Sumco Corporation (OTCPK:SUMCF) Q3 2022 Results Conference Call November 8, 2022 1:00 AM ET
Company Participants
Hiroshi Shibaya – Managing Executive Officer
Mayuki Hashimoto – Chairman & Chief Executive Officer
Michiharu Takii – Vice President & Chief Financial Officer
Hiroshi Itoh – General Manager of Accounting
Conference Call Participants
Takashi Enomoto – BofA Securities
Atsushi Ikeda – Goldman Sachs
Yu Yoshida – CLSA Securities Japan
Go Miyamoto – SMBC Nikko
Shigeki Okazaki – Nomura Securities
Mikiya Yamada – Mizuho Securities
Hiroshi Shibaya
I am Hiroshi Shibaya of Sumco Corporation. Thank you for your participation today. This is the results briefing for Q3 of the fiscal year ending December 2022.
Before starting the presentation, allow me to confirm today’s materials, which consists of 3 items: the brief statement on consolidated financial results for Q3 fiscal 2022, the announcement regarding revision to dividend forecast and the presentation deck entitled Results for the Third Quarter of fiscal 2022. This will be a roughly 1-hour briefing, which will end at 5:00 p.m.
Next, a disclaimer. The estimates, expectations, forecasts and other future information discussed here and shown in today’s materials were prepared based on the information available to the company as of today, and on certain assumptions and qualifications, including our subjective judgment. Actual financial performance or results may differ substantially from the future information contained in this material due to risk factors, including domestic and global economic conditions, trends in the semiconductor market and foreign exchange rates.
We will have presentations today from Representative Director, Chairman and CEO, Mayuki Hashimoto; and Vice Chairman, CFO, Michiharu Takii. Hiroshi Itoh, General Manager of Accounting is also on hand. Chairman and CEO Hashimoto will discuss our forecast and operating environment, to be followed by an explanation of the financial results by CFO Takii. We have set aside time for a Q&A session as well.
I will now hand over to Chairman Hashimoto.
Mayuki Hashimoto
I am Chairman Hashimoto. I will start with Slide 5 of the presentation. These are the Q3 results, as you will have already seen. Q3 sales were ¥116.2 billion; operating income, ¥30.2 billion; ordinary profit, ¥32.5 billion; and net profit attributable to the owners of the parent was ¥20.4 billion. The average dollar-yen rate for Q3 was ¥136.6.
The yen was ¥1.6 weaker than our assumptions, but it wasn’t a huge difference.
Turning to our Q4 forecast. We are guiding for sales of ¥115.5 billion and operating income of ¥28.5 billion, which is down slightly Q-on-Q. Our guidance for ordinary income is ¥27 billion and ¥16 billion for net profit attributable to owners of the parent.
Our ForEx assumption for Q4 is ¥145 to the dollar. Optically, profits appeared down slightly Q-on-Q.
However, the forecast reflects the impact of periodic maintenance at the Kubara plant at Imari in November, which will have a relatively significant impact on production volume. Next slide, please.
This is shareholder returns. We had already disclosed this, but we are guiding for a full year dividend of ¥78, which represents a full year payout ratio of 40.3%. The year-end dividend per share guidance has been set at ¥42. Next slide, please.
This is the market environment for silicon wafers. In Q3, demand for 300-millimeter wafers was still very strong and continued to exceed supply. 200-millimeter wafer production remained at full capacity utilization. However, for 150-millimeter and smaller diameters, we saw an easing of previous tightness. On prices, LTA prices were maintained, of course.
Spot prices rose.
Next are our expectations for Q4. In the current Q4 already underway, in 300-millimeter, we are already seeing a substantial correction in memory. Conditions vary widely by customers for both memory and logic. We are seeing a polarization between customers, with a clear separation between winners and losers. However, despite a mixed picture, on an overall basis, we expect total demand to outstrip supply.
For 200-millimeter in Q4, demand from automotive and industrial applications remain strong. As with 300-millimeter wafers, we are maintaining full capacity utilization for 200-millimeter wafers. For 150-millimeter and smaller diameters, the correction in end demand in consumer applications is ongoing, so we are no longer producing at full capacity.
On Q4 prices for both 300- and 200-millimeter, LTA prices are being maintained. There is not much change in spot prices either. What is key is the outlook going forward. Although customers have already started production adjustments on the back of weakness in end markets like smartphones and PCs, demand from data centers and automotive applications has remained strong. As a result, although there are differences between customers, as noted earlier, overall, we expect to continue producing at full capacity utilization for 300-millimeter given the demand for volumes we are seeing.
For 200-millimeter, automotive and industrial applications continue to drive the market, so current high capacity utilization rates are likely to be maintained. However, for 150-millimeter wafers, we expect a slightly softer trend. Next slide, please.
Slide 8 shows the wafer trend for 200-millimeter by quarter. As you can see, production volume remains capped at the ceiling. I believe that the 5 majors have pretty much done all they can in terms of improving productivity. These levels probably represent the limit. Volumes are largely unchanged from last year. I think the industry is running at full capacity utilization.
The applications for 300-millimeter are different than 200-millimeter, so chipmakers can simply shift to 300-millimeter if there are shortages in 200-millimeter. Unlike 300-millimeter wafers, where the range of applications is small but the scale of volume per application is large, 200-millimeter demand tends to consist of a wide array of applications in small volumes.
Given this, I think a significant migration to 300-millimeter is unlikely as the economics for small lot production would be challenging. As a result, demand for 200-millimeter waivers has been persistently strong to this point. Next slide, please.
This is the same chart for 300-millimeter wafers. On this chart as well, you can see that volumes have remained largely flat at the ceiling from Q1 through Q3. Effectively, this is the proof that the industry has reached the limits of brownfield expansion. To this point, there had been some room to expand brownfield production, which is why the trend had been sequential improvement.
It is rare that we see a flat quarterly progression. However, the flat trend for Q1 through Q3 is not the result of a weak market, but is a sign that the industry has reached its limits and cannot increase production any further. The next increase will only come when the greenfield capacity currently under development comes online. Next slide, please.
This is our outlook for smartphone shipments. There has been a lot of talk about declines in handset volumes. However, although total handset volumes are indeed down, we expect 5G to account for 80% of the total by 2026. Next slide, please. This shows the share breakdown for application processor makers for smartphones.
We can’t disclose the specific identities of these players, but what I can say is that half is leading edge and the other half is more commoditized product that is typically being used in China smartphones. Next slide, please. This chart looks at the trend in the number of cores in application processors. The number of cores has been rising steadily, but at the same time, over the last 3 years, wafer consumption surface area for processors is also increasing. The increase reflects the continuing need for high-performance application processors in high-end 5G handsets.
Please turn to Slide 13. This slide compares surface area trends for leading-edge smartphone application processors. The companies we show here, i.e., company A, company B, are companies that fabricate advanced application processors. As you can see, the volume of silicon being consumed for application processors, effectively surface area, is rising. Next slide, please.
This looks at 5G in comparison to 4G. First of all, the transition from 4G to 5G leads to a 1.5x increase in wafer consumption. On top of this, even within 4G, we are seeing a trend of growth in surface area on the back of increased functionality. The reason why we included this explanation is because although smartphone handset volumes are falling, this is outweighed by the positive impact on wafer consumption as a result of improved functionality. The key takeaway is that a decline in handset volumes will not translate into a decline in epi wafers.
Please turn to Slide 15. This bears out our thesis, we continue to see new functionalities be incorporated in smartphones. Memory is more of a nice to have, but a handset without an application processor isn’t a smartphone, so application processors are essential components and therefore less prone to sudden declines.
Memory is a nice to have in that, with cloud storage, you don’t need to have a handset with a huge amount of memory to store content like pictures. As a result, when economic conditions start to look a little shaky, demand shifts to lower memory densities, depressing memory. Logic, on the other hand, plays a central role so it doesn’t go away.
Let’s move on to the next slide to look at what is happening with the wafer inventory. It is true that inventory has increased a little recently. Next slide, please. If we break out customer 300-millimeter wafer inventory between logic and memory, the bar chart seem to suggest that logic inventory has risen while memory has not.
But in actual fact, on the logic side, there is one player that is struggling and has seen a very significant increase in inventory, which led to the overall increase in inventory for logic. Logic players that are doing well have not seen inventory increase very much. Obviously, we cannot provide a breakout by individual company, but that is what is happening. Next slide, please.
So how should we analyze what is happening with inventory? This chart plots trends for inventory, price and demand for 300-millimeter epi wafers. The volatile orange line, which shows some dramatic swings is inventory volume on a monthly basis. The blue line shows trends in demand. The gray line is price.
The data is our estimate of market prices rather than Sumco’s own prices.
What this chart tells us is that for logic, for example, there was a dip in 2015, but there was a rebound after 4 quarters. If we look at the near-term conditions for inventory, which is the orange line, we are at record lows. So hypothetically speaking, even if there is a drop, it is unlikely to get too severe before there is a recovery and any dip is likely to be quite modest.
Right now, there is a player that is struggling. This is due to factors specific to this company. This player has a strong focus on PC applications, so as being directly impacted by the weakness in PCs. Conditions at other players are not that bad. So even if there is a correction, we would expect the correction to be of relatively short duration and modest in terms of the magnitude of the downturn.
With regard to wafer purchases, given the low level of inventories, we haven’t seen any major changes so far. Also, if you look at the gray line, which is the price trend, shortages started to emerge from the second half of 2016, which led to LTAs. The chart clearly shows the impact of LTAs. Since LTAs were initiated, you can see price volatility has gone away. Next slide, please.
This is the same chart, but for polished wafers. I forgot to mention one point on the previous slide, but with epi wafers, typically when wafer consumption goes up, inventory also tends to go up. In other words, the 2 tend to move in tandem. This is because foundries account for a high proportion of epi demand. So when demand picks up, the foundries tend to be proactive in locking in wafer supply.
However, with polished wafers, customers do not increase inventory until demand increases. So at the time of an increase in demand, inventory typically tends to decline. So if you look only at inventory, the behavior patterns are very simple. Epi customers tend to anticipate demand, buying wafers in excess of immediate demand which lead to increases in inventory levels in the short term before inventory starts to decline. So there are differences in behavior between epi and polished wafers.
As you can see on this chart, inventory has been in a down trend. Here as well, there was a correction in 2019, but roughly speaking, the recovery began after 4 quarters. In terms of past examples, although the chart does not show data prior to 2013, generally speaking, corrections bottomed and rebounded after 4 quarters. Given this, it appears that memory is in a correction phase, but probably the correction phase last 4 quarters before we see a bounce back in my view. In other words, sometime in mid- to late 2023, and I believe we are likely to see a recovery.
That said, for polished wafers, inventory levels are relatively elevated. So I imagine we could see a slight correction in wafers as well. However, given this is the nature of our business, we recognize that there is a cycle of overshoots in supply followed by supply shortfalls. We operate in a growth industry, so a misreading of growth can lead to excess supply. A misread in the other direction can lead to a supply shortfall.
Because it is not possible to perfectly predict the pace of growth, growth industries are unfortunately prone to lumpy trends in my view. Next slide, please.
This is the outlook for demand. The red dotted line is our forecast for demand based on what we were hearing from our customers at the beginning of the year when everyone was bullish. If we look at current conditions, growth now looks likely to be more modest, dropping to close to GDP-like levels. We expect 2023 to be flat with a return to a growth trajectory from 2024. This coincides with the ramp-up of greenfield capacity.
That said, probably we would still have a situation where there are slight shortages. Next slide, please.
This is a completely different topic. We recently announced the acquisition of Mitsubishi Materials’ semiconductor polysilicon business. Although this business was deemed noncore by Mitsubishi Materials, obviously, our objective was to acquire a business which focuses on the production of essential raw materials for Sumco. For us, this represents a core business.
We are acquiring not only polysilicon production capability, but also trichlorosilane as well, which is essential to the fabrication of epi wafers. There are only 2 companies in Japan that can supply these materials. We chose to take over these operations given that they are essential to our business.
Vertically integrating will give us an advantage when there are supply-demand imbalances in polysilicon. 10-plus years ago, when polysilicon prices surged, we were forced to lock in supply at high prices. With in-house capacity, we won’t need to scramble for supply.
Obviously, this does not cover all of our needs, but we will have a situation similar to crucibles where we have in-house capability and also procure externally.
In terms of the transaction structure, we are acquiring equity in a new company engaged in the production of semiconductor polysilicon to be created by Mitsubishi Materials, as well as buying Mitsubishi Materials shares in Mitsubishi Polycrystalline Silicon America Corporation and NIPPON AEROSIL. We plan to complete the acquisition of the above on March 31, 2023. The contract has already been signed.
This completes my section of the presentation. I will hand over to CFO Takii to talk about details of our earnings.
Michiharu Takii
I, Takii, will present the earnings in more detail. Please turn to Slide 23. These are the Q3 results. As we show here, we were able to exceed ¥30 billion in operating income in Q3. Under nonoperating income and expenses, we recorded a significant ¥2.3 billion in income.
This was underpinned by a positive ¥2.4 billion translation gain from FST as a result of the significant weakness of the New Taiwan dollar to the U.S. dollar. As a result, ordinary income also surpassed ¥30 billion. Net profit was a hefty ¥20 billion. Further down the table, we are making solid progress on CapEx, which rose to ¥34.2 billion.
Looking at the various KPI at the bottom of the table, OPM was virtually over 25%, and the EBITDA margin was 38% or 39%. On a 9-month basis, ROE was close to 14% and 15.9% for Q3 on a stand-alone basis. Please turn to the next slide.
This is the analysis of changes in operating income. First, the sequential analysis. As shown in the waterfall chart on the lower left, costs rose ¥2.5 billion, mainly the result of an approximately ¥1 billion increase in materials costs, a ¥1.7 billion increase in electric power costs and a further ¥1 billion rise in labor costs. The increase in electric power costs is due to higher unit prices. Higher prices also pushed up materials costs.
With the ramp-up of new facilities, depreciation was also up Q-on-Q.
Sales-related variance improved ¥3.5 billion Q-on-Q. This breaks down into a roughly ¥1 billion contribution from price. There were increases reflecting the timing of price resets under LTAs and the rise in spot prices in Taiwan, mainly from 300-millimeter wafers.
The impact of an increase in volume was ¥2.5 billion. Foreign exchange had a positive impact of ¥4.2 billion, of which dollar-yen accounted for ¥3 billion and New Taiwan dollar to U.S. dollar was around ¥1 billion.
On the right, we show the year-on-year change for 9-month results. 9-month OP improved ¥43.4 billion year-on-year. There was a significant year-on-year increase in costs of around ¥13 billion. The breakout is similar to the factors in the Q3 sequential analysis reflecting higher unit prices for materials, which had a ¥2 billion impact, and electric power unit prices for an impact of ¥3 billion year-on-year.
Labor costs also rose by ¥3 billion. The higher labor costs reflect hiring with headcount increasing, including some hiring in anticipation of the greenfield capacity for a contribution of ¥3 billion. There was also an increase in bonus payments as well as higher unit labor costs, which contributed ¥2 billion to the higher labor costs. Sales-related variance had a roughly ¥43 billion year-on-year positive impact.
The impact from volume was around ¥17 billion. Price increases had an impact of ¥26 billion, which breaks down into 70% from 300-millimeter and the remaining 30% from 200-millimeter and smaller diameters. Foreign exchange had a positive impact of ¥20 billion, ¥18 billion due to dollar-yen and ¥1 billion due to New Taiwan dollar versus the U.S. dollar. Next slide, please, for a discussion of the balance sheet and cash flows.
Slide 25 shows the balance sheet. Cash and time deposits on the balance sheet increased ¥52.9 billion. Accounts receivable rose ¥14.3 billion on the back of sales. Finished product and work in progress also increased. Raw materials and supplies fell ¥0.7 billion.
The reduction in polysilicon inventory within this was ¥7 billion, which was offset by an increase of around ¥7 billion in materials inventories, particularly stock of maintenance-related materials which we considered important from a BCP standpoint, given we are running at full capacity.
Tangible assets rose on the back of higher CapEx. Other liabilities increased. We discussed this previously, but our joint venture subsidiary in Taiwan for. FORMOSA SUMCO TECHNOLOGY, took in advanced payments and deposits on LTAs of roughly ¥29 billion. This contributed to the increase in other liabilities.
If we look at the capital account, retained earnings increased, reflecting the impact of net profits after the payment of dividends. There was an increase in noncontrolling interest and others of ¥25.5 billion, of which ForEx translation accounted for more than ¥13 billion. You can see the various metrics below.
On the cash flow statement, operating cash flow was ¥143.4 billion, while investment cash flow was ¥79 billion. Free cash flow was ¥65.3 billion. After dividends paid and other adjustments, we get to the net increase in cash and time deposits highlighted earlier. Next slide, please.
These are our forecasts for Q4 on Slide 27. As alluded to by Chairman Hashimoto at the outset, we project slight declines in sales and OP in Q4. Under nonoperating income, we have factored in typical ForEx translation impacts. With the management metrics shown at the bottom of the table, on a full year basis, we continue to see improvements.
In particular, the EBITDA margin was able to achieve a 39% level in Q3 and is expected to do so in Q4 as well. CapEx is not shown on the slide, but for now, we expect to finish the year at around ¥100 billion. EBITDA has also improved significantly. We project the average ForEx rate for the full year to be ¥130 to ¥131 to the dollar. On the next slide, we provide the analysis of sequential change for Q4 and year-on-year change for the full year operating income.
I will start with the sequential change from Q3 into Q4. We expect to see a continuation of rising costs for materials and electric power. The other point I will highlight is the negative ¥2.3 billion in sales-related variance. This is the result of reduced production on the back of periodic maintenance at the Kubara plant, which will be shut down for 1 week in November. We expect a positive exchange rate impact of ¥3.5 billion.
Looking at the analysis of year-on-year change for the full year OP, we project a year-on-year improvement of ¥56.9 billion. Costs are expected to rise ¥19 billion year-on-year. The key factors are an extrapolation of what we saw in Q3. Materials and supplies costs are expected to rise ¥3 billion year-on-year. We expect electric power unit prices to rise for an impact of ¥5 billion.
Labor costs, including headcount increases, will rise ¥7 billion year-on-year. Repair and maintenance costs and logistics expenses combined are expected to increase ¥5 billion year-on-year. For sales-related variance, we expect a ¥51 billion positive. Here again, the key factors are an extrapolation of the 9 months trend.
Volume should contribute a positive ¥18 billion, while price contributes ¥33 billion, with the split of price impact between 300-millimeter wafers and 200-millimeter and smaller diameters, around 70% and slightly less than 30%, respectively. We expect a ¥33.5 billion positive from ForEx for the full year, reflecting a ¥28 billion contribution from dollar-yen and ¥5 billion from New Taiwan dollar to the U.S. dollar. Next slide, please.
As you can see on Slide 30, from 2021 onward, we have continued to report growth in revenues and profits. From this year onward, ForEx has also been a tailwind. Volumes are up as our prices, so sales have grown significantly. This completes my remarks.
Question-and-Answer Session
A – Hiroshi Shibaya
Thank you. We will now open the floor to questions. [Operator Instructions] We will start with Mr. Enomoto of BofA Securities.
Takashi Enomoto
My question is about CapEx and capacity expansion. This time, you revised your demand forecast for silicon wafers. Is it possible that you might delay your expansion plans or slow down the pace of capacity expansion? Or given the broad tightness in materials and labor, is the progress on capacity expansion already lagging? Is it the case that you want to move faster, but have fallen slightly behind?
Also, if possible, given that the impact of ForEx is likely to have a significant impact on CapEx, could you provide additional color?
Mayuki Hashimoto
First of all, with regard to the possibility of delaying CapEx, actually with suppliers already indicating they expect delays, we have been focused on trying to find ways to ensure we can stay on track with our schedule. It has been a scramble, but we have been able to get back on schedule. I have no intention of delaying CapEx. The customers with whom we have contracts have not made any sort of indication that they want to delay.
I believe we have a responsibility to our customers to ramp up our new capacity as quickly as possible and to adhere to the schedule. So no intention whatsoever to delay CapEx or reduce volumes. If you are asking whether ForEx has had a negative impact on CapEx, that is not the case. The vast majority of what we require has already been purchased. So we would expect to see very little further impact from ForEx.
There will likely be an impact on investments to expand capacity beyond 2026. Obviously, however, prices will also be pushed up by ForEx, so I believe they should cancel each other out.
Takashi Enomoto
Understood.
Hiroshi Shibaya
Thank you, Mr. Enomoto. Next is Mr. Ikeda of Goldman Sachs.
Atsushi Ikeda
I would like to ask about your view of semiconductor demand in 2023 and how this will relate to wafer procurement. You suggested there will be an impact on wafer procurements from memory.
However, for epi wafers, should we expect there will be little impact given the solid conditions at your main customers? Hypothetically, if there were to be changes in demand volume, your Japanese competitor indicated that they were prepared to adjust prices provided the overall value of the contracts remained unchanged. Please comment on your price strategy.
Mayuki Hashimoto
First of all, with regard to price, we have already made investments, so I expect our customers to respect the terms of the agreements we have. I believe our customers understand this. In the past, where we did not incur CapEx, and it was simply a case of prices going up, there might have been room to be accommodating. But even so, in 2017 and ’18, as we showed in the charts, prices did not change. So no intention whatsoever to change price.
However, with regard to delivery volumes, in a situation where the customers’ warehouses are already jampacked, we can’t just simply force our customers to take delivery. In that situation, if it comes to that, then we might entertain a discussion of possibly allowing the customer to push out the shipment schedule. That said, none of this has happened yet.
Atsushi Ikeda
Understood. Your competitors suggested they were prepared to be accommodative in exchange for higher prices, but am I correct to assume that you are not currently thinking about such measures? So for 2023 contracts, you aim to have customers respect the 10% price increase as stipulated in the LTAs, is that correct?
Mayuki Hashimoto
Yes, we will proceed on that basis.
Atsushi Ikeda
Understood.
Hiroshi Shibaya
Mr. Ikeda, thank you. Next is Mr. Yoshida of CLSA Securities Japan.
Yu Yoshida
My question is about Slide 20 and the demand forecast. When I look at the solid red line, it looks like you expect demand to be weak in the early part of 2023, but then expect a recovery in the second half of the year. Is that the correct way to read this? Why is the line not flat for all of 2023?
Mayuki Hashimoto
You are correct. Ours is an industry where demand is lumpy. It will be great if capacity expansion materializes in the way we hope. But by nature, we often see mismatches between the volume required by the market and the increase in our production volume. When I look at 2023, for now, it feels like memory is weak.
Given this, I would expect to see a slight correction. That is reflected in the chart.
The biggest question is how long the correction will persist. If we take into account past trends and inventory levels, as well as penetration rates for LTAs as it impacts prices, we estimate a correction would typically last 4 quarters at most. So this is our assumption for the portion of the red line that is flat.
Yu Yoshida
I see. If possible, could you share your forecast for estimated volume for 2023 and the split between memory and logic?
Mayuki Hashimoto
It’s tough for me to comment on volume. What I can say is that, as discussed, we believe the probability of a correction in logic is low, but the probability of a correction in memory is high. In fact, customers are already adjusting production.
Yu Yoshida
Yes. Understood.
Hiroshi Shibaya
Thank you, Mr. Yoshida. Next is Mr. Miyamoto of SMBC Nikko.
Go Miyamoto
In the last correction in 2019, players like Shin Etsu and SK Siltronic were able to raise their share on a value basis with strong revenue growth, but Siltronic lost share in value terms. Do you think it is possible that Sumco will be able to increase its share this time? Compared to Shin Etsu, do you think you can achieve a higher rate of sales growth? Hypothetically, if the market were to head into a downtrend, please talk about your confidence in your earnings.
Mayuki Hashimoto
In the last correction, Shin Etsu and SK Siltronic were able to grow share because they are strong in polished wafers. Sumco did not have much polished wafer capacity to begin with, so we lagged. We are not really that concerned about what the others may be doing. We are strong in epi. We have an interest in polished wafers, but if you look at the top 5 players, there isn’t really room for differentiation in polished wafers.
Given that it is not possible to show technological differentiation in polished wafers, we are not really aiming to expand aggressively.
All of the current round of capacity expansion for Sumco is in epi. So it may well be that our share of epi will increase, but the flip side is that our share of polished might decline. After factoring in all of the puts and takes, at a minimum, we would aim to maintain our share. We are not aiming to take share away from others. There may be some switching if you look at polished wafers and epi.
Go Miyamoto
I see. So you expect to be able to maintain your share even in a down cycle. And the reason is because epi will grow, is that correct?
Mayuki Hashimoto
Yes.
Go Miyamoto
This is because you have strength in epi?
Mayuki Hashimoto
Yes.
Go Miyamoto
Understood.
Hiroshi Shibaya
Thank you. Next is Mr. Watanabe of Mitsubishi UFJ Morgan Stanley Securities.
Unidentified Analyst
I have a question about Slide 12, where you talk about application processors and surface area. In talking about 5-nanometer, you mentioned that die sizes got bigger. On the back of a change within the same generation where there is an increase in the number of transistors, there is a choice between chip shrink and increasing surface area. I get that in the latter part of a generation, surface area increases, but when the transition is from 5-nanometer to finer design rules of 4 or 3 nanometer, why would die sizes increase?
Mayuki Hashimoto
This is really more about the customer’s design philosophy, so I don’t really know. What we do know is that functionalities continue to be added. In the transition from 10-nanometer to 5-nanometer, for example, die sizes did not get smaller. I suspect that the customers are being asked to provide more functionality. So in the transition from 5-nanometer to 3-nanometer, I suspect chipmakers are being asked to add even more functionality.
Even though we, as individuals, are not fully able to tap into all of the functionalities embedded in application processors, I think new functionalities keep being added.
Unidentified Analyst
At this stage, do you have a sense that die sizes could get even larger going forward?
Mayuki Hashimoto
No, I think this is it.
Unidentified Analyst
Understood.
Mayuki Hashimoto
In the next step, the technology will move to gate all around, which will be a completely different world.
Unidentified Analyst
Understood.
Hiroshi Shibaya
Thank you, Mr. Watanabe. Next is Mr. Okazaki of Nomura Securities.
Shigeki Okazaki
Based on the discussions today, I would like to ask about how to think about price trends in the medium term. I would like to confirm the 300-millimeter wafer LTA prices will rise until 2024 and then will be flat to 2027. Is that correct??
Mayuki Hashimoto
Actually, prices for 2027 have not been set yet. Prices are fixed up to 2026 and will gradually rise over time.
Shigeki Okazaki
My understanding was that LTA prices would rise to 2024 and remain flat?
Mayuki Hashimoto
The LTAs are not all the same. Some LTAs are flat after 2024 after a relatively large increase at the outset, some will rise gradually over time. There are some where the prices rise in a step-wise function, while others are linear. Ultimately, however, all of the LTAs will end up at a similar level in 2026.
Shigeki Okazaki
In terms of price increases, is it true that the increases will be relatively modest up to 2024?
Mayuki Hashimoto
I couldn’t hear you very well, but generally speaking, the magnitude of the price increases is larger in the period up to 2024, with a more flattish trend in the subsequent years.
Shigeki Okazaki
Is it fair to say that regardless of demand trends, LTA prices will rise?
Mayuki Hashimoto
With regard to demand, it is quite rare to see such a clear bifurcation between the winners and losers. In terms of corrections, there are customers that are making very significant adjustments and others that aren’t adjusting at all and only finally catching their breath.
Conditions at the individual chip makers are actually quite different. We have seen the emergence of clear winners and losers. One more thing I will say is that customers in auto-related and power semis are doing very well with no weakness at all in demand.
So there are significant differences between individual customers. This makes it very tough to generalize. But if you bring it all together, my sense is that the overall situation is not that bad.
Shigeki Okazaki
Previously, you had indicated that the increase in LTA price to 2024 is around 10%. Is that correct?
Mayuki Hashimoto
For our LTAs, yes.
Shigeki Okazaki
Understood.
Hiroshi Shibaya
Thank you. Next is Mr. Yamada of Mizuho Securities.
Mikiya Yamada
Everyone has asked about 300-millimeter, so I think I have a good understanding of the situation. I would like to ask about 200-millimeter and smaller diameters. IGBT demand will clearly increase going forward, in my view. Also with regard to automotive, a transition to 300-millimeter seems unlikely given the need for scale to make the economics work, as you alluded to earlier. From that perspective, I think the tightness in 200-millimeter may become even more acute in some instances.
Is there a possibility that some applications might choose to move back to 150-millimeter wafers, which would be supportive for smaller diameters overall? Also, if the shortages in 200-millimeter become truly serious, do you think that some chipmakers might choose to migrate to 300-millimeter epi despite the suboptimal efficiency? Please comment on 200-millimeter and smaller epi.
Mayuki Hashimoto
That is a very good question and something that I have been tracking. For instance, with IGBT, the strong leading player is proactively migrating to 300-millimeter. So no possibility of a reversion to 150 millimeters. For power management and analog ICs, 150-millimeter capacity is being shut down and a transition to 300-millimeter is being attempted. So the analog and power IC related, which are both strong, are being actively shifted to 300-millimeter of this customer.
They have indicated that they will want more 300-millimeter wafers in future.
Mikiya Yamada
I see. So in other words, for IGBT, given the greater thickness of the required epi layers, said to be 20 to 25x higher than logic, it would seem even more meaningful to be expanding epi capacity early. Is this the right way to think about this?
Mayuki Hashimoto
Yes.
Mikiya Yamada
Thank you.
Mayuki Hashimoto
Sumco is the top supplier, and we have a very strong relationship with this customer. We have won the best supplier award every year. So I do think this customer has high expectations for Sumco.
Mikiya Yamada
I have high expectations, too.
Hiroshi Shibaya
Thank you, Mr. Yamada. Thank you to everyone. We are grateful for your participation today. We will end the meeting here.
Be the first to comment