TOPPAN : Q&A of Electronics Business Strategy Briefing

(Held via webcast)

Q: Could you once again explain your strengths? While your overall market share in FC-BGAs is not particularly large, page 11 of the presentation slides shows that you hold the number three share for high-end switch applications. Could you please elaborate on what aspects of your products are strong, and what customers value in your FC-BGAs for high-end switches?

A: Our strength lies in applying high-speed transmission technology used in the communications domain to FC-BGA substrates for high-end switches. Our customers recognize the value of that capability, and even though our overall production scale is not large, we have been able to secure the number three share for FC-BGAs used in high-end switches.

Q: Could you talk about your growth outlook? On page seven of the presentation slides, your semiconductor-related sales are shown with a CAGR of 26%. Meanwhile, on page 14, the CAGR of the high-end segment within FC-BGA substrates looks closer to around 20%. Could you explain the background behind why your sales would grow faster than the market segments you're focusing on?

A: The areas we intend to pursue, as I mentioned earlier on page 10 of the slides, are primarily high-end switches and AI applications. The growth of those areas is relatively large compared with the overall market. By concentrating on such segments and leveraging our technological strengths, we expect to achieve a relatively high growth rate for sales.

Q: Looking at your operating margin, it is 26% for FY2025 and 30% for FY2030. Given that the proportion of AI-related business will increase from next fiscal year onward, I get the impression that margins might rise close to 30% as early as FY2026. But after that, it seems margins may not rise much further. How should we view that?

A: We are currently building new lines in Niigata and Singapore. Once production begins, depreciation will also begin, so margins will not jump up all at once. As those lines gradually fill, overall profit and the operating margin will rise over time.

Q: Regarding the JPY270 billion figure projected for semiconductor-related sales in FY2030, could you explain what you envisage in terms of the proportions of your focus areas of high-end switches, AI ASICs, and CPUs? Could you comment on the relative growth expectations for each and how much contribution you expect from next-generation areas within that JPY270 billion? And how much have you factored supply capacity, such as Singapore and Ishikawa, into the projection?

A: The areas we intend to target going forward are high-end switches and AI ASICs, but it is difficult to estimate the exact composition for FY2030, so I would like to refrain from commenting.

As for market growth, high-end switches are expected to grow about 1.6 times from FY2024 to FY2030, and server CPUs about 1.8 times from FY2025 to FY2030, according to published estimates from market research firms. We estimate that our growth will generally follow those trends.

Regarding the JPY270 billion for semiconductor-related sales in FY2030, we estimate that roughly 80% will come from semiconductor packaging (FC-BGAs and advanced packaging). We expect FC-BGAs to account for more than half of that.

In terms of supply capacity in FY2030, after the startup in Singapore in late 2026, the plan includes capacity up to the contribution of mass production at the Ishikawa Plant from FY2028 onward.

Q: About the near term, you mentioned that qualification for AI switches and AI ASICs is progressing smoothly. Specifically, when in the next fiscal year will this begin contributing to your results? And what risks could delay or prevent approval?

A: We expect approval to come around February. From that point, factoring in lead times, we believe the contribution will start from the beginning of next fiscal year. We believe the risk of not receiving qualification is extremely low.

Q: On photonics-electronics convergence, you showed that unit prices for network switches increase over time, and I believe the biggest driver of those rising unit prices is CPO (Co-Packaged Optics). My understanding is that your company has been working on CPO from the earliest stages of development. I want to confirm that with optical waveguides and new optoelectronic conversion devices being added onto the substrate, it is correct to understand that the substrate size becomes quite large. If so, how big are substrate sizes likely to get?

A: Substrate sizes will increase, but they vary depending on the type and number of optoelectronic devices mounted, so it's difficult to specify what substrate sizes will be.

Q: Industry experts are saying that CPO will be essential for 3.2 Tbps cables and beyond. Could you comment on the specific timing of CPO adoption, mass production, and launch? Could you speak a bit more, in technical terms as well, about photonics-electronics convergence?

A: We expect CPO-based FC-BGA substrates to appear in the market starting from switches with performance at the 102 Tbps level. I would prefer not to comment on the specific market launch timing.

Q: If you have, say, 16 or 18 optical engines mounted, compared with today's 51.2 Tbps generation, would it be wrong to assume something like three to four times the size?

A: I don't think it would reach three times.

Q: When a design becomes CPO-based, is there a possibility that the number of wiring layers will increase? Or should we assume the change will not be that significant?

A: The number of layers depends on the semiconductor chip that sits at the center of the package. For example, in the case of an FC-BGA substrate with an interposer, the layer count will increase.

Q: With regard to page 15 of the slides, presenting your thinking on production capacity expansion going forward, I understand that the Niigata line has already been successfully brought online, and the next one is Singapore at the end of next year. You are increasing your production capacity, but my sense is that demand will be very strong, with growth exceeding your planned capacity increases. My understanding is that customers may already be pressing you to accelerate production launch plans. Is that correct?

If that is the case, in terms of room for further expansion in the future, could the second Singapore line possibly come as early as 2027? Or might you add additional FC-BGA lines in Ishikawa as well?

A: About AST (new Singapore Plant) Phase I, construction of the building is progressing smoothly and we are starting to install equipment in preparation for the opening ceremony next summer. Given that market demand is currently increasing, we want to bring the facility online as early as possible. At this moment, we are targeting operational start-up toward the end of 2026.

Next, regarding capacity expansion, launch of the new Niigata line is running smoothly. AST Phase I is, as noted, scheduled to begin operation at the end of 2026. After that, we plan to establish a new FC-BGA line at the Ishikawa Plant. This will not simply be an extension of Niigata's existing capability, rather, we are designing it as a new line for advanced packaging. Our current plan is to begin mass production at Ishikawa sometime after FY2030.

Q: If I understand correctly, even if the Ishikawa Plant could be used to handle additional demand, it would not be like Niigata's second or third lines, but rather something closer to next-generation processes, and you would handle any upside demand through that. Is that right?

A: Yes, exactly.

Q: From the business division's perspective, with demand being this strong, I imagine you would like to secure future capacity and invest early, but from the overall perspective of TOPPAN Holdings, there may be concerns about investing too far in advance, creating a tension between stepping on the accelerator and applying the brakes.

A: Our strategy in this area is aligned between TOPPAN Holdings and the business division. Fundamentally, even though our FC-BGA market share has historically been small, we have built our business by specializing in high-end, large-size, high-layer-count applications where technical and customization capabilities matter most. This focus has enabled us to maintain a high-profit business model, and we do not intend to change that strategy. So even if demand increases for ASICs or switches, our strategy is not to expand capacity blindly in an attempt to increase market share. The next line will be launched as a new, higher-value-added line that differs from the existing ones. Our approach is to consistently stay at the technological forefront of the industry, build our competitive advantage, and develop a high-profit business.

Q: You have several competitors, but why can't they replicate these strengths? And to reconfirm, what is behind your ability to grow both sales and profit?

A: Our strategy is fundamentally about establishing solid technology and then advancing business while remaining fully aligned with our partners. Another strength is our ability to build trust with partners by consistently delivering technology development capability, quality, and cost performance accordingly.

Q: In view of the capacity expansion, is there discussion of any type of support from customers, such as financial support?

A: We are not in a position to comment on that. But our priority is to manufacture products that satisfy customers in terms of quality, technology, and cost.

Q: On page 13 of the slides, within the AI accelerator domain, there are two types: AI ASICs and GPUs. Why did you choose AI ASICs? Is it influenced by factors

such as how well your technology fits or capacity constraints? I would appreciate more background on this.

A: On page 12 of the slides, we show the focus areas, but rather than extending our existing technologies, we intend to focus on AI ASICs and high-end switches, areas where, through new technological development, we can compete at the high end and demonstrate clear advantages. We intend to concentrate on areas where we can achieve high profitability.

Basically, the more customization involved, the better the profitability. With highly general-purpose products, scale is large but there are many competitors and fierce competition. Our view is that harnessing customization technologies to deliver exactly what customers want will be the source of profit. In that sense, the business will not continue to be viable without those technological capabilities, so we will keep strengthening in that area.

Q: I'd like to ask about the synergies that can be generated with Tekscend Photomask now being accounted for under the equity method. Given the focus areas of semiconductor packaging discussed today and the evolution of the technology road map, I assume the ability to exchange information from customers is the most important synergy. Is that the case?

A: There are synergies in terms of being able to exchange information. As semiconductors become more advanced, semiconductor packages also need to be miniaturized and have higher densities, and the boundary between front-end and backend processes is disappearing. In that sense, information exchange with Tekscend Photomask, which operates at the front-end, is highly beneficial.

Q: I'd like to ask about next-generation substrate technologies and your thinking on the investment burden associated with them. With the adoption of the damascene process for wiring patterns and formation of lines with widths below 1 micron, the fabrication process itself is becoming increasingly front-end-like, which prompts concerns about the investment burden rising. I assume you will basically be able to recover this with pricing, but how much greater is the investment burden likely to become?

A: As you correctly inferred, because we will be adopting the damascene process used in semiconductor front-end manufacturing and moving to a level where line width and spacing are below 1 micron, existing FC-BGA exposure tools will not be applicable. Since the process becomes much closer to front-end manufacturing, the capital investment burden will indeed be heavier than for FC-BGA thus far. However, because our policy is that ensuring profitability is the first condition, we will not invest until we have established a process that ensures profitability and carefully confirmed pricing arrangements with customers.

And we will clearly separate development investment from mass-production investment. Development investment has already begun. As shown on page 34 of the slides, this development phase has been selected by NEDO. We will advance development in collaboration with multiple stakeholders, and our priority now is to refine the technology thoroughly. As for mass-production investment, including funding arrangements, we will determine the appropriate structure in the next phase.

Q: With regard to AI ASIC demand, which is very strong, roughly how many customers do you expect and how much of your capacity might be filled next year or further out?

A: We are currently in the process of obtaining qualification for AI ASICs and are nearly at the point of receiving approval. Production should begin next fiscal year. We expect multiple customers to qualify our products.

Q: With respect to the content in the bottom-right section of slide 15, showing business scale expansion, the timing for the second Singapore line is listed as undecided, but under what conditions would you be able to make that decision? For example, if qualification for ASICs seems likely to be obtained, is it possible you might proceed with investment in Singapore before the Ishikawa Plant?

A: AST is a joint venture, so how the second line would be utilized is something to be discussed going forward. The Ishikawa Plant, on the other hand, will be decided based on our technology road map and our discussions with customers. Comparatively speaking, AST has more constraints, and we will proceed in discussion with our customers.

Q: Products geared towards focus areas accounted for 83% of the business in FY2024, but what was the situation in H1 of FY2025? And looking ahead to FY2027 and then FY2030, do you expect that proportion to increase? What will the impact on margins be then?

A: In Q1 the proportion of products for consumer applications increased somewhat, and overall profit declined. From Q2 onward, the proportion for high-end switches, servers, and AI applications is gradually increasing. From FY2026 onward, we expect the consumer share will decrease further. AI ASICs will begin contributing from FY2026, so that ratio should also rise.

Q: What yen-per-dollar exchange rate are you assuming for the sales and operating margin figures on page seven? Also, how sensitive is your 2030 operating profit to changes in the yen-dollar exchange rate?

A: We built the plan assuming JPY145 per USD, and the calculation assumes the same rate through FY2030. In terms of FX sensitivity, currently a JPY1 depreciation in the yen increases operating profit by roughly JPY500 million. About 80% of that is reflected in the Electronics segment.