Podcast: Japan’s quest for the most powerful microchip leads semiconductor revival

[Daniel Nenni]

• In the 1980s, Japan was the dominant player in the semiconductor market. But huge shifts in the way the chip supply chain operated eroded that leading-edge.
• Japan is now on a drive to revitalize its semiconductor industry. Tokyo has unlocked billions of dollars of subsidies for its chip companies.
• 
  
• At the heart of Japan’s efforts is Rapidus Corporation, a company founded in 2022 by the Japanese government and eight domestic companies to develop and manufacture advanced semiconductors.

In the 1980s, Japan was the dominant player in the semiconductor market. But huge shifts in the way the chip supply chain operated eroded that leading-edge.
Countries like Taiwan with TSMC, the world’s largest and most advanced semiconductor manufacturer, began to dominate in the production of chips. The U.S. is key in areas like design. While the Netherlands has ASML, a company that makes tools required to manufacture cutting-edge chips.

Japan is now on a drive to revitalize its semiconductor industry.
Tokyo has unlocked billions of dollars of subsidies for the sector. At the heart of its revival efforts is Rapidus Corporation, a company founded in 2022 by the Japanese government and eight domestic companies to develop and manufacture advanced semiconductors.

Rapidus Corporation aims to manufacture 2 nanometer chips by 2027, as it looks to catch up to TSMC and Samsung.

In this episode of Beyond the Valley, Tom Chitty and I are joined by Fei Xue from the Economist Intelligence Unit (EIU), to talk about Japan’s efforts to reignite its domestic semiconductor industry.

If you have any thoughts on this or previous episodes, please email us at beyondthevalley@cnbc.com.

You can subscribe to “Beyond the Valley” by clicking the links below to your chosen platform:
Apple Podcast
Google Podcasts
Spotify

Here is a transcript of the “Beyond the Valley” episode released on August 6, 2024. It has been edited for clarity and brevity.

Tom Chitty
By now, you’re probably aware of the importance of semiconductors. They’re in everything from smartphones to running shoes, and are a vital component in the development of Gen AI. But you may not know that back in the 1980s Japan occupied more than half of the global semiconductor market. Since then, however, other countries have led the way as they scramble to secure their chip supply chains. Now Japan, with the backing of both government and business, is planning to revitalize their semiconductor industry and have begun a quest to develop the world’s most advanced microchip. Our guest this week is a senior analyst at the Economist Intelligence Unit. Created in 1946 the EIU, as it’s otherwise known, is a research and analysis division of The Economist Group, the sister company to The Economist newspaper. He focuses on North East Asia, providing forecasts and analysis on economic development, public policy and International Relations in the region. Fei Xue, thank you for joining Beyond the Valley.

Fei Xue
Thank you. Glad to be here today.

Tom Chitty
Well, let’s get cracking for our listeners. Summarize how important semiconductors [are], I did at the start, but just to give an understanding of why people should care about what we’re talking about today.

Fei Xue
Essentially semiconductors [are] more like a commodity nowadays, and we use this very small component everywhere in not only electronics devices, but also, like you said, even running shoes. We now use that to measure things. And eventually, I think, because the moving trend of the world is a connection of everything together, we need semiconductors for that to be achievable. And semiconductors, largely, there are two groups of it. One is logic chips, the other is memory chips. And logic chips is what we call processors, and memory chips is where we store data. Yeah, the two biggest segment of the semiconductor markets and each segment has their dominant powers in it.

Tom Chitty
So before we get back to it, Arjun, stat of the week.

Arjun Kharpal
The stat of the week is 97 billion U.S. dollars.

Tom Chitty
Let’s talk about specifically then, Japan. Give us a little bit of background on Japan’s semiconductor industry.

Fei Xue
Japan used to be the pioneer of global semiconductor research and development. At the end of [the] 1980s, Japanese companies accounted for like 51% of the global semiconductor market. Why? Because in 1980, a consortium of Japanese companies made a breakthrough in terms of creating or developing the electron beam lithography technology in terms of making semiconductors much easier and at a cheaper price, and also with the help from state fund and financing powers, they quickly take up most of the global share of semiconductors and become a powerhouse in that area. But also, like you said, after the 1980s starting from 1990s their market share quickly descended, and right now their collective market share is like below 10% in the world. And most of the semiconductors made in Japan are either specialty semiconductors, related to especially automotive industry, or legacy semiconductors, which is bigger than 28 nanometers.

Arjun Kharpal
Fei, what [were] the changes [that] happened that meant that Japan lost some of its leadership? I know part of it was political, right? I think under the Reagan administration in the U.S., they imposed some very big tariffs on Japanese memory semiconductors, which effectively hurt. But were there other reasons as well?

Fei Xue
Two big reasons. One is political, as you mentioned, the 100% semiconductor tariffs imposed by the U.S. But also from U.S., in 1986 the U.S. and Japan signed the U.S.-Japan semiconductor agreement. Essentially it set the bottom limit for Japanese-made semiconductors’ price sold in the U.S., but also it forcefully increased the share of foreign semiconductors sold in the Japanese market. So by doing this ... one thing undercut Japanese semiconductor chip makers’ competitiveness in the global market. Another thing [is] that [it] forced open [the] Japanese semiconductor market to foreign players, and this creates opportunity for the U.S., South Korea and Taiwan. Speaking of the latter two, around 1990s there is a generational evolution in semiconductor development – the specialization in sector. So traditionally, one single company, a big company, takes care of both chip design and also the manufacturing, which is the traditional Japanese model. But starting from 1990s specialization made it possible for one company to focus on chip designing, which is a strong point for U.S. companies. And some companies started to just focus on manufacturing. So essentially, TSMC, the Taiwanese chip maker, is just a contractor, they focus on manufacturing semiconductors designed by whoever are the players, and that gave all these companies the opportunity to leapfrog Japanese companies by investing heavily into a certain specific segment or sector in this industry, and many of these Japanese semiconductor makers, they could not catch up, with a huge amount of fixed investment and government support to the industry also declining a little bit. So that’s why Japanese chip makers gradually lost competitiveness in the global market.

Arjun Kharpal
So just to summarize, I mean quite a few things there, but it was one; that the politics ... the sort of decline of investment, but also the increase in competitiveness of foreign players who are allowed to enter the market. But also, I think, one of the biggest shifts, and you heard it, and if our listeners ever come across, is the fabless model, right? The rise of the fabless model. You probably hear these days about Taiwan Semiconductor Manufacturing, Co. or TSMC, who is the biggest semiconductor manufacturer in the world. They’re a big beneficiary of this move for companies that wanted to not manufacture their own chips, but actually just design them and outsource that manufacturing to TSMC and others as well. And so that’s one of the big shifts. What is happening now then Fei, in terms of Japan’s bid and desire to revive its own industry? What are they doing?

Fei Xue
They are adopting a very comprehensive approach, I would say. And it’s again, [a] state-led initiative pulling together private sector resources and development capabilities and [on the] one hand, they want to step into the chip designing sector with collaboration program with IBM in the U.S. But also they want to attract international investment in terms of semiconductor making or chip making, and to establish fabs in Japan’s territory. And they have achieved a certain degree of success in early stages. They have attracted TSMC to establish fabs on the Kyushu island, Kumamoto Prefecture. And TSMC has already said that they will set up another, the second fab there. And also eight Japanese electronic companies put resources together to create a consortium called Rapidus with state funding support. And this consortium aimed to be able to make the most advanced semiconductors, two nanometers semiconductors by 2027.

Tom Chitty
For our listeners that don’t know the difference between two nanometers and three nanometers, and I believe that 40 nanometers is kind of where Japan’s semiconductor industry is at the moment in terms of its producing of the best microchip it has, but the leading microchip is three nanometers. So when someone says two nanometers, it’s just like, one nanometer better, but it’s vast, yes?

Fei Xue
Yes. The difficulty of miniaturizing the semiconductors increase exponentially by generation. It requires a lot of investment, a lot of research and development, and so it’s hard. Not to mention that actually, according to Rapidus’ latest plan to be able to make two nanometer semiconductors by 2027, they will already be behind the most advanced cutting edge semiconductors at that time, because TSMC and Samsung already have plans to make 1.5 nanometer semiconductors by 2026. Another point is that actually some Japanese companies are trying to bypass this and to use a different set of technology, like on the back end, in terms of chip packaging, advanced chip packaging, so to stack chips together or to incorporate different types of chips to improve the performance of processes. So they are trying to make [the] most advanced or miniature semiconductors on one hand, but on the other hand, they are actively seeking [a] different approach to improve a chip’s performance.

Arjun Kharpal
Fei why do we need to keep going smaller and smaller in terms of nanometers?

Fei Xue
Because it will increase the usage or the usefulness of semiconductors, and also it will automatically improve the performance of the processors.

Arjun Kharpal
I think one of the other fascinating things you said was that even when Rapidus, if it manages to commercialize two nanometers in 2027, it will still be behind the leading edge from TSMC and Samsung, etc. That’s really the difficulty, right? For a lot of countries, they still, as much as they want to boost their own domestic industry, they’re still going to rely heavily on production from TSMC and from Samsung, just because of, naturally, the amount of R&D these companies put in over the years, but [also because of] so much of a head start they have as well.

Fei Xue
Yeah, definitely. But to be fair, for Japanese companies, they are playing the catch-up role. They don’t necessarily need to be able to make the most cutting-edge semiconductors, because there will be a huge market for legacy chips, and also the chips they are making can provide to domestic demand, and that will be a huge market for them to explore. And another issue is that even though TSMC and Samsung can make the most cutting edge 1.5 nanometer semiconductors by 2026, their capacity won’t catch up to the demand. So there will still be a huge demand for even two nanometers semiconductors.

Arjun Kharpal
I think a lot of the semiconductors, at the moment, the cutting edge ones, they’re going into high-end smartphones, right? When you think about some of the smartphones from Apple, from Samsung, as well, some of, I think Nvidia’s GPUs are on slightly more advanced nodes as well. But actually, there’s still a big market for legacy nodes, right? When you think about cars and various other areas, they don’t need, you know, three nanometer chips necessarily in them.

Fei Xue
No, not necessarily. Right now, I think for a lot of car specialty semiconductors, 28 is good enough.

Tom Chitty
You mentioned earlier that this has been driven, this revival by the government, largely with support from business. How much of that is a burgeoning industry that we need to be, you know, leading, but also national security concerns?

Fei Xue
Yeah, it’s a very good question, because these government Initiatives are not just targeting economic growth. Of course, by reviving the semiconductor industry, [the] Japanese government wants to provide another source of economic growth to the country by facilitating the development or revival again of supporting industries and technologies. However, semiconductors, like I said previously, has already commoditized in the global market, and it’s become so critical to many advanced technology areas, including artificial intelligence, including various electronic devices and even EVs, and it’s [an] essential part of the global technology race. And all countries want to, many countries want to gain space in this niche area and also, they don’t want to rely too much on foreign companies to do that, so they would like to establish domestic companies to at least be able to create and design manufacturing semiconductors for domestic uses. At the moment, we can see that the most cutting edge semiconductors, 90% of these are manufactured by TSMC, or Taiwanese players alone. The others are made by Samsung from South Korea, and a small amount made by the U.S. companies. So Japanese companies, if they require these cutting edge semiconductors, they have to rely on foreign supplies. That’s a national security issue.

Part II next......

 

[Daniel Nenni]

Part II:

Arjun Kharpal
Fei the geopolitics, or the semiconductors, have been so dragged into geopolitics over the past few years, and there’s been so many different things happening, involving in particular the U.S. and China, with many other key semiconductor players brought in to this geopolitical mix. The U.S. has embarked on a strategy of three things, it seems. One is trying to cut off China’s access to certain critical tools, selling critical semiconductors. The second is trying to invest, via the Chips Act, into its own domestic semiconductor industry, and the third seems to be striking up partnerships with allies, the likes of Japan, South Korea and other places in order to secure supply chains, work together on semiconductors, etc. But at the same time, the U.S. is more recently, there have been reports where they’re suggesting they may expand some of the restrictions on China through a rule known as the foreign direct product rule, which would potentially mean some Japanese firms, even Dutch firms, may not be able to export some semiconductor equipment out to China. So it feels like between this battle between the U.S. and China in semiconductors, there’s a lot of countries, including Japan, that, in many instances, could be caught in the middle, where, at the same time they’re trying to have sort of alliances with the U.S. How does a country like Japan navigate these difficulties on the geopolitical level?

Fei Xue
Geopolitics is always a concern for the semiconductor industry. Actually, the restrictions imposed by the U.S. has already been in effect since last year, and by now, Japanese companies cannot sell the most cutting-edge semiconductors or chip making tools to Chinese companies. And interestingly, before the restrictions came into effect, there was a huge surge of Japanese exports to China especially in chip making tools, because Chinese companies want to avoid the cost of these restrictions, and also they are trying hard to build its own semiconductor industry and to achieve self sufficiency. However, back to Japan. [The] Japanese government has already introduced a set of restrictions on semiconductor exports to unfriendly foreign countries. That includes over 25 different components and materials critical to chip making process and Japanese companies because of these increasing geopolitical rivalry between U.S. and China and technological war and trade conflicts certainly will lose a part of the market. However they will explore or actively explore other parts of markets, especially given that many countries are trying hard to reshore or establish chipmaking capacity within their own soil or in friendly countries, and that will increase demand, not to mention the continued technological advancement in AI, in EV and data centers. These will all amplify the demand for semiconductors and chip making tools.

Tom Chitty
Let’s talk about AI, then. How much of this revival is an urgency to be a leader in generative AI for Japan?

Fei Xue
For Japan in terms of the semiconductor sector, AI needs two kinds of semiconductors. Back to the basics; one is the logic chip, processors like the one right now, designed by Nvidia. Japanese companies, unfortunately, have no capacity at the moment to make them. They want to be able to make them by the end of this decade so that they can benefit from the surge of AI demand. On the other hand, AI development and cloud computing needs a huge capacity of data centers, which requires a large amount of memory chips. And Japanese companies are trying to both develop their own capacity but also attract phone companies to establish fabs there. So Micron, the U.S. chip maker, and also Samsung, has already established plans to set up either research centers or fabs in Japan to make advanced memory chips there. So the plan for Japanese companies is to catch up in this area gradually.

Arjun Kharpal
I thought one of the most interesting stories that kind of underlines Japan’s attempt to revive this industry actually involves one of its most oldest and traditional companies, Canon, obviously known for printers, known for cameras. I think last year, they said they’re trying to develop what they dubbed a nano imprint lithography system to rival ASML. Now we’ve spoken on the podcast before, a lot about ASML and their extreme ultraviolet lithography machines. They effectively have a monopoly on this market, only one in the world that really can do these systems required to manufacture the most advanced chips in the world. This Canon story is fascinating, as I mentioned, one of the oldest, well-known tech companies in Japan trying to reinvent itself as well. How likely is it going to be able to do that? But is this also something we’re going to see other more traditional Japanese companies try to do as well? Japan has such a rich history in innovation in technology. You think of the brand names that still survive these days, Sony, Nintendo, even Canon as well. What’s happening in that area?

Fei Xue
Absolutely, I mean, Japan still holds a huge advantage in many niche areas in semiconductor sector, and Japan is an active player in the precision equipment making and a lot of semiconductor materials. And back to the Canon. This nano imprint lithography technology is not new, actually. But they have somehow improved this technology so that it can be mass commercialized. However, I would say that they need to first achieve a certain degree of success rate so that they can be competitive to the existing leader in this area. Another thing, many existing fabs design their layout to accommodate to the extreme ultraviolet lithography technology. So some modifications might be needed, and this (is an) actual cost for the existing chip makers and so they will need to surpass this hurdle as well. According to Canon, the main competitiveness lies in cheaper production cost, and also less energy hungry. So that’s two main attractiveness for their technology. However, I would say in general, surpassing the existing players is one thing, but more likely Japanese companies will use their niche technology and manufacturing techniques to collaborate with existing market dominant players and to explore market share and also to try to develop new technologies. One example is Sony, which actually is a big reason why TSMC decided to establish a fab in Kumamoto, is that they want to collaborate with Sony to develop image or next generation image sensors. And Sony, of course, has its own technology there. But by pooling together these two companies’ resources and technologies, they want to increase their competitiveness against South Korea’s Samsung Electronics. So by collaborating with international companies, Japanese companies can leverage their existing technology and manufacturing techniques to expand their global share as well.

Arjun Kharpal
So we’ve spoken about Japan’s historical strengths, the way that the global semiconductor landscape changed and now attempts at the revival. And you’ve mentioned also the fact that Japan has niche areas of semiconductor. Where is it particularly strong at the moment? Because we have spoken again on this podcast about how complex the supply chain is for semiconductors, from software, design, manufacturing, the materials in between, testing, packaging, everything like that. Where do you see Japan’s strength at the moment? I was reading some stats saying it holds an 88% market share globally for coaters, developers like Tokyo Electron and Screen Holdings, 53% for silicon wafers, 50% for photoresists. These are all very technical. So what do these all mean? Where are Japan’s strengths?

Fei Xue
Japan’s strengths lies in niche materials, components and precision chip making tools and photo resist is a good example. Not only do Japanese companies, including one called Shin-Etsu Chemical, they dominate the global market share of photoresist as a material which is used in lithography technology and also the coating technology, or coating technique to coat the photoresist around a wafer. It’s dominated by Japanese companies. So they hold up some niche areas or segment in the long semiconductor supply or technology chains, and they can utilize these kind of materials and chipmaking technology and tools to their advantage.

Tom Chitty
Fantastic. All right, well, that’s all we have time for. But if, if you’d like to play stat of the week, I’ll leave it to Arjun. He’d like to be a game show host one day.

Arjun Kharpal
One day, that’s the dream. So, contestants, 97 billion U.S. dollars is the stat of the week. Fei, since you’re our special guest today, I’ll give you the first guess.

Fei Xue
Is that the cost required to make fabs by TSMC?

Arjun Kharpal
Your guess, Tom?

Tom Chitty
Estimated total investment that the government of Japan say will require to make two nanometer chips.

Arjun Kharpal
You’re both wrong. I mean, those are far more interesting than my stat, to be honest, if we knew the stat for that. It’s the market cap as of today, of recording 1st August 2024, of Tokyo Electron, which is one of the big Japanese semiconductor firms.

Tom Chitty
Fei, thank you so much for joining us. That’s it for this episode. Before we go, please follow and subscribe to the show and leave us a review if you’d like. Thank you, Fei.

Fei Xue
Thank you, Tom. Thank you, Arjun.

Tom Chitty
Thank you, Arjun

Arjun Kharpal
Thank you, Tom.

Tom Chitty
We’ll be back next week for another episode of Beyond the Valley. Goodbye.

TV
Squawk on the StreetWATCH LIVE
UP NEXT | Money Movers 11:00 am ETListen

 

[Daniel Nenni]

It is hard for me to take Rapidus seriously. They will be late on 2nm, they do not have the capacity for the whales, and the foundry ecosystem is just getting started. I'm sure they will get the Japan 2nm business but TSMC has such a huge advantage in all regards I just don't see it. Someone please convince me otherwise because I would certainly like them to succeed.

 

[MKWVentures]

It is hard for me to take Rapidus seriously. They will be late on 2nm, they do not have the capacity for the whales, and the foundry ecosystem is just getting started. I'm sure they will get the Japan 2nm business but TSMC has such a huge advantage in all regards I just don't see it. Someone please convince me otherwise because I would certainly like them to succeed.

I think I have made my position known.... from what I hear

1) they are getting their tech from IBM. IBM is not a semiconductor Fab manufacturer. they are a Lab
2) They made claims on single wafer processing and fast cycle time. This only works in theory
3) Their timelines seem to be low confidence. they had very low headcount for a company planning pilot line next year
4) they are "jumping in" with 2nm GAA as first node. Seems like 7nm would be better start


I would expect repeated delays with hype about what is coming "2 years from now" for the next 5 years. I also hope to be wrong. seems to be a research fab

IF they show something real (Large 2nm SRAM wafer and characterization data) in 2025, then I will be wrong.

 

[milesgehm]

Fei Xue
Japan used to be the pioneer of global semiconductor research and development. At the end of [the] 1980s, Japanese companies accounted for like 51% of the global semiconductor market. Why? Because in 1980, a consortium of Japanese companies made a breakthrough in terms of creating or developing the electron beam lithography technology in terms of making semiconductors much easier and at a cheaper price, and also with the help from state fund and financing powers, they quickly take up most of the global share of semiconductors and become a powerhouse in that area.

What is this person talking about?