerrilla force. "When the world is in chaos, we have opportunities."

Be down-to-earth and understand the needs!

Naineng's chairman lived in a customer's dormitory for half a year and got a big order

Secondly, you have to lie on the ground to be truly down to earth.

Naineng's core technology is the AI ​​chip that makes various smart devices "smarter". Their latest product is a collaboration with its investor Himax to create an affordable version of the 3D sensing lens on the iPhoneX, which can be installed in ATMs, robotic arms and other devices to give the devices 3D vision. In the future, people can withdraw money by "scanning their faces" or let robotic arms directly screen out defective products with minor defects on the production line. Its current customers include the top three Internet giants in mainland China by market value, China's leading home appliance manufacturers and other manufacturers.

The main team members of Kneron are from the United States and Taiwan, but they can survive in the Chinese mainland market. Liu Juncheng said that the key is to squat low enough and get close to your customers.

Kneron's most important customer at present is a leading Chinese home appliance company. Currently, orders from this company alone can cover its entire year's operating costs.

Once, he didn't understand why customers set inexplicably high acceptance standards for consumer electronic products. For example, the phone lens had to be able to be opened and closed tens of thousands of times and could be used even on rainy days. So he spent half a year living in the customer's employee dormitory every day, eating with the employees, playing ball with the workers, chatting whenever he got the chance, and understanding their ideas.

He finally realized that his client, who originally made home appliances, was using a 10-year warranty mindset to make consumer electronics products. "If you don't understand your customers' needs and the limits of their requirements, your products won't impress them."

Chen Huiming, a former foreign analyst, believes that being close to customers like Kneron is a new skill that IC design companies must possess in the era of AI and the Internet of Things. "What (AI chip company) Cambrian has, small and medium-sized IC design companies in Taiwan are capable of making. The question is, can you get close to the application and know what to put in the IC? That is the key."

In the early days of his business, Liu Juncheng spent half of his time in the United States. Now, he devotes three-quarters of his time to the Asian market, mainly in mainland China. He said, "Wherever the market is, I will be there."

"Small companies need to be guerrillas. Only when the world is in chaos will we have opportunities." Lin Youming, CEO of Netlink Communications

Grasp the niche market thoroughly!

Powertech has customers in the US, Japan and China, with net profit up 21%

Finally, only when you are thorough in the niche market can you have the leverage to move around. Old companies like Licheng are very thorough in this regard.

Last year, the bases of its peers Siliconware Precision Industries and Chip Microelectronics declined, while Powertech achieved a 23% increase in revenue and a 21% growth in net profit.

This is certainly related to the surge in the memory market last year. More than 70% of Powertech's revenue comes from memory packaging and testing. Compared with mobile phone processors, power management chips, panel driver ICs and other products, this is a niche market. The major manufacturers in the world that have entered this field have only been around for a short time. Although memory leaders such as Micron and Samsung have their own packaging and testing capacity, it is still not enough to support their needs and they have to outsource some orders to memory packaging and testing factories, and Powertech is the largest among them.

"(Powertech's) strategy and tactics are certainly worth referring to!" Huang Rican, a partner at Jones Day, believes that Powertech's sticking to the niche field of memory packaging and testing can avoid a head-on battle with manufacturers such as Jiangsu Changdian and Tianshui Huatian that specialize in general IC packaging and testing in mainland China.

At the same time, not allowing mainland China to invest in it also prevents Powertech from being labeled, allowing it to move more freely between the camps of Micron, Toshiba, etc. "It will take some time for Powertech to catch up with mainland China (memory), and Powertech is now in an invincible position." said a supervisor of a packaging and testing factory that allowed Tsinghua Unigroup to invest in it.

Of course, Powertech will also face risks in the future. If Yangtze Memory Technologies Co., Ltd. of Tsinghua Unigroup rises in the future, and Tsinghua Unigroup has already built memory packaging and testing capacity through Chip Microelectronics and Siliconware Precision Industries, it will no longer need to cooperate with Powertech, which may cause the latter to miss out on the Chinese mainland market.

In the future, in an era of reverse trade with severe tariffs and non-tariff barriers, both mainland China and other countries will need more companies or mercenaries to accelerate their industrial development and eliminate the restrictions brought by trade barriers. In an era of more serious protectionism, only by seizing a niche market and defining oneself in a more cross-border way can one survive and thrive in this chaotic era.

The content of this article is selected from "Business Weekly". If there is any infringement, please contact us to delete it. Thank you!

"Xinshiye's media"