Q&A at Analyst’s Meeting (June 12,2019)

Once we sell ABF to customers, we are not sure what application it ends up being used for, so I will answer your question based on estimates derived from information we get from customers. We are projecting equal growth in demand for server applications and network applications, including 5G. We refer to network applications as ASICs (application-specific integrated circuits). We do not anticipate much growth in demand for PC and communication hardware applications, but demand for social infrastructure applications will grow equally.

(Q: With respect to 5G-related applications, are you concerned that trade tensions may dampen demand from major base station suppliers? Would such a scenario, were it to arise, pose a certain degree of downside risk to your sales?) We are closely monitoring the news on a day-to-day basis given concerns about US-China tensions, but based on the situation as it currently stands, we do not envision growth in demand for social infrastructure applications grinding to a halt. While incoming news may seem quite bearish for demand, we forecast demand based on information from the PCB (printed circuit board) makers to which we actually sell ABF and set production quantities accordingly. Based on such information, US-China trade tensions may trigger a short-term dip in demand but we see little likelihood of a major demand downturn.

Strictly speaking, there may be differences, but with demand from end-user semiconductor makers growing rapidly, they seem to be ordering to Japanese, Taiwanese, and other PCB makers equally. Looking at incoming orders, we do not see any major differences between Japanese and Taiwanese PCB makers, except that highly specific applications have so far been confined to Japan or, conversely, perhaps they are being requested only in Taiwan. Based on ABF shipment data, we do not see much difference between Japanese and Taiwanese PCBs for major applications.

 

 Let me start with your third question. Although we anticipate substantial growth in ABF demand, we outsource the coating processes. Our capital investment is therefore limited to varnish production processes only. We aim to remain an asset-light producer by manufacturing varnish very efficiently. We are set up to be able to meet growing demand without investing heavily in production capacity and expect to be able to continue to do so.

Circling back to your first question, we refrain from disclosing country-by-country sales breakdowns to honor our confidentiality commitments to our customers. I am sorry I cannot be more helpful.

As for your second question, CPUs have long been used in PCs and servers. Earlier I used the term “logic ICs,” which include CPUs, GPUs, ASICs and FPGAs. Looking at current trends, we anticipate growth in demand for FPGAs or ASICs and, in the automotive space, GPUs for graphics processing. In short, we expect more growth in demand for GPUs, ASICs, FPGAs, etc., than for CPUs. FPGAs can be modified by customers to their own application specifications, but we feel that demand for non-CPU applications are in a growth trend.

(Q: How will ABF change in response to such trends?) We have a varied lineup of limited-edition ABF varieties with different specifications tailored to specific applications or customer requirements. I mentioned earlier that a wide range of different capabilities can be realized. The mix of properties differs for each variety depending on the application and customer needs. Combining these properties in the right proportions is our core technology and key know-how.

 

(Q: You said you are minimizing capital investment with an asset-light approach. How much production capacity do you ultimately plan on having?) We have spare production capacity even today. One crucial issue is managing delivery lead times given the numerous ABF varieties we produce. Because customers operate in rapidly changing industries, they want their orders manufactured and delivered right away. Such being the case, shortening order turnaround time and formulating quick-turnaround production plans are important in addition to production capacity. If we can manage order turnaround time well enough, we believe we can grow sales without making any additional investments for the time being.

 

I cannot answer your questions with specific numbers, but our forecast is largely predicated on data from customers on their expected ABF usage. Based on such data combined with various assumptions, we estimate the application mix and calculate future demand by application. In short, we simply tally up information from customers on how much of our products they expect to need.

We sell products to all PCB makers. The numbers we receive from them, when aggregated, tend to be somewhat on the high side. Because PCB makers sometimes provide us with numbers based on, for example, optimistic-case assumptions, we sometimes discount their numbers a bit. That said, we calculate projected demand using our own estimation models. We recently experienced a minor, short-term dip in orders from customers from late 2018 when US-China trade tensions started to become a topic of discussion. At the time, we thought customers were perhaps paring down inventories to some extent. Currently, however, we feel that orders are tracking nearly in line with end demand.

Current pricing is completely unremarkable from our perspective. Prices are always linked to volume. We price our products in line with their sales volumes. However, the recurring upgrade cycle of end-product applications is an important factor in the pricing equation. CPUs, for example, have evolved through a number of generations. Our lineup of ABF varieties likewise changes in response to such evolution. If a certain ABF variety becomes outdated, its price naturally declines. Conversely, new varieties are priced higher than older ones. CPU prices tend to exhibit the same pattern. In sum, you might say the two main determinants of price are volume and product lifecycle dynamics.

(Q: In other words, the product upgrade cycle is currently a tailwind for prices but you are pricing your products as usual, so nothing special is happening pricing-wise. Correct?) Yes. Even the product upgrade cycle is the same as always. We expect the current pricing environment to persist.

The specialty chemicals business has grown substantially since FY2017, so I think growth in FY2019 may appear relatively moderate. Even World Semiconductor Trade Statistics, an organization that issues international semiconductor market forecasts, is projecting a dulling of growth in FY2019. We have accordingly formulated a somewhat cautious forecast relative to growth in recent years in light of information from customers and objective data. Trend-wise, however, even objective market forecasts bode favorably for resumed growth from FY2020. Our outlook more or less coincides with such forecasts.

(Q: Then the electronic materials business’s costs have not changed much?) Correct.

(Q: You are forecasting the specialty chemicals business’s FY2019 marginal profit ratio at around 25% but business profit growth of only about ¥500 million. What accounts for this discrepancy in the absence of growth in fixed expenses?) We do not disclose information on electronic materials alone. The specialty chemicals business includes operations other than electronic materials. Disclosed information alone does not provide a detailed picture.

Although the environment will presumably change in various ways, we assume that pricing and profit margins will more or less remain unchanged from their status quo. We do not envision notable growth in fixed expenses.

Yes.

Both. The fact that applications change over time is very important. There has been a significant increase in the number of demands from end-product makers or, in some cases, PCB makers. How we meet each and every such demand has become a key issue for us. (Q10a: Back when CPUs were ABF’s main application, I assume processes basically changed about once every two years. Even with multiple ABF varieties, you were presumably producing a relatively limited product line by basically producing the varieties as a single series. If, however, ASIC demand in particular grows, am I correct to assume that ABF varieties would have to be replaced much more frequently because ASICs are completely custom designed and their packaging changes every time their design changes? Or are the ASICs to which you refer a new type of chip produced in relatively higher volumes like FPGAs or TPUs instead of conventional custom ASICs?) First, the number of ABF varieties is currently increasing quite considerably over time. As they do, order turnaround time becomes more important. In other words, although we have sufficient capacity, we have to produce a wide range of ABF varieties to order, so efficiently utilizing available capacity becomes more of a challenge. We would prefer customers to accept mass-produced varieties, so we strongly encourage customers to use those as much as we can. At the other end of the spectrum, we have some user-specific varieties also. We consequently have a vast product lineup comprising a mix of relatively high-volume varieties and small-niche varieties.

Our ABF prices are determined by the specific volume of the variety in question. We can charge high prices for little-used varieties. Conversely, if you buy a larger quantity, you get a better price.

(Q: Over the past year or so, I have frequently heard from Taiwanese contacts that prices are rising because of product shortages, but you say this is not the case. Are PCB makers saying that prices are rising because their demand mix is shifting as a result of increased purchases of custom varieties or small-lot varieties?) In our ABF business at least, prices on shipments to Taiwan have not risen notably. Any price increases are due to the correlation between a variety’s volume and its price.

Among logic ICs, we generally have very high market shares for CPUs or GPUs and for ASICs and FPGAs used in a wide range of applications, including PCs, servers and game hardware. In the smartphone space, by contrast, we have not captured much market share. We recognize that the reason for this is the crucial importance of the size of smartphones’ logic ICs. Because smartphones have such a small form factor, the request elements are slightly different. Their logic IC is called an application processor (AP), not a CPU. APs are not really our forte. Our competitors may be selling their materialsmostly for APs. Assuming this is the case, while demand for social infrastructure applications is in a growth trend, smartphone demand is highly sensitive to economic conditions. We are insulated against the impact of such a scenario. This may be the reason our growth rate appears to be higher than competitors’ if that is indeed the case.

The semiconductor makers that produce CPUs are different from those that produce APs. That is the decisive difference. We have strong, long-standing relationships with semiconductor makers that produce CPUs and ASICs. Our relationships with AP makers are not very strong yet.

(Q: Between 2008 and 2012, your ABF sales plateaued in response to PC market stagnation. Meanwhile, the smartphone market was booming. I imagine you might have been feeling a sense of crisis. How did you approach AP makers back then? And how have you remained strong in infrastructure applications even after ABF sales resumed growing from 2013?) We have still not made much progress in cultivating relationships with leading AP makers. ABF sales have been growing by virtue of the growth of semiconductor makers that are strong in CPUs. Our relationships with these CPU makers ended up driving growth in ABF sales for social infrastructure applications also.

We have excellent relationships with not only CPU makers but leading GPU makers also. Although we are weak in the AP market segment, we have built good relationships throughout the rest of the semiconductor industry. A huge driving force behind our close ties with the semiconductor industry is that we set up a base in Silicon Valley in 2015. With many of our customers clustered in Silicon Valley, we are able to meet with numerous customers daily. We have been building excellent relationships with a wide variety of customers.

We undertake maintenance investment every year, but we do not invest much. Think of it this way: While forex factors also perhaps boosted margins on exports from Japan, it was ABF sales growth that led to a higher business profit margin.

(Q: With customers’ capital investment expected to increase from FY2020, is it safe to assume your business profit margin may rise further?) Barring major exchange rate volatility, our business profit margin should steadily increase in such an environment. That said, our new ABF variety launch rate has a major impact on pricing. New variety launches vary in number from year to year. Such variation will likely affect our business profit margin to some extent.

Our specialty chemicals business encompasses electronic materials and personal care ingredients. The former belongs to the Life Support segment; the latter is part of the Healthcare segment’s “other” operations. The personal care ingredients business, like electronic materials, is also growing very rapidly. Both are asset-light businesses that share the same Company-wide management objectives. Namely, they aim to boost ROA and business profit margin and speed up the cash conversion cycle by pursuing specialization under an asset-light model. Both businesses share a common mindset and foundation.

Customers’ applications are very hard to ascertain. A given ABF variety is often used for different applications by different customers. I apologize if I sound like I am making excuses about the data we presented, but the boundaries between the ASIC and server or game categories are sometimes blurry. The customer may use a certain variety for a server application while we automatically classify it in the ASIC category based on its attributes. The same holds true between PC and server applications. As electronic hardware becomes more complex, multiple logic ICs will increasingly be mounted on a single motherboard, making it practically impossible for us to accurately track ABF sales by application.

 I will answer as best I can. The only thing we are sure of is that PCs’ share of demand will decrease. To reiterate, we collect information on end-product applications from semiconductor makers in the course of our day-to-day communications with them, but actual ABF orders are placed by PCB makers. We have to draw inferences on applications based on the ABF varieties on incoming orders. Based solely on ABF variety-mix trends, we anticipate growth in ABF use for server and 5G applications. Demand for game hardware applications seems to be slowing.

As one example, interconnect circuitry is progressively shrinking. Process technology has currently progressed as far as the 7nm mark, but technical challenges with processing progressively intensify as width decreases. ABF is highly process-friendly, including for copper deposition processes. As process geometries continue to shrink, ABF could be used for applications for which it has yet to be used so far, perhaps even smartphones—our weakness. With smartphone batteries getting bigger and bigger, other smartphone components have to keep shrinking. If this trend continues, it may present opportunities for us. There are some other technical requirements, but we believe we stand to benefit from such a miniaturization trend, for example.

Another driver of change is energy efficiency. I cannot go into detail, but we are developing new formulations able to meet more demanding energy efficiency specifications.

On page 13 of our presentation materials, you will find the term “low dielectric tangent.” This is another technical requirement. When electricity flows through a circuit, electrical load increases in tandem with signal throughput. If electrical load exceeds a certain threshold, it can prematurely burn out circuitry and/or impair energy efficiency. A low-dielectric-tangent material is required to minimize electrical load. We are developing formulations capable of meeting the requisite low dielectric tangent specifications.

I believe the companies involved in manufacturing APs are very knowledgeable about ABF. However, I suspect they may be resistant to using ABF because they are heavily cost-focused given the fierce price competition in the smartphone market.

We rolled out quite a few new varieties in FY2018. I doubt the number will be larger in FY2019. We do not have many major new variety releases planned in FY2019.

“Functional materials” is a very broad label that can be applied to lots of different things. From such a perspective, it is likely that other hit products will emerge in the future. With assemblies of various micro-components being manufactured for use in smartphones and other electronic hardware, our adhesive shipments, for example, could increase substantially if our adhesives are used in popular products. Dispersing agents also tend to be used for diverse applications. While sales of products such as adhesives and dispersing agents might increase in the future, we do not anticipate much growth in sales of epoxy curing agents or flame retardants, both of which are legacy products.