During the COVID-19 pandemic and in the post-COVID world we have seen shortages in consumer electronics and associated factory disruptions, which really highlighted to companies and governments the fragility of the global chip supply chain. We now see a concerted effort to regionalize onshore production. Between 2021 and 2023, ground was broken for 84 new fabs worldwide. If we talk about the US alone, the CHIPS and Science Act has seen 18 fabs constructed or under construction. Could you tell us what opportunities these developments are presenting in the semiconductor industry, and how you are reacting to these developments for your company in particular?
For sure, it’s a great business opportunity, and I can actually experience the huge opportunity that is presented to Japanese companies. The possibilities are expanding, and I can see that not only equipment and materials providers but also service providers are increasing the number of locations and human resources. And there are some Japanese manufacturers that now have overseas production bases as well. In the US pre-COVID, especially, it was more of an exclusive market, and now it has opened up more to Japanese companies, so Canon is currently strengthening our service and our sales capacity there. Even though there’s a shortage of labor and wages are quite high there, we are currently making investments for the future.
Japan was a global leader in chip production in the 1980s and 1990s. Since then, the government has made multiple efforts that have failed to revitalize production. Today, it’s all about onshoring, new technology development, and also collaboration. These attempts failed for several reasons, but people say there is now a unique window of opportunity for Japan. Since policymakers in the West are very concerned about the political tensions between Taiwan and China, Japan is a real alternative in many people’s eyes. Do you agree with that sentiment, and if so, why?
I agree with that sentiment, even if it’s not an easy path. In Japan, many companies lags behind TSMC, Samsung, and other global players, but the government is taking the lead, and supporting the big Japanese companies, as well as supporting collaboration so Japan can once again flourish as the semiconductor hub.
It was interesting to see TSMC open foundries in Kumamoto and Arizona more or less at the same time. The one in Kumamoto should begin operations on schedule, maybe even a little ahead of schedule, according to industry insiders. The one in Arizona has been plagued with delays, and we think it will be two years behind the original plan. Obviously, there are differences between these two. The one in Arizona will focus on nano design, and the one in Japan will focus on legacy chips. Nevertheless, it’s an interesting comparison. Why do you think these differences arose?[h1]
There are multiple reasons. One reason could be that the Japanese government’s subsidy execution was quite fast. Although the US has the CHIPS and Science Act and the Japanese subsidy was introduced later than that, the execution was faster. It could also be attributed to Japanese culture, where people are very precise and move according to the schedule, so when things are planned out, and such a structure is required, people adhere to the schedule. I think it is also an aspect of Japanese culture where the importance of collaboration is stressed.
The national government, the local Kumamoto government, and the companies have all worked together in unison to strongly promote the project and push it forward. This is just my assumption, but the US already had Intel, which is a rival company to TSMC, so having both on the same ground perhaps involved some political adjustments that needed to be made, whereas in Japan there’s no such issue. Therefore, it was easy for Japan to welcome TSMC and dedicate all its resources, including subsidies, to its development.
Sony also might have played a vital role. I went to the opening ceremony for the Kumamoto factory, and Morris Chang was there. He explained how they have had a good relationship with Sony for a long time, so I learned about their relationship then.
Another factor is that Japan already had an ecosystem of material and equipment provision, so it was easier to make progress with the project.
Over the past two years, increasing inventories and the drop in demand have seen the semiconductor industry tail off. However, the first few months of 2024 have been positive. We’ve seen foundries increase profitability and revenues. The World Semiconductor Trade Statistics predict that 2024 will see a 30% increase in the market. The long-term outlook is for a USD 1 trillion industry by 2030. Over the next 12 months, what is your take on the developments in the industry?
I expect this year to see a good recovery from the downturn last year, but I feel that the speed of the recovery is slower than expected. I think one reason is that the recovery of NAND flash memory is slow. Another reason is that the demand for smartphones is also not increasing that much. I expect maybe the latter half, the end of the year, or early next year when we will see much more investment. One of the drivers will be AI, and there’s a huge demand for Nvidia’s AI logic and high bandwidth memory (HBM). This is actually a positive driving force for our business, too. I feel that now is an inflection point, from smartphones as the major driver of demand to newer applications like AI. Although I mentioned AI, there is a growing need for semiconductors for sensors and other devices for vehicles and communication or power devices, so I have a high expectation for a big recovery at the end of this year.
As for power devices, the material currently used is silicon, but that’s now shifting to a silicon carbide compound, and although there’s talk that EVs may be dwindling, there will still be a growing demand for power devices. There are so many companies that are currently engaged in the development of semiconductors using silicon carbide, so excessive demand is a concern, but we have many products related to it so we are able to meet market demands.
Canon is very well-known for cameras and imaging equipment, but the industrial business is perhaps less well-known. You introduced the first lithography equipment to Japan in the 1970s, and you acquired Tokki in 2007 to become a leader in OLED systems. You also have Canon Machinery doing things like die bonders and Canon Anelva doing thin film deposition. Could you talk about some of these technological developments and any key technical outcomes that you have contributed to the industry?
Lithography was the epitome at the early stage, but to develop that required so many technical developments. There are too many to discuss in detail, but to sum them up, the commonality is ultra-precision technology. In particular, optics and mechatronics are our strengths, and these are used across a diverse business lineup. For example, the technology that was developed for lithography is now used in glass polishing for cameras. Likewise, technology developed in other departments is applied to semiconductor lithography equipment development as well.
The strength that Canon has is a diverse range of business lines that all have their own R&D and production capability. The outcomes or advances are shared among the Canon Group companies and applied elsewhere, creating a synergistic effect.
Lithography equipment
Canon grabbed headlines in 2023 when you introduced your new nanoimprint system. This will use a stamping process in order to transfer the circuit patterns onto the wafer. If you look at the history of nanoimprint lithography (NIL), it’s been around for about two decades, yet it has failed to be adopted at a large scale by manufacturers. Why do you think that this technology will succeed where others have failed for the past twenty years?
This technology has existed for quite some time, and there are many ways of utilizing it. However, with five nano and four nano nodes, our technology can overcome the issue of resolution and overlaying accuracy and defects—in fact, Canon is the only company that has overcome all of these obstacles. This nanoimprint technology was originally developed by a US company, Molecular Imprints, which we acquired in 2014, and we have been co-developing it with them. Inkjet plays a key role in this technology, and, of course, Canon has a long connection with inkjet technology. And we had lithography technology as well, so by amalgamating all the technologies that we had accumulated, we could create a one-of-a-kind lithography. It is especially outstanding in overlay accuracy. We can even achieve accuracy at a level of two or three nanometers.
If I go into the details of technology, I can continue talking for the whole day, so let me just say that our technology is the amalgamation of all the efforts that we have made in the past.
Without giving a very technical explanation, could you explain how you achieved that accuracy?
Some chips have as many as 20 layers, so it’s very important to accurately lay out each layer. As the foundation, we use a glass sheet for the mask, and we have four direction actuators to deform the glass by pushing and pulling it. We also match the underlayer pattern. Magnification is adjustable, but if it’s distorted, then you cannot adjustments. The technology is called highorder correction, and we developed a new technology based on that. through digital micromirror devices (DMD), we focus laser light on the wafer. DMD is the technology used for projectors, and with it, we can select parts for reflection and parts that are not reflected, so we can control where we have the laser light emitted. Of course, wherever there’s a laser, there’s heat, so applying heat through a laser creates a difference between the surface material silicon and the ceramics underneath, and as the rate of expansion changes, you can control the form of the area. So, with that technology, we were able to achieve high-order correction.
Proprietary matching system. DMD laser focusing
There are a few things that are interesting about your new technology. First, it supports five nano and could potentially go down to two nano design nodes. Second, it’s an alternative to EUV, a market that has been monopolized by ASML. Third, because it is a completely different process, it doesn’t use the high-performance light that EUV does. Yet, it remains a new technology in the sense that there are no big fabs that have it. When you go to sell this nanoimprint system, how are you going to convince customers of the business advantages of changing their processes and challenging the EUV monopoly?
First and foremost, our product is much cheaper than EUV, so that’s a good sales point. It also does not consume as much electricity, and EUV is very complex, so replacing even part of EUV with nanoimprint would simplify the process.
As an example, let’s say we want to make a grid pattern. With optical lithography, you cannot do it at once. You must do it in steps. With nanoimprint, you can print it at once because it is a stamp.
Nanoimprint equipment actually simplifies the steps and can stamp out complex geometry as well, so we’re not trying to replace or rival EUV completely. Instead, we are trying to appeal in terms of cost reduction and increased productivity.
From a technical perspective, EUV baking results in edge roughness, whereas with nanoimprint that does not happen. It may be a small thing, but the accumulation of small things matters a lot.
The technology may be good, but you have to show the advantage for the customer to introduce the equipment, so we are trying to strengthen and highlight the charms of our product.
You touched upon the environmental impact of nanoimprinting in your answer. EUV requires a huge amount of energy at very high temperatures, which is a big concern. TSMC, for example, uses 7% of all the electricity in Taiwan, which is a massive burden on the country. Could you highlight how your technology is superior to EUV for some of those environmental aspects?
Our lithography consumes one-tenth of the energy that EUV equipment requires, so it’s very environmentally friendly.
The development of nanoimprint technology is pivotal for Canon. In 2021, you launched the second stage of what you call your grand strategic transformation, which will transform Canon through M&A activity and corporate restructuring into what you call an industry-targeted business. How will that impact Canon’s industrial division, and what role will the semiconductor sector play within that grand strategic vision?
Canon Anelva, Canon Machinery, and Canon Tokki were subdivided before, but now we are all in one group, and this grouping creates a synergy by allowing us to do things that separate companies cannot do. Being under the same umbrella means we can share resources and technology and work together. By organizing ourselves in a group within the semiconductor division of Industrial Group, each company carries out its own technology, like lithography or sputtering, but we can now establish a platform for resource and technology sharing, creating synergies, and it is my mission to lead the way in strengthening these ties. As Canon’s strategy is to shift from a B2C to a B2B model, including medical, commercial printing, and network camera businesses, the role that the Industrial Group plays is vital in this evolution.
Canon has announced that under this grand strategic transformation, you will move into a model that doesn’t just offer equipment but also offers services. A good example of that is your IoT-based Lithography Plus, which goes beyond maintenance. It also includes prevention, prediction, auto recovery, auto installation, and even engineering to assist your customers to use your devices best. How important do you think it is for Canon’s industrial division to move from just equipment to services, and what strategic investments or initiatives do you have to strengthen that service solution?
In the manufacturing industry, the trend is now towards not only manufacturing but also providing a holistic aftermarket service, taking care of the equipment and thus adding new value to the business. For Lithography Plus, the plan is to increase its functionality as it grows, and it’s a common service system that we developed. Data handling is the basic functionality, and on top of that, the system itself grows, whereby you can add new features such as data visualization, process visualization, and prediction. This system adds new value to our products and aftermarket services, providing efficiency and high-quality service to our customers. This is of value to us as well as to our customers.
As new production fabs are established in Japan, we are enlarging our pool of engineers. However, in order to match the speed of transformation, so a combination of technology and a cyber/physical approach is important in providing a comprehensive service to our customers. Our plan is to continue heavily investing in the development of these service platforms, and we will apply this as a role model to other part of the Industrial Group. Part of it is now executed in the flat panel display (FPD) production as well.
Looking at the next three to five years, what do you see as the next steps in the R&D strategy? Are there any particular products, technologies or services that you’re looking to further develop?
It’s very hard to predict the next three to five years since it’s an industry that fluctuates a lot, but our mission and vision in the semiconductor lithography business is to strengthen our lithography and deposition technology and apply these technologies. Our focus is also on advanced packaging, more on the back-end processes, and I see growth in this field. I also see chiplet technology expanding. There’s also been a shift in the mindset about back-end processing towards highly sterile environments, and the reason TSMC is successful in AI chips is that they have included this back-end packaging in their in-house processes.
Fortunately, we already have our lithography equipment in many of the customers’ sites, so we want to leverage these existing relationships to learn and meet the needs of the market. Design-in is a word we coined. It’s a similar concept as market-in, but we are customer-centric so we see it from the perspective of the customers and provide whatever is needed by them.
Imagine that we come back to have an interview with you again in 2030, when semiconductors have become a USD one trillion industry. What are some goals that you have set for yourself as the director of Canon’s industrial transformation that you hope to achieve by then?
In Canon’s transformation, we want to be a strong player to support the growth of the company. I foresee our semiconductor equipment business as being Canon’s main business, and we want to be a vital part of the Canon Group in developing new technologies that could be applied by other departments. Hopefully, we will also be a profitable business that contributes to the overall growth of the company and one that plays an indispensable role in its overall development.
