President of Tokyo Electron Device Ltd.
[Newsweek (International版), March 25, 2022号]
Leveraging the synergy of a trading firm with manufacturing functions, Tokyo Electron Device (TED) provides total solutions that encompass leading-edge devices with a mission to drive digital transformation. We sat down with president Atsushi Tokushige, who gives more details on the company’s solutions in the fields of robotics and artificial intelligence with technologies such as TriMath, a 3D vision robot system, and the RAYSENS automatic semiconductor wafers inspection system.
Major electronics manufacturers were market leaders in the semiconductor industry in the 1980s, but in the early 2000s, their market share declined rapidly due to effects from regional competitors. Japan, however, is still very strong when it comes to functional materials and machinery for semiconductor production. What is your assessment of Japan’s current semiconductor industry and its strengths?
I think that newer competitors such as Korea, Taiwan and China took the market share for digital products, such as PCs and LCD TVs, away from Japan. Conversely, some Japanese firms have been able to maintain high market shares for a long time when it comes to industrial products, like FA (factory automation), robotics, semiconductor production equipment and automobiles. I think the difference between digital and industrial products is the number of technological elements. While PCs have relatively few technical elements, automobiles have so many technical elements that it is difficult to achieve high performance and quality. The more technological elements there are, the more alignment is required, and Japanese firms perform this very well. Furthermore, in Japan, market leaders partner with SMEs who have leading niche technologies. I think this capability of exchanging views among engineers in different sectors within the supply chain is key for product development and is also a strength of Japanese companies.
The semiconductor industry is a little different from what I’ve just discussed because we have followed what is called a ‘silicon cycle’. This is basically tied to the requests of customers—when they have demand, they place a lot of orders and you produce a lot, but when there are no incoming orders, production slows—and this cycle is repeated. Now, I believe that the US is a leader in the semiconductor industry because they are able to conduct business at their own pace.
Japan is well-known for having an ageing population and a declining birth rate. In anticipation of this trend, smart data-driven, human-centric and next-generation cities are emerging. What do you envision as the key technologies for such a trend?
As I mentioned, Japan tries to concentrate the know-how from people in different sectors, in an effort to transition the functionality of human beings to machines or for computing and learning. I think the products that Japan can bring to the world, especially through cooperation with machines, will contribute greatly to societies with ageing populations.
The ageing population is a problem that’s too big for one company to solve, but we try to contribute as much as we can through our products. We are focusing on developing products to replace the functionality of human workers with machines, not just for heavy or dangerous work, but also, more recently, for work that requires precision.
Although Japan is a high-tech society, it has been relatively slow in digital transformation (DX) in comparison to other OECD countries, being ranked 28th in the World Digital Competitiveness Ranking 2021. How do you cater your services according to the different degrees of ability to adapt to digitalisation? How do you make it accessible?
Our mission statement is “DRIVING DIGITAL TRANSFORMATION”. There are many different kinds of technologies in DX, but one of the key technologies is AI. We opened the TED AI lab are our Engineering Centre and installed two types of ultra-highspeed AI accelerators for deep learning. We offer this AI environment to customers for evaluation of how to utilise it, and they then sell it as verification systems or time-based subscription services to their customers. It is better for customers to use AI—we can support everything for the AI environment for the customer. Through this experience, we deepened our knowledge and understanding of AI and customer requirements regarding it. This is our investment in the market.
You began as a trading firm and now you offer your own design and manufacturing service, the inrevium brand. Can you explain the synergistic benefits from operating as both a trading and manufacturing firm?
We have built a customer base of over 2,500 clients domestically through our business. Almost all of our customers have their own factories and subcontractors in Japan, China or elsewhere. As I mentioned, we are focusing on developing machinery to replace human workers, especially for manufacturing customers, and therefore we are able to sell not only semiconductors, but also our private brand products to the same customers. That is our biggest synergy.
TED provides its contract design and mass production services under the inrevium brand. How does this diversification and having the private brand business help strengthen business moving forward?
Our brand business (private brand business), which is one of our unique features, is based on the idea that we should make the products that our customers need and that cannot be found anywhere else in the world. Based on this idea, we established our design development centre in 1985 and since then have been providing contract design and semiconductor substrate manufacturing services for over 35 years.
The current contract design and mass production services provide one-stop services for substrate design, development, and manufacturing by leveraging TED’s development capabilities to design and develop substrates, which are then manufactured by TOKYO ELECTRON DEVICE NAGASAKI.
TOKYO ELECTRON DEVICE NAGASAKI, which handles substrate manufacturing, was made a consolidated subsidiary of TED in 2017 so that TED would have its own substrate production function. In addition, we made FAST CORPORATION a consolidated subsidiary in 2018 to develop products for the robot vision field using image processing technology. In this way, we have spent the past few years developing the infrastructure to become a manufacturer.
In our mid-term management plan, we have set “becoming a manufacturer with technology trading company functions” as our “VISION”, and the private brand business will be the core division for achieving this vision.
The healthcare sector is shifting from treatment to prevention, with early diagnosis becoming an important factor. Why should your company be the go-to partner for ODM services regarding medical devices and software? What are your competitive advantages in these areas?
When we talk about providing ODM services to the medical industry, of course the market itself holds appeal for us, but another reason is that we are very good at high speed image processing, which is one of our strengths as a company. We can condense the digital signals through networks and also support high speed transmission of those signals through telecommunications. These fit well with the needs of the medical industry, and that’s why we suit the industry.
Usually, customers will tell you what they want, and you create what they need, but many of our orders are intended for use in medical equipment, and after having discussions with clients, we realised that they would rather focus on reagents and medicinal drug development, rather than electronics. As a result, we handle building the systems after we understand what they need, and we are shifting part of our business model to that of an ODM.
TED provides, as its own products, the CX series, which are equipment abnormality detection and failure prediction systems, and TriMath, which is a vision robot system. Can you discuss these technologies and explain how you are leading the FA revolution that is taking place worldwide?
TriMath is a work robot system that combines robots, hands, and system control with optical equipment and unique image processing and AI technologies at its core. TriMath vision enables flexible transportation and sorting of a wide variety of irregularly shaped objects, which has been difficult at manufacturing and distribution sites in the past. A single operation can be easily constructed by combining the necessary functions from the three pre-defined operations: picking, recognition/measurement/inspection, and sorting, according to the target objects and on-site operations. This drastically reduces the amount of work that has to be designed and integrated at each site and enables quick introduction and efficient system operation.
So far, it has been installed on manufacturing lines, but we want to expand this to agriculture, construction work and even the medical and nursing fields, to give just a few examples.
You adopted VISION 2025, which looks to the post-COVID era, when you have target sales of ¥200 billion. What is behind this target? Can you explain some of the short-term goals that you have set and how will you achieve them?
Last year, our VISION 2020 plan came to an end. In fiscal year 2021, we had sales of ¥170 billion yen, and our target for fiscal year 2025 is ¥200 billion. The semiconductor market is very strong this year, which is the catalyst for the increase in sales, but it will slow down at one point or another. Our biggest challenge is to sustainably increase profits and sales. Normally people say, “increase sales and profit”, but we are saying “increase profit and sales” because we prioritize the profit growth rate over sales. The way we will achieve this is through our private brand business.
What vision do you have for your private brand business?
In the private brand business, we will position the manufacturing automation segment as an important field and focus on the development of “Monozukuri-Systems”. In the outsourcing business, we will improve profitability by shifting from the conventional service of developing one part of the customer’s products on consignment (DMS) to ODM, in which we sell final products developed independently by us under the brand of the customer. In this context, we are emphasizing the development of Monozukuri Systems. Our development concept is “the fusion of image processing, data science, and robotics”, and we will promote this through collaboration among group companies and with leading companies.
One specific example is the “RAYSENS” macro inspection system. Instead of visual inspection, RAYSENS can automatically inspect semiconductor wafers. It is already in operation at a wafer manufacturer, and going forward, we will promote larger diameter wafers. The second is “TriMath”, the 3D vision robot system I mentioned earlier, which has the ability to recognize shapes in real time and is being used at home appliance recycling plants. In the future, we will evolve its functions as we accumulate experience in a wide variety of manufacturing processes.
You’ve acquired two companies through M&A—AVAL NAGASAKI CORPORATION (now TOKYO ELECTRON DEVICE NAGASAKI) in 2017 and FAST CORPORATION in 2018. What is your M&A strategy to make inroads in the ODM field?
When we conduct M&As we do not simply pursue scale. The decision is based on whether synergies can be expected with the technology owned by the target company.
You have been international since 2005 when you expanded to Hong Kong and have since further established business operations in China, Korea, Singapore and Thailand. Moving forward, where will you expand your business next?
We are focusing our foreign businesses on the international semiconductor market, with key clients in the industrial and automotive sectors. In terms of overseas expansion, the challenge is to expand into Europe.
For our private brand business, we want to strengthen business in China, as we have sales bases there. We also want to enhance the functionality of those bases to provide maintenance services, especially in the southern regions of China, where we expect demand will be high.
Imagine we come back in five years, in TED’s 40th year, and interview you again. What goals would you like to have achieved by then?
We have been called a trading company that has manufacturing capabilities. In five years, we want to significantly change our business model and be a manufacturer with trading company capabilities.