Let's talk about semiconductors. In Japan, there's a belief that while Japan has outstanding technology, it has lost to South Korea in semiconductor competition due to management issues and diplomatic pressure from the United States. This statement is, in fact, correct. However, in various videos, I have argued otherwise. If we look at the situation, it appears that Japan has faltered not only in technology but also in management, which is essentially an excuse. The reason Japan initially thrived in the semiconductor industry was that the United States provided all the technology and offered various conveniences. Why did this happen? It was during the Cold War, with the Soviet Union posing a threat, and to counter that, the U.S. needed allies. Since it was overwhelming for the U.S. to handle everything alone, they hoped to unite with many allied countries, which led to Japan being supported and developed as an ally.

At that time, the United States provided more semiconductor technology to Japan. However, Japan suddenly developed and later tried to surpass the United States. As a result, semiconductor factories in the U.S. ultimately couldn't withstand the competition and started closing down. In 1984, the U.S. created a semiconductor chip law. Following that, in 1986, the U.S. established the Semiconductor Agreement. At that time, they imposed a 100% tariff on Japanese semiconductors and mentioned that only 10% of products made overseas could be used. Then, five years later, in 1991, the second U.S.-Japan Semiconductor Agreement was signed, which required that 20% of semiconductors used in products made in Japan had to be produced overseas.

Through that process, Japan has significantly become more withdrawn. This topic is frequently discussed, and another aspect is the ability to win through technology and management. So, usually, when we look at why Japan's semiconductor industry, which was thriving, has fallen behind, it is often attributed to pressure from the United States, suggesting that the U.S. controlled Japan, leading to a complete decline in its semiconductor sector. However, that is not the only reason. In fact, there were internal issues within Japan. 

So, what were Japan's problems? There was an excess of technology. What does excess technology mean? Until the 1980s, the mainframes were where a lot of Japanese semiconductors were utilized. Typically, mainframes are quite large computers.

When semiconductors were involved, we requested that they be made so that they wouldn't break down for about 30 years. We asked for semiconductors that wouldn't fail for around 30 years, especially for important enterprise computers. These semiconductors would be used in high-quality products. In Japan, there are also entities involved in communication, and the devices they produce are quite large. The semiconductors required for those devices demand high quality. At that time, in Japan, we often talked about materials, parts, and equipment. There are many excellent materials and equipment available, so they invested heavily to create high-quality semiconductors. They successfully produced high-quality semiconductors, but the problem arose in the 1990s when the era shifted dramatically. The age of mainframes transitioned to the dominance of PCs, which took over the world.

It's about family computers, so small computers like laptops are mainly used. The number of semiconductors in these devices is overwhelmingly greater than those in mainframes or large computers. The demand has shifted significantly. As a result, there should be a transformation, what they call giri in Japanese, but that transformation hasn't happened. When asked why it hasn't, it turns out that Japanese engineers have become accustomed to making semiconductors that don't break down for the past 30 years. This has led to the formation of a culture of extreme technology and high-quality standards. In Japan, engineers have a strong pride in their technology, believing that Japanese technology is the best, and they often talk about the spirit of craftsmanship.

I, who am quite mediocre, can create high-quality work, but deliberately making something of lower quality is simply unacceptable. Is that even reasonable? The current era cannot afford to settle for cheap alternatives. This is not acceptable. Typically, RAM and PC components are replaced every two to three years because they keep advancing in versions. Initially, the quality was poor, but it has continuously improved. Therefore, for PC users, RAM can be designed to last around three years, and it's okay if it breaks after that. If we aim to create something that lasts about 30 years without breaking, that would obviously cost a lot more. So, it is clear that using advanced technology to consistently produce high-quality products is not feasible.

This is what is referred to as a failure to transition. So, even if the product changes, the illness has not been cured; there is no change. This person, while drawing a graph, criticized Japan sharply regarding Japanese semiconductor engineers. They said things like We won in technology but lost in management, which is absurd. If this kind of perception persists, Japan will fail again in semiconductor production. What I mean is that one must accurately understand where the problem lies in order to find a treatment method, right? However, currently, Japan is talking about building rapid factories again, and that is what they are addressing. The problem is not that; if you look below, it shows that Korea has a market share with low-cost RAM. Korea has simply provided the level of quality that the market demands at a low price. Korea is precisely creating the deal that the market wants, while Japan is still...

As high-quality products are made, they are being broken down in the market. There is no need for those expensive items, so it’s not necessary to have them. If we can sell at a lower price, can we do that? The cost involved in producing them is higher, making it difficult to produce at a reasonable price. So, it has failed. 

Now, Doctor, Japan is providing huge subsidies to TSMC in Taiwan and IBM in the United States, and they are attempting to boost the semiconductor industry. How is this progressing? 

First, take a look at the number of semiconductor factories under construction globally. The number has increased from seven to eighteen from 2021 to 2023. This is more than what was built between 2018 and 2020. As you can see, Europe is also increasing its production, and let’s take a look at the United States during this time.

We're only making three, but the production has significantly increased. China, well, it hasn't increased that much, but in the past, they produced a lot. Japan has increased a little, South Korea has increased a little, and Taiwan has also increased. Overall, the entire world is making factories at this level. 

Now, if we compare the semiconductor support policies of each country, Japan is at 4 trillion yen. From 2021 to 2023, they are allocating about 4 trillion yen from the national budget to provide support. How much is it for the United States? It's 88 trillion yen. The U.S. is currently overwhelming in its investments, so Japan is looking at the U.S. and trying to follow suit. The UK is at 0.2 trillion yen, Europe is around 2 trillion, and China is about 5 trillion yen. 

Japan is currently providing government subsidies to major semiconductor companies, and this is ongoing.

Currently, the investment in Lapis is around 3.3 trillion won, and then there's an additional 670 billion won. How much is that in yen? It's slightly over 1 trillion yen. This investment is being made by investors, with a contribution of 929 billion yen from Kioxia, 235 billion yen from Micron, and 476 billion yen from TSMC for the Kumamoto factory. They are also constructing a J2 factory, which is expected to receive support of several hundred billion won, amounting to around 2 trillion won for building the factory. That's an enormous amount of subsidies. 

Now, regarding the private investments in Lapis, Kioxia has contributed 1 billion yen, Sunny Group 1 billion yen, SoftBank 1 billion yen, Denso 1 billion yen, Toyota 1 billion yen, NEC 1 billion yen, and NTT 1 billion yen as well. Additionally, UJBC UFJ Bank has invested 300 million yen. The total private investment amounts to 7.3 billion yen. How does this compare to the 10 billion yen raised?

100 billion yen, 1 trillion yen, right? Less than 1%, it’s not even 1%. It's about 0.7%. The amount invested by the private sector is only 0.7%, while 99.3% comes from the Japanese budget, funded by Japanese taxes. So why is there so little private investment, Doctor? Let’s take a look at this. The factory for Kumamoto is being built here, at 28 or 22 nanometers. Initially, they were planning to build the first factory. This technology is from 10 years ago, and we've already come this far. Currently, we have companies like TSMC and Samsung at 3 nanometers, but this technology is from 10 years ago. Now, where is Japan in all of this? Japan has only progressed to this point; Japanese engineers can only produce 40 nanometers, which is technology from about 20 years ago. Using that 20-year-old technology...

있으니까 이것도 감지덕지해야겠죠. 그런데 이게 사실은 28나노, 22나노 기술로 만드는 제품들을 보면, 대부분이 우리 주변에 있는 제품들입니다. 일상적으로 많이 쓰는 전자제품들이 다 28나노, 22나노 기술이 들어간다는 거죠. 그래서 3나노, 2나노 기술이 들어간다면, 그건 아주 고퀄리티 핸드폰이나 앞으로의 전기차, 자율주행차에 사용될 것입니다. 자율주행차가 2나노 기술을 사용하게 되면, 그게 중요한 거죠. 그러니까 그 외에는 일반적인 제품에는 다이 기술만 해도 충분하다는 이야기입니다. 그래서 드론도 있고, 그다음에 아까 언급한 공장은 여기입니다. 7나노 정도의 공장이고, 3나노 공장은 지금 그 정도를 생각하고 있다고 하더라고요.

일본 정부가 TSMC를 적극적으로 유치하고 있습니다. TSMC 입장에서 보면, 이런 공장을 짓는 데 돈을 투자하고 싶어 합니다. GSMC도 여기에 들어가는 공장을 짓고 싶어 하죠. 최신 기술 예산을 여기에 돌리고, 다른 곳에는 덜 투자하고 싶지만, 일본 은행이 그만큼 보조를 해 준다고 합니다. 여기서는 거의 45% 정도 보조를 해 주고, 다른 곳은 30% 정도 보조금을 많이 줍니다. 그래서 이제 1, 2 공장을 만든다는 것이고, 일단 1단계에서는 TSMC를 유치했죠. 그러니까 다른 나라의 공장을 유치하고, 2단계에서는 이제 자국의 공장을 만드는 것입니다. 일본의 국산화, 즉 히노마루 반도체가 결국 이루어지는 것이죠. 아까 봤듯이, 처음에는 일본의 기라성 같은 일곱 개 대기업들이 출자를 해서 그 돈으로 공장을 만들려고 했는데, 각자 10억원씩 내고 그대로 감감해버렸습니다.

There wasn't a reason for those large corporations not to pay. I think it's because they feel they can't. The large corporations are using their own money, while the money from the government comes from taxpayers, and no one takes responsibility for misusing tax money. That's right. So, the governments of the United States and Japan keep providing funding, but the corporations don't contribute. This is the situation. 

In August 2022, the Rapid Corporation was established. The Ministry of Trade, Industry and Energy provided 70 billion won in support. At that time, the companies contributed 7.3 billion won, so they were planning to start with a total of 77.3 billion won. However, they also needed to purchase factory land and equipment. The equipment from ASML costs between 600 billion yen to 600 billion won. In Japanese currency, that’s almost 70 to 80 billion yen. So, the funding for the factory is insufficient.

So, again, they plan to invest 280 billion yen in the Ministry of Economy, Trade and Industry for the year 2023. At that time, they noticed that companies weren't investing, so they quickly decided to provide an additional 260 billion yen in support. So, that makes a total of 330 billion yen. With this 330 billion yen, they are building factories and hiring employees, moving forward in the initial stages. They aim to create nano prototypes by April 2024. So, they will have completed the nano prototypes by next April. However, the factories aren't fully completed yet. This goal seems a bit challenging. Originally, for the ASML EUV machines, they need to make a third generation. The machines from the Netherlands are expected to arrive at the end of 2024, and they need to figure out how to proceed by next April.

There are concerns about this. The ASL machine is really sensitive, so can we handle that? We need to acquire the technology and then by the end of the year, we will have a prototype ready by April next year. Isn’t that too tight? It seems quite tight from a common sense perspective. So, I wonder if that’s possible. Anyway, they say they will have it ready by April, and then after that, if we succeed here, we will continue to upgrade the mass production technology and aim for mass production in 2027. However, TSMC and Samsung have mentioned that they will start production in 2025. So, there is about a two-year gap, even if Japan produces it. This is related to the same context, and what I’m illustrating here is how much money is required for the mass production.

It amounts to 5 trillion yen, but how much is it now? So far, it's been 700 yen, 2.6 trillion, 5.9 trillion, totaling around 9.2 trillion in support. The figures in the graph vary a little; earlier it was just over 1 trillion. So, this has come out, but money has to keep flowing, right? Therefore, the government needs to guarantee that loans can be made. We need to get loans from banks because companies aren't paying. Now, looking at these companies, why aren't they paying? They should be contributing 1 billion yen each. What are they trying to do? So, this is the situation. They are looking to raise additional funds from overseas, and there are talks about that. Also, since the government guarantees it, banks can provide money if the government backs it. However, in reality, Japanese banks require the company to have performance from the previous year when dealing with such large investments.

Do you have any results? Of course not. So, in the end, if you want to get a loan from the bank, there has to be government support. That's what I'm saying. Now, how is Lapis planning to create the Inano product? This is really important, right? If they can make it and earn money well, then it will become a successful investment model. 

Looking back, Japanese companies used to adopt a vertical integration model, while the current global trend is horizontal integration. What does this mean? If you look at it, semiconductor manufacturers are involved in planning, design, and both front-end and back-end processes to produce the final product. They handle everything. However, starting from the late 1990s, this began to change globally. At that time, manufacturers started to fragment, for example, like Apple.

설계만 하잖아. 핸드폰 설계만 하잖아. 그 제품을 누가 만드냐? 음, TSMC가 다 만들어. 파운드리. 어, 파운드리. 그러니까 그게 이제 파운드리인데, 여기에서 애플이 자기 핸드폰에 들어가는 반도체를 자기들이 설계하자. 그 설계해 가지고 기획하고 설계해 가지고 TSMC 파운드리 공장에 맡기는 거예요. 거기서 입만 만드는 거죠. 입만. 그게 공정이에요. 그럼 칩을 만들어. 칩을 이 한 원판에 여러 개, 뭐 많을 거 아니에요? 그걸 다 떼 가지고 하나씩 하나씩 떼 가지고 이제 어떤 선을 가지고 조립을 하잖아요. 쌓기도 하고 막 선 연결하고, 이걸 후공정이라고 그래요. 음, 그게 OSH에서 후공정에서 이게 그러니까 설계 단계, 전공정에서 칩 만드는 거, 칩 가지고 여러 가지로 이제 적절하게 제품을 만드는 과정. 이렇게 해서 하는 과정이 전부 다 분리가 돼 있어요. 음, 특화는 거죠. 그랬는데.

When asked about Japan, as I mentioned earlier, it has a vertically integrated system. In the past, there were about ten general trading companies in Japan. Within these trading companies, there are various divisions, and one of those divisions is responsible for semiconductor manufacturing. Each of these ten general trading companies has its own semiconductor production. This is somewhat like small, independent shops compared to what we see here. In contrast, the system has completely diversified and undergone mass specialization globally. For instance, companies like Mitra or Saira manage everything internally, from planning and design to both front-end and back-end processes, leading to the complete production of their products. This results in inefficiencies and higher prices. In mass production, prices tend to decrease, so this is the situation.

Since discretionary production is not possible, the prices are high. Therefore, it's somewhat manageable for domestic consumption in Japan, but exporting is nearly impossible. In the past, when Japanese semiconductors were doing well, it was because there were companies within Japan that needed semiconductors. This included large computers, and then there were electronics and home appliances in Japan that were thriving. Japanese home appliances used to be incredibly famous. Back then, semiconductors were widely used in those products. However, as Japanese electronics began to be gradually overshadowed by Korean companies like Samsung and LG, the market rapidly declined. The market that had existed disappeared. Then, Korea...

To sell in the global semiconductor market, we need to develop technology that can lower prices and create products, but we couldn't do that, which is why we ended up in this situation. Now, what is the difference between the old model and the current Lapis model? The planning is done here. The only difference is in the planning. In Lapis, we handle design, manufacturing, and post-processing to produce the final product. This process takes about three months—three months for design, manufacturing, and post-processing. Is this possible, Doctor? So, is it feasible? The large corporations in Japan are shaking their heads and not investing. They are questioning whether this is even possible. However, where is the planning done? Who is responsible for it? In our case, we are saying, Let’s create our semiconductor.

기업들이 기획을 하며 어려움을 겪고 있습니다. 그럼 이게 어디냐면, 애플은 지금 TSMC와 모든 것을 진행하고 있습니다. 이미 공급처가 있고, 모든 작업이 완료되었습니다. 퀄컴 같은 대형 기업들도 모두 TSMC와 협력하고 있습니다. TSMC는 그런 것을 많이 만들어 보았기 때문에 전문가이고, 가격도 저렴하며, 완전히 전문화되어 있습니다. 반도체의 품질도 거의 차질 없이 잘 만들어낼 것입니다. 그동안 쌓인 노하우가 있기 때문입니다. 그러나 대형 업체들이 새로운 신생 기업에 맡기겠다고 한다면, 불량품이 나오거나 문제가 생기면 큰 위험에 처할 수 있습니다. 그래서 소규모 회사들이 AI와 관련된 기업들을 실리콘 밸리에서 알아보고 있는 것입니다.

What can I say? If we consider that this Lapis is using AI semiconductors, then the customers are like this. So that could be the case. However, there are many new AI companies, especially venture companies, right? If you go to Silicon Valley, there are many venture companies. These AI venture companies are still in the early stages, so they are designing their products, but they aren't immediately producing them in large quantities. They will first try in small amounts and experiment with various things. This means they are targeting those emerging venture companies for sales. I think the market size might be a bit small. If the market size is small, it seems like even if there are a few small-scale customers, it would be fine, but I wonder if there are really tens of thousands or hundreds of thousands of AI ventures in Silicon Valley. That doesn’t seem likely.

That's right. The question is how much can be ordered from there. Even if we assume that it can be made within these three months, is it really feasible to do so? As I mentioned earlier, is it economically viable? It is said that the total cost to make this will be at least around 5 trillion won. If we invest that much, we will need to make a profit from sales to repay the investment and also to reinvest. The question is whether that level of profit will actually come in. Currently, wholesalers around the world are operating on an economy of scale, but it feels a bit like retail. Regarding AI, there are discussions about companies, and another point is that there are many companies that are customers here in Japan, such as Toyota, which recently invested 1 billion yen.

Many of the companies we invested in are included here. I see that customers will emerge from this as well. But to be honest, do you know what they are saying here? There are no products among our offerings that can use 2-nanometer technology. We don't produce high-quality semiconductors like that. We only use electronic products that cost around 260,000 won. It's sufficient to have something priced between 220,000 and 280,000 won. That's the problem. These companies aren't paying because there’s nothing for our companies to use. They can't produce products of that caliber. In Japan, they are making semiconductors for smartphones, right? If they are making smartphones, then that technology should be included. But do they even have decent smartphones? So, this is labeled as a potential customer, but there’s little chance that any real customers will emerge from this. Just look at TSMC and Apple.

Qualcomm, AMD, and Intel, as well as Sony, have secured major customers here. Therefore, I don't see any possibility of them leaving. I believe that these customers currently don't have any products that can utilize 2-nanometer technology. Toyota might use it in the future when they develop fully autonomous vehicles, but I think that is still a long way off. It could be five years or even ten years down the line; if it happens within ten years, that would be fortunate. So, if they manage to create 2-nanometer technology by 2027, will Toyota actually use it? There are also issues regarding skilled technicians. Right now, the economic situation is a problem as well, and as I mentioned, there are only about 400,000 skilled technicians in Japan. So, what will happen then?

When it comes to recruitment, you need to offer a lot of money to attract talent. As we saw earlier, semiconductor factories are being built all over the world, and they all require skilled technicians. So, honestly, how many technicians capable of producing semiconductors are there worldwide? There will be fierce competition among countries for these skilled technicians. The United States and Europe are investing heavily in building facilities to attract the best talent. However, can Japan secure enough skilled technicians? To do that, they will need to offer substantial salaries, right? The question is whether this is feasible. Earlier, the doctor enthusiastically talked about the yen depreciation, didn't he? If they want to pay a lot, it ultimately needs to be in dollars, which means that their compensation could be significantly reduced when converted to yen.

If you set a standard, then it will completely not come. Ah, then you have to invest a lot of money. So, the budget will be quite significant. The question is whether it can be economically viable. This issue is quite substantial. For now, based on the current situation, it seems that the revival of the Japanese semiconductor industry is not easy. Currently, everyone evaluates it this way; experts in Japan all see it clearly. More than anyone, it is evident that if companies do not invest, the revival of the Japanese semiconductor industry will be quite challenging. It seems difficult at present. However, from the government’s perspective, they cannot afford not to act. This is a core industry for the future; semiconductors are essential in all industries. There is no field that does not require them. If Japan falls behind, just look at the neighboring country, South Korea, which is producing a tremendous amount of semiconductors. TSMC in Taiwan is also significant. China will continue to develop in this area as well.

It will come out, Panda. Then, countries around us will all become core semiconductor production nations, except for Japan. So, Japan has materials and equipment, right? They always talk about that. They keep mentioning Japan's materials, parts, and equipment. But let’s consider this: Picasso painted a picture using amazing Sony products, brushes, palettes, and paints. Just think about the cost of the paint and brushes that went into that single painting. How much would that be, right? The final product is expensive. So, there’s a limit to the price of raw materials, components, and equipment. However, the value-added in creating semiconductor products using those materials is overwhelmingly high. The reason the Japanese government provides subsidies for something that isn’t working is that if they fall behind, it’s the end for them, because the economy itself relies on semiconductors.

I understand, Doctor. You spoke well about the reality of the semiconductor industry in Japan. I'm looking forward to March or April of next year to see if mass production of 3-nanometer and 2-nanometer semiconductors will be possible. I will keep an eye on that in April next year. Doctor, you have shared insights with our Dalchaekbang audience for such a long time. Lastly, could you please say a few words for the viewers of Dalchaekbang? 

Dear viewers of Dalchaekbang, I hope to see you many times. My comments are based on analyzing Japan and are intended to provide you with various pieces of information. In some cases, you may perceive me as criticizing or condemning Japan, but that is not necessarily the case. It's my analysis, so please take a good look and make your own judgments.

So, I provided various judgment materials, and I hope that after reviewing just a few millimeters, you can make a good judgment and that it helps you take action. Thank you. With this, we will conclude the interview with Dr. Lee Myung-cheon. Thank you for your hard work during this long time.

"삼성에 비하면 구멍가게 수준" 일본 반도체 부활 와장창 깨졌다. 구걸해도 일본 기업들도 망한다고 투자 안 한다 (이명찬 박사 / 4부)

🌕 웅달과 함께 하는 "고래 돈공부" 모임
📌 네이버 카페 - https://rb.gy/nxu5i
📌 "고래 돈공부" 신청하기 - https://rb.gy/y0y0x

🌕 JUST READ and DO IT.
"웅달 책방"과 함께 읽고, 쓰고, 말하고,
나누는 삶을 실천해 보아요.

여러분들의, 무엇보다 건강, 경제적 자유,
그리고 자아실현을 응원합니다.

🌕 웅달
현) "웅달 책방" 운영 중
전) "체인지그라운드" 창업
카이스트 경영대학원 SE-MBA 졸업
연세대학교 경제학과 졸업

🌕 비즈니스 문의 :)
cclee9@naver.com


"삼성에 비하면 구멍가게 수준" 일본 반도체 부활 와장창 깨졌다. 구걸해도 일본 기업들도 망한다고 투자 안 한다 (이명찬 박사 / 4부)

Scripts