2nm's "League of Losers"

There is a new company in Japan, Rapidus, that claims to mass-produce 2nm chips in Japan by 2027.

At first glance, it sounds like a joke, but then everyone realizes that this little-known Japanese company may be serious.

Rapidus was listed and opened in 2022, and received a big red envelope of 70 billion yen from the Japanese government at the beginning, and later received a total of 7.3 billion yen of funding from 8 domestic companies such as Toyota and Sony.

To the surprise of many people, the Japanese government is also rare serious, and the Ministry of Economy, Trade and Industry has paid 260 billion yen in subsidies just to help it build a 2nm chip factory in Chitose City, Hokkaido.

What exactly is the origin of Rapidus, and what confidence does it have to shock TSMC and Samsung?

1. Lian Po is old

If you want to talk about Rapidus, you can't avoid two key figures, one is the chairman Tetsuro Higashi, and the other is the president Junyoshi Koike.

Tetsuro Higashi was not a semiconductor engineer by profession, but graduated from the International Christian University (ICU) with a master's degree from the Graduate School of Tokyo Metropolitan University, where he majored in modern Japanese economic history, and in 1977, he joined Tokyo Electron, which had only 200 employees at the time.

In an interview with Japanese media, Tetsuro Higashi said that he is neither a mechanical boy who likes to fiddle with machines, nor an engineer, before he was 10 years old, there was no telephone or TV at home, and the only electrical appliances in the house were an electric light hanging from the ceiling and a vacuum tube radio placed on a chest of drawers.

In 1996, at the age of 46, Tetsuro Higashi became the president of Tokyo Electron, and under his leadership, he became the No. 1 semiconductor equipment company in Japan and the No. 3 in the world.

In 2013, Tetsuro Higashi, then chairman and president, began to promote the merger of Tokyo Electron and Applied Materials of the United States to form a new company jointly controlled by both companies, and the combined company will significantly surpass other companies in semiconductor equipment sales and become the world's No. 1 company.

However, the acquisition has not been negotiated for two years, including the opposition of regulators in China and the United States, after this heavy setback, Tokyo Electron and Applied Materials have declined by different ranges, and then in 2019, Tokyo Electron has suffered losses for two consecutive years, according to unwritten regulations, Tetsuro Higashi was forced to resign as chairman and president.

For Tetsuro Higashi, who had experienced the glory days of Japanese semiconductors, he probably thought about how to make a comeback, and it happened that the then chief technology officer of IBM called and said: "IBM has completed the development of 2nm technology, do you want to cooperate with us and choose to mass produce in Japan?"

Tetsuro Higashi said that if Japan does not do it, other countries will do it, and this is Japan's last chance. So he approached Junyoshi Koike, then president of Western Digital Japan, and Akira Gan, chairman of the Ministry of Economy, Trade and Industry and the chairman of the Caucus of Parliamentarians for the Promotion of Semiconductor Strategy, to discuss the feasibility of cooperating with IBM to produce 2nm chips.

Unlike Tetsuro Higashi, Koike was an engineer himself, and he was fascinated by radio in junior high school, and he took the relevant qualification exams for it, and later he was admitted to the Faculty of Science and Engineering of Waseda University and received a doctorate in electrical engineering from the graduate school of Tohoku University in Japan.

In 1978, Junyoshi Koike joined Hitachi Manufacturing Co., Ltd., entered the semiconductor division, and officially started his semiconductor career, and in more than 20 years, he started as an ordinary semiconductor engineer and eventually became the head of Hitachi Semiconductor's production technology department, which can be described as one of the most senior semiconductor engineers in Japan.

In 2000, Hitachi Manufacturing Co., Ltd. and Taiwanese foundry giant UMC established Trescenti, which means "300" in Latin, that is, wafers with a diameter of 300mm.

At that time, the Japanese semiconductor industry put forward a very bold plan, what is the plan? That is, 11 major Japanese semiconductor manufacturers will cooperate with each other to jointly develop the next generation of semiconductor processes, and each technology will be gathered in a "main factory", which will be responsible for the foundry production of Japanese semiconductors, and Hitachi's Trescenti to develop the foundry business is naturally the most suitable choice.

However, the 11 companies were not as united as everyone thought, and everyone chose to develop their own 90nm technology, but the plan was ultimately aborted, and Trescenti was eventually merged into Renesas Electronics with Hitachi's spin-off of the semiconductor business, and Japan's hopes for a wafer foundry were finally dashed.

After that, Junyoshi Koike served as president and representative director of SanDisk Inc., representative director of HGST Corporation of Japan, and president of Western Digital Japan until he received an invitation from Tetsuro Higashi to experience the glory and decline firsthand, and he became the president of Rapidus without hesitation when he wanted to revive Japanese semiconductors.

It is worth mentioning that whether it is Trescenti (meaning 300) or Rapidus (meaning fast), both are from the hands of Junyoshi Koike, although the two are separated by more than 20 years, but they are both at the forefront of the Japanese government's semiconductor strategy, shouldering the mission of revitalizing the industry, and are born to achieve mass production of advanced processes, Japanese semiconductors seem to have never given up the dream of a comeback.

It's just that, there is an old saying called Lian Po Lao, can still eat, Tetsuro Higashi was born in 1949, Koike Junyoshi was born in 1952, one is 74 years old, one is 71 years old, and it belongs to the elderly players all over the world, not to mention that Rapidus is still a newly established company, not other mature enterprises, Japan pins its hopes on semiconductors on such two old people, on the one hand, it can be seen that the Japanese society is aging to a high degree, on the other hand, it also reflects the lack of semiconductor talents in Japan today.

2. Talent declines

Some people may say that the president and chairman do not represent Rapidus as a company, as long as there is capital and technology, there will be more new blood injected.

Is this really the case?

As of December this year, Rapidus has grown to 280 employees, mainly made up of former employees from other companies, and in the next fiscal year, Rapidus will hire new graduates for the first time, but even now there are few foundry talents among those 280 employees.

In August, Rapidus held an induction ceremony for more than 30 semiconductor engineers, many of whom are in their fifties, and although they have been working for many years at major semiconductor and electronics companies, they are determined to become front-line employees rather than managers, but this is probably not something that can be achieved through hard work.

In a Reuters interview, Yumi Suzuki, 55, who joined Rapidus this year, has worked in the semiconductor industry for more than 30 years since graduating from college in 1991, initially working in front-end manufacturing equipment development, and then moving to back-end processes such as packaging and testing, and working for a Japanese image sensor company for more than 20 years.

Naoto Yonemaru, 31, who joined the company at the same time as Yumi Suzuki, is one of the few young people in Rapidus who jumped into the industry after a period of stagnation in Japan's semiconductors, but his former company was a semiconductor materials manufacturer, and the lithography technology he will develop in the future is a far cry.

There are many similar Rapidus employees, they are either from the equipment field, or from the materials field, or from the design field, among which there are all kinds of executives in the original company, but there are very few in the front-end field of wafer manufacturing, after all, Elpida has gone bankrupt since then, Japan has been in the gap of advanced manufacturing for more than ten years, and now it is not easy to find the right talent.

What if there is no talent, we can only start from scratch.

In order to achieve the ambition of mass production of 2nm in 2027, Rapidus began sending engineers to IBM's U.S. R&D base - Albany Nanotechnology Park, and sent the first batch of 7 engineers in April this year, and so far, about 100 Rapidus engineers live in the park, which is divided into 6 teams in the fields of "design", "patterning (circuit formation process)" and "device (complete semiconductor evaluation)" with IBM The ultimate goal of joint R&D efforts is to bring semiconductor mass production technology back to Japan.

A group of forty or fifty-year-old employees have to learn from IBM, which has withdrawn from the field of wafer foundry, in three or four years, and complete the work that others have spent more than 20 years to complete, which is not so much an inspirational story as a satellite of the semiconductor industry, which is no different from the Arabian Nights.

Moreover, even if these engineers return to China after completing their studies and become the key force driving Rapidus to produce 2nm chips, a qualified wafer foundry still needs several times the number of grassroots engineers who are managers, and this problem is probably a little more difficult than solving mass production, because there are not so many engineers with foundry experience in Japan.

In fact, the competition for semiconductor engineers in Japan is intensifying. According to Recruit, the number of semiconductor talent in Japan has been declining since 2010, but the number of hires is increasing rapidly, with the number of job openings for semiconductor-related engineers in Japan increasing by as much as 12.8 times over the past decade.

Hiroyuki Araki, a strategy executive at semiconductor equipment manufacturer SCREEN, revealed, "From around 2013 onwards, it became increasingly difficult to recruit. Even 3D semiconductor packaging technology, which is regarded as a next-generation technology, "has no human resources" (Yasumitsu Orii, senior executive officer and general manager of Rapidus' packaging division), and in academia, "companies need semiconductor talent the most, but universities also need it" (Tetsuro Endo, director of Tohoku University's International Integrated Electronics R&D Center).

We have previously discussed the dilemma faced by Hong Kong in the development of semiconductors, and also discussed the lack of newcomers in TSMC, in this regard, Rapidus is more difficult than the first two, in the context of declining birth rate and aging population, the original plan to mass produce 2nm in 2027 with 1000 employees can be achieved, there is a big question mark.

Of course, many experts have long questioned Rapidus. Mr. Takashi Yugami, director of Japan's Institute of Precision Machining, once wrote that if an elementary school baseball boy wants to become a player in the future, he wants to play baseball like Shohei Ohtani (a player of the Los Angeles Angels of Major League Baseball). It was naturally impossible to join the major leagues after three years, so you had to play in Koshien first, then enter the Japanese professional baseball team, and finally try to transfer to the major leagues.

The gap between Rapidus and 2nm is like that of Baseball Boy and Shohei Ohtani, and it is impossible for Takashi Yugami to complete mass production in three years.

The hurdle for Japan to want mass production is "quite high" due to a lack of experience in mass production of cutting-edge semiconductors, with TSMC having a wealth of experience in technology, while Rapidus seems to rely on only IBM and a group of veteran employees in the wafer manufacturing industry, according to the consulting director of British research firm Omdia.

In the 80s and 90s of the 20th century, the electronic products produced by Japanese companies were enough to consume most of the production capacity, and now the 2nm process is so expensive, how many companies are willing to use it?

Even Hisashi Kanehashi, director of the Equipment and Semiconductor Strategy Office at the Ministry of Economy, Trade and Industry, said that while he has been encouraging Rapidus investors, the biggest challenge is how to connect with those who use the most advanced semiconductors, "starting with the automotive industry, we will create use cases for advanced semiconductors," and his comments are a little too optimistic given the development of new energy vehicles in Japan.

3. The optimism of the "loser".

Tetsuro Higashi, Junyoshi Koike, and Rapidus employees who have experienced the decline of Japanese semiconductors, unwilling to accept Japan's current marginal position in the world's semiconductor development, have gathered under one signboard to regain its former glory.

Their optimism and fighting spirit come from past experience, from the fact that they were once a member of the "losers" of Japanese semiconductors, and also from the common direction and goals of each other, Sony, Renesas and Kioxia are both Japanese semiconductor giants, which are also dominated by elderly employees, but they do not have this kind of sharpness, the emergence of Rapidus, has added a breath of fresh air to the twilight of Japanese semiconductors, and has also made many people no longer immersed in the pessimism of the past.

Rapidus does not exist as an independent enterprise like the previous Trescenti and Elpida, but the Ministry of Economy, Trade and Industry undertakes the funding of the national semiconductor company, as long as the Japanese government does not stop the financial support, Rapidus can continue to develop at an astonishing speed, from this point of view, Rapidus is more like the Japanese DRAM five in the 70s of the last century, allowing the country to ensure the competitiveness of its own semiconductors.

Is 2nm technology important, of course it is important, Intel, TSMC and Samsung are all crazy about the progress of the 1nm process, and the annual wafer factory investment alone is tens of billions of dollars.

But to a certain extent, 2nm is just a banner waved by Rapidus, Tetsuro Higashi and Junyoshi Koike, as semiconductor veterans in their seventies, it is impossible not to understand the difficulty of achieving this goal, but as long as they play the card of 2nm, they can get more subsidy funds, attract more people with lofty ideals, and give Japanese semiconductors a little more hope.

Rapidus, an alliance composed of "losers", may not be able to create a real 2nm chip, but the so-called "money buys horse bones", Japan's choice to establish Rapidus may have already been achieved.