Sumco Corp

TSE:3436

Earnings Call Transcript 2024-Q1

[Takayuki Komori, executive]

Thank you for your participation today. This is the results briefing for the first quarter of the fiscal year ending December 2024. Before starting the presentation, allow me to confirm today's materials, which consists of 4 items: the brief statement on consolidated financial results for Q1 fiscal 2024, the announcement concerning disparity between forecast and actual figures for the first quarter of fiscal 2024, the announcement regarding revision to dividend forecast, and the presentation deck entitled Results for the 3 months ended March 2024. This will be a 60-minute 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 President, CFO, Shinichi Kubozoe. Chairman and CEO, Hashimoto, will discuss our forecast and operating environment to be followed by an explanation of the financial results by CFO, Kubozoe. We have set aside time for a Q&A session as well.

I will now hand over to Chairman, Hashimoto.

[Mayuki Hashimoto, executive]

I am Chairman Hashimoto. I will start with the overview on Slide 5. This is a summary of the results for the first quarter of the fiscal year 2024. Both sales and profits for first quarter significantly overshock our forecast. Sales exceeded the forecast by JPY 6.5 billion. As shown here, operating income and ordinary income were roughly double the forecast. If we look at operating income, which we planned by JPY 4.1 billion, half of the overshoot is the result of higher-than-expected volumes. In terms of what was behind the variance on volumes, Overall, our forecast was too pessimistic as well, earnings were supported by the market bottoming out with 300-millimeter epitaxial wafer shipments relatively stronger than we had anticipated. Costs and foreign exchange also contributed by JPY 0.9 billion and JPY 0.8 billion, respectively. All of these factors combined to generate an overshoot of around JPY 4 billion.

Turning to the earnings forecast for the second quarter of 2024, we project a sequential increase in sales of JPY 5.5 billion to JPY [ 99 ] billion. Operating income is projected to be JPY 9 billion, ordinary income JPY 8 billion and net income JPY 5 billion. We assume a dollar-yen rate of JPY 150 to the dollar. In terms of sales, we expect 300-millimeter volumes to improve, but expect volumes for 200-millimeter and 150-millimeter to be flat Q-on-Q. We believe 300-millimeter wafers have bottomed out. That said, although I will go into more detail later, customer wafer inventory levels remain very high. I expect that a recovery in wafers for us may take some time.

Please turn to Page 6 for shareholder returns. This slide shows shareholder returns. We are guiding for an interim dividend of JPY 10. The fiscal year-end dividend is to be decided. The guidance represents a consolidated dividend payout ratio of 35.4%. The Note that we have excluded subsidies received after tax from the calculation of the dividend. In value terms, subsidy received are not that large, but we are excluding it from the base for calculating dividend.

Next page, please. Looking at current market conditions for silicon wafers and the outlook, in Q1, 300-millimeter wafers appear to have bottomed out for the market as well. The driving force behind this is, as you already know, logic and DRAM for AI, which showed strong growth. Customers continue to adjust production for other applications. For diameters of 200-millimeter and smaller overall market demand was weak and the correction continues. We expect shipments will continue to decline. So 200-millimeter is bottom crawling and 300-millimeter appears to have bottomed out.

On prices, customers continue to respect LTA prices. In the spot market, although we have no 300-millimeter spot, spot prices for 200-millimeter fell to a certain degree. We think this is inevitable. Prices are not plunging dramatically, but given the weak demand, they appear to have softened somewhat. For Q2, while 300-millimeter demand has bottomed out, customers continue to carry high levels of inventory and are producing from wafer inventory. As such, the improvement in demand hasn't reached wafers yet. For 200-millimeter, customers continue to adjust production, so low shipment levels are likely to continue. On prices, as noted earlier, LTA prices are being respected. But for spot, although we have no 300-millimeter spot, the picture is mixed, depending on region and application.

On the outlook going forward, demand related to AI is likely to remain strong but customer capacity is limited, which is constraining production and limiting wafer consumption. The customers are actively moving to increase capacity, but it is likely to take time. PC and smartphone appear to be picking up slightly, but this is being offset by a weakening of demand for automotive and industrial. Given this, the overall picture is for a gradual recovery over time in our view. My sense is that we are unlikely to see a sharp V-shaped recovery as had been seen in the past. So that is the situation in terms of customer consumption. Given that customers are also still adjusting inventory, a recovery in silicon wafer demand is likely to be gradual in my view.

Page 8, please. This slide shows the wafer trend for 200-millimeter wafer by quarter. The black line is 2023 levels and the red dot is Q1 2024. As you can see, this is the lowest level we have seen in the last 7 or 8 years. Applications for 200-millimeter are completely different than for 300-millimeter. What we are seeing reflects the weakness of consumer applications at this time. So until we see a recovery in consumer applications, although there are some applications where demand for devices is strong, Conventional applications appear to be very weak. Addressing this will require a recovery in the macro backdrop. I believe U.S.-China trade tensions are clearly having a significant impact. We believe these conditions will persist for some time to come.

Page 9, please, for 300-millimeter wafers. This is a trend for 300-millimeter. Unlike 200-millimeter, 300-millimeter is not at 7- or 8-year lows but feels like it is only down slightly. It is currently close to 2020 levels. I believe we will see a gradual improvement over time. In terms of what we are expecting to see, please jump forward to Page 11. So this is the much founded AI chip for use in servers. AI chips do a high number of computations storing huge volumes of data in memory instantaneously for rapid retrieval for data optimization. The more complex the calculations, the higher the number of learning parameters. Each year, the number of parameters is said to be increasing by a factor of 10.

AI chips require DRAMs and GPUs. The need for AI chips will continue to increase. Currently, supply is falling well short of demand. There has been talk that production out to 2025 has already been fully sold out. In any case, AI chip makers have not been able to keep pace with demand. As a result of this demand, the surface area of logic used in AI chips continues to grow. In order to raise processing speeds, AI chips have adopted HBM and are implementing ever higher levels of stacking. However, production capacity for HBM has not kept up with demand. As a result, the HBM growth is not having an immediate impact on wafer demand, which is unfortunate. That said, I believe AI chips will definitely be a driver of growth.

Slide 12, please. This slide shows demand forecast for leading-edge wafers used in servers. Demand for AI servers is expected to grow dramatically. Compared to conventional servers, the CAGR for AI servers is expected to be much higher at 14%. Moreover, if you look at the chart on the lower left, AI servers consume high levels of silicon in surface area terms. That is clear from the diagram of the structure of AI chips shared earlier. AI servers consume close to 4x more silicon. AI servers require leading-edge wafers for both logic and DRAM, given the very fine design rules required for devices. We look at the expected growth of AI servers over time in the chart on the right. Currently, 300-millimeter wafer consumption is estimated to be around 800,000 wafers per month. As AI servers increase, leading-edge wafer demand is projected to exceed 1 million wafers per month in 2027. Sumco has a high share in leading-edge wafers, so this is very good news for us.

Slide 13, please. This slide shows expected trends in adoption of AI functionality in smartphones and PCs. This is another topic that is being intensively covered by the press now. We are seeing increasing strong take-up of AI functionality in smartphones and PCs. The adoption of AI functionality will drive higher levels of complexity in chips which will drive up silicon consumption as covered earlier.

Slide 14, please. So how are we thinking about the impact of AI on overall demand? For the leading-edge wafers used in logic and AI chips, which is epi wafers, we estimate average consumption for this year to be around 420,000 to 430,000 wafers per month which feels right given our sales. We expect to see further growth. In 2025, wafer consumption should rise to 570,000 to 580,000 per month. When we get to 2027, it is likely to be around 800,000 wafers per month. As reflected in these figures, we expect a very rapid pace of growth. Of course, the number of wafer makers capable of producing such leading-edge wafers is limited. Obviously, Chinese wafer makers are not at all competitive in this area. There aren't that many wafer makers fabricating leading-edge wafers. I think competition in this area will be limited to the top players.

Slide 15, please. At first glance, your impression is likely to be that the customer wafer inventory months have peaked out. It is true that device demand has finally bottomed out, but it doesn't necessarily mean that it will lead to an immediate pickup in wafer purchases. It will probably take a little more time.

Slide 16, please. This shows the breakout of customer inventories between logic and memory. You can see that inventories for logic appear to be steadily declining. However, if you look at individual chip makers, there appears to be significant divergence, making it tough to generalize. The same could be said of memory as well. We are seeing an increasing trend toward polarization with the gap between winners and losers widening out.

Just as we are seeing winners and laggards emerging in HBM, the same could also be said of logic as well. Only a certain player has the capability for 5-nanometer or 3-nanometer. Those that are falling behind are struggling. So there is a wide range of variance between individual companies which makes it difficult to talk about the industry in general terms. That said, those customers to which Sumco is supplying leading-edge wafers are doing very well, so we have high expectations.

Slide 17, please. Slide 17 is my final slide. In terms of how we see the market over the longer term, if we look at the chip market in 2023, it fell 12% year-on-year. Reflecting the low base for comparison, it is expected to grow 17% to 18% year-on-year in 2024. We Growth in the following year is expected to be around 14% to 15% year-on-year as well. Wafers typically lag chip trends. So 2024 could either be flat year-on-year if we see a recovery in second half or could potentially drop year-on-year if the second half recovery is very slow. However, for 2025, assuming that consumption improves and inventory reverts to more normal levels, we could see a return to normal conditions where consumption equals purchases. If so, we could see year-on-year growth of 17% to 18% in 2025. This completes my section of the presentation.

I will hand over to CFO, Kubozoe, to talk about details of our Q1 earnings.

[Shinichi Kubozoe, executive]

I, Kubozoe will present the earnings from this quarter. Please turn to Slide 19. The Q1 results are shown in the third column from the right. Sales were JPY 93.5 billion and operating income was JPY 8.6 billion. I will talk about the year-on-year and Q-on-Q comparisons on the next page. If you look further down the page, ordinary income was JPY 9.1 billion in Q1. Nonoperating gains and losses was a positive JPY 0.5 billion. Among other things, last year, we received an insurance payment in response to an incident leading to the slight positive in nonoperating gains and losses. Below this, under extraordinary gains, as you know, last year in Q1, we reported a JPY 20.1 billion gain on negative goodwill incurred in conjunction with the acquisition of the polysilicon business. In the absence of this gain in the current Q1 on top of the year-on-year declines in ordinary profit, net profit attributable to owners of the parent declined sharply by JPY 32.6 billion.

CapEx in Q1 was JPY 76.2 billion. We continue to incur greenfield investments on an acceptance basis. Although it is down slightly year-on-year, overall CapEx continued to run at a high level in Q1. In contrast, reflecting the start of a new fiscal year, depreciation was JPY 17.8 billion, down roughly JPY 4 billion Q-on-Q. However, on a year-on-year basis, given continued progress to expand capacity with investments in facilities and the completion of physical structures, Q1 depreciation was up slightly by around JPY 4 billion. Further down the table, the OPM, EBITDA margin and ROE are as shown on this page.

Please turn to Slide 20. This is the analysis of Q-on-Q and year-on-year changes in operating income. First, the sequential analysis on the left. Sales fell JPY 11.6 billion, and operating income dropped JPY 2.6 billion. The yen strengthened slightly versus Q4 2023 by JPY 1.7. The analysis of the JPY 2.6 billion decline in OP is shown below. The positive was the impact of a start of the new fiscal year on depreciation. However, the decline in sales led to a negative impact of JPY 6.1 billion in sales-related variance. Most of the decline is due to lower volumes. In terms of costs, materials and maintenance costs were up slightly. As a result, OP fell JPY 2.6 billion sequentially. Looking at the year-on-year change for Q1, sales fell JPY 16.4 billion and OP dropped JPY 17.3 billion. The yen weakened from JPY 14 depreciating from JPY 133.2 to JPY 147. As a result, as shown in the bar chart below, there was a year-on-year JPY 4.4 billion positive from ForEx. Costs also improved on a year-on-year basis with electricity costs and labor costs down slightly. In contrast, in addition to the increase in depreciation, sales-related variance was a negative JPY 21.2 billion. There was a significant decline in volume.

Even with the positive impact of ForEx on sales, the large negative from sales-related variance reflects the fact that production was running at relatively high levels up to first half 2023, despite some production adjustments. This led to relatively higher levels of inventory from high production levels and finished product inventory. The large-scale negative from sales variance also reflects the impact of lower inventory, but is largely due to the decline in volumes.

Slide 21, please. Slide 21 shows the balance sheet and cash flow statement. On the balance sheet, total assets increased JPY 107.4 billion. This is mainly due to the hefty JPY 63.4 billion increase in tangible and intangible assets on the back of progress on the acceptance of greenfield CapEx and completion of physical structures. Cash and time deposits rose JPY 42.3 billion. I will cover this in explaining the cash flow statement on the right. On liabilities, interest-bearing debt increased JPY 105.1 billion to JPY 329.5 billion. Given expected monetary policy changes from BOJ this year, we chose to front-load our borrowing in Q1 in Japan ahead of anticipated increases in interest rates to cover our needs for 2024.

BOG did, in fact, make a move in March. This was behind the large increase in interest-bearing debt. Going forward, we will fund our remaining greenfield investments for this year from the proceeds which have pushed up cash and time deposits. As shown at the bottom of the table, the equity ratio and D/E ratio have deteriorated slightly versus the end of December 2023, as a result of the front-loading of borrowings. On the cash flow statement to the right, relative to operating cash flow of JPY 19.1 billion Capital expenditures on an acceptance basis were roughly JPY 80 billion, resulting in a negative free cash flow of JPY 60.2 billion. As touched upon earlier, borrowings in Japan pushed up interest-bearing debt. This will be used to cover expected expenditures of roughly JPY 40 billion. The proceeds from the borrowing have pushed up Q1 cash and time deposit as a result.

Slide 23, please. On Page 23, we show our forecast for second quarter. As covered earlier by Chairman, Hashimoto, we project Q2 sales of JPY 99 billion, operating income of JPY 9 billion, ordinary income of JPY 8 billion and net profit attributable to owners of the parent of JPY 5 billion. Depreciation is expected to increase slightly sequentially. The we project JPY 18.3 billion for Q2. While we expect sales to increase Q-on-Q, as you can see, we expect ordinary profit and net profit to be largely flat Q-on-Q. If we look at the year-on-year comparison for first half, sales are expected to fall JPY 28.1 billion and operating profit to drop JPY 29.1 billion. we expect year-on-year declines for both sales and profits for first half.

Slide 24, please. Slide 24 shows the analysis of change in operating profit. First, the Q2 sequential change on the left. Sales is projected to rise JPY 5.5 billion Q-on-Q and OP to increase by JPY 0.4 billion. Our ForEx assumption is for JPY 150 to the dollar, a depreciation of the yen of JPY 3 Q-on-Q. Looking at the expected JPY 0.4 billion Q-on-Q improvement in OP we expect positives from sales-related variance on the back of higher sales and ForEx impact. However, we project an increase in cost with electric power unit prices rising from April and labor costs to rise on annual increases in salary and the intake of new graduates. The net result is a JPY 0.4 billion sequential improvement in OP.

Looking at the analysis of year-on-year change for first half and Q2, as noted in the explanation of Q1 year-on-year changes while there are positives from ForEx and costs, the resulting decline in volumes from lower sales is expected to lead to a significant negative in sales-related variance. As such, we expect first half OP to decline JPY 29.1 billion year-on-year. The reference materials show trends for a number of metrics. Please refer to this at your leisure. This completes my section of the presentation.

[Takayuki Komori, executive]

Thank you. We will now open the floor to questions.

[Takayuki Komori, executive]

We will start with Mr. Enomoto of BofA Securities.

[Takashi Enomoto, analyst]

How should we think about silicon wafer demand? I refer to Slide 17 in asking my question. Your projections for silicon wafer demand were unchanged this time. flat year-on-year. What changes have you seen compared to 3 months ago? Earlier, you referred to the strength of AI, but also mentioned that other conventional applications have not recovered. Can you talk about your overall view, whether you feel that the recovery in demand is earlier or later than you had expected? The reason why I ask is because recent comments made by your peers very widely. Some have revised down while others have said that they are seeing a recovery. Frankly, it is difficult to get a sense for the direction of the silicon wafer market. Could you share your view, please?

[Mayuki Hashimoto, executive]

Our sense is that the market has bottomed out. What is weak is automotive applications. Automotive was actually relatively strong last year, but appears to have subsequently deteriorated. Actually, in terms of 300-millimeter wafer consumption, however, the auto industry is not that large. Instead, much of auto wafer consumption is 200-millimeter. Our sense is that visibility for 200-millimeter is poor. But for us, 200-millimeter wafer does not account for a large part of our sales. The scale of business is less than half of 300-millimeter for Sumco. So for Sumco 300-millimeter is the mainstay.

In 300-millimeter, auto applications are weak but there is a slight recovery in smartphone and PC applications and a solid recovery in AI applications. But for AI, there are capacity constraints on the part of customers which means that volume is unlikely to grow significantly. Even if we talk about HBM and DRAM or logic for AI applications, while the customers are working hard to increase their capacity. These efforts are not translating into an immediate increase in wafer volume, so we are treading water at the moment. That said, our feeling is that demand has bottomed out. As such, we expect wafer consumption for 300-millimeter to gradually improve.

It is a little more difficult to read 200-millimeter demand, although it does feel as if we are crawling the bottom. When we get into 2025, I think we should see an improvement as the macro environment should get better. I don't think it is a situation where we have no visibility at all. That's our sense of the market. Conditions vary by customer as well. So it's tough to generalize. Our customers are strong players that are doing well and have strength in the area of AI, so we would expect to see a recovery. Does that answer your question?

[Takashi Enomoto, analyst]

Compared to 3 months ago, is your sense that conditions have improved?

[Mayuki Hashimoto, executive]

Yes, that's correct. That's reflected in the fact that we overshot our forecast. 3 months ago, when we formulated our forecast, we were more bearish, so we do have a sense that conditions have improved.

[Takayuki Komori, executive]

Next is Mr. Ikeda of Goldman Sachs.

[Atsushi Ikeda, analyst]

I would like to ask about Page 12 and wafer demand forecast related to AI servers. I have a slightly detailed question. You suggest that demand in 2024 would be around 300,000 wafers per month. When you think about DRAM, there are significant differences depending on whether you are basing your assumptions on input or output. What is your yield assumption? Are you assuming that DRAM on an output basis will be around 100,000 wafers per month and that logic accounts for the larger part. Can you talk about the breakdown of the 300,000 wafer per month figure between logic memory and silicon interposers.

[Mayuki Hashimoto, executive]

What we show here is on an output basis. The yields on DRAM and HBM vary significantly by chip maker. So it's hard to generalize, but yields on HBM are worse than conventional DRAM. Our forecasts are on an output basis. If customer yields are taken into account, it may well be that there is upside to expected wafer consumption. That said, fundamentally, customer device capacity is limited. Customers are trying hard to increase capacity, but until the capacity increases, the sample size is too small to draw meaningful conclusions. However, there are DRAM use wafers for HBM, demand for these are very strong with customers clamoring for more and faster.

This is an area where we are seeing increasing polarization amongst customers, particularly for logic and HBM for AI servers. There are top companies that are particularly dominant here, which reflects this trend of polarization. The other customers are trying very hard to catch up. when we talk about logic at 3-nanometer or 5-nanometer, at this time, almost none of the other customers have been able to compete.

[Atsushi Ikeda, analyst]

I have 1 follow-up. With regard to the adoption of AI functionality in smartphones and PCs as shown on Page 13, can you talk about the silicon contents required for this, similar to what you have shown for AI servers?

[Mayuki Hashimoto, executive]

I don't have data for this yet. However, for logic, they would certainly be using leading-edge logic. I think it is unlikely that HBM is being used. Perhaps they are using upgraded conventional DRAM. I don't really know, but as functionality becomes increasingly sophisticated, I would expect to see increased silicon consumption. However, at the moment, the sample size is still too small to get sufficient information to draw any sort of conclusions.

[Atsushi Ikeda, analyst]

Yes, but inquiry levels are high. You indicated that you are feeling more positive.

[Mayuki Hashimoto, executive]

Demand for HPM and leading edge logic is high.

[Takayuki Komori, executive]

Next is Mr. Watabe of Morgan Stanley.

[Takato Watabe, analyst]

Your initial view was that there would be a correction in wafers in the March quarter, but a recovery in the June quarter. It appears that there wasn't much of a correction in the March quarter and profits overshot your forecast. Is this because there were solid LTA deliveries during the quarter or were customers anticipating the recovery in demand in 2025 and 2026. As touched upon earlier, is carrying higher levels of inventory becoming the new normal. Why wasn't there much of a correction in the March quarter?

[Mayuki Hashimoto, executive]

Actually, Sumco saw overall demand drop by around 10% in Q1. In terms of what was weak, there was a significant drop in PW. So there was a sizable correction. But if we look at Q2 and beyond, in logic where the correction has not been that large, demand for leading edge has been very strong. also mainly driven by HBM, overall DRAM has improved, impacting PW. The reason for this is that, as you know, the fabrication for HBM involves many steps, so HBM tends to consume production capacity, taking capacity away from other DRAM. So capacity for conventional DRAM shrink, which tightened the supply of conventional DRAM. So apparently, there are synergistic benefits at work, although it doesn't apply to NAND. DRAM is doing better. So PW is doing better, mainly as a result of what is happening with DRAM.

[Takato Watabe, analyst]

You mentioned that there was a 10% decline in volume. Was that in 300-millimeter?

[Mayuki Hashimoto, executive]

We were down 5% Q-on-Q and the market was also down 5%. The earlier 10% decline was for all diameters, so not just 300-millimeter, but also 200- and 150-millimeter as well. If it is just 300-millimeter, the NAR decline was 5% Q-on-Q in line with the overall market. 200-millimeter and 150-millimeter were both down 10-plus percent.

[Takato Watabe, analyst]

I see. With regard to customer inventory levels, rather than going back to the previous levels of 2 weeks or 1 month, are you seeing customers now shifting to carry inventory levels of around 2 months?

[Mayuki Hashimoto, executive]

I think customers are aiming for inventory levels of around 2 months. Currently, there are some that have inventory of 3 months or more. There is a wide variation between customers with some apparently carrying inventory of close to 6 months. There is significant polarization between winners and losers recently.

[Takayuki Komori, executive]

Next is Mr. Okazaki of Nomura.

[Shigeki Okazaki, analyst]

I would like to ask about how to think about LTA pricing for 300-millimeter wafers. Three months ago, you indicated that it was possible that revenue generation from greenfield capacity could start in the September quarter and that there was a possibility that LTA prices could rise from the September quarter. Can you provide an update on this and also comment on near-term demand trends?

[Mayuki Hashimoto, executive]

Our 300-millimeter at nonconsolidated Sumco is 100% covered by LTAs. There are a variety of contracts. Although it accounts for only a small proportion, there are some where fixed contracts have an adjustment range of plus/minus 5% every year. Some are fixed flat pricing. This accounts for a relatively large number of contracts. Others have step-up pricing. What I was talking about was what might happen if step-up contracts are pushed out. So for example, this would be where a customer was supposed to purchase a certain volume of wafers over a 5-year period which is then pushed out into a 6-year period. With the step-up contracts, prices rise each year. So an extension represents a pushout in timing of price increases. If a customer pushes out delivery, it's tough to get the customer to agree to the price point that applies to the year when they actually take delivery. Instead, if the push out was for volume that was supposed to be delivered in the previous year, we are applying the previous year's price when the delivery is made. What this means is that the actual price point in a given year becomes the weighted average of the prices.

So as deliveries get pushed out for this year, what that means is that deliveries that happen in the second half of this year may be deliveries of last year's volume. So the step-up in prices under LTAs is only realized gradually and prices during a given period are the average of the prices on the deliveries. So there is a combination of different types of contracts the flat pricing, the 5% range adjustment contracts, which is a very small part and the step-up pricing. We understand that some of our peers have 5% range adjustment contracts which put the wafer maker at risk of downward price pressure. However, we are allowing customers to defer volume. Given that customers are not that bullish and they are asking for favors, I think the overall process has been very gentlemanly.

[Shigeki Okazaki, analyst]

So given what you've said, do you think gradual increases in prices will come through in the September quarter, as you had previously suggested?

[Mayuki Hashimoto, executive]

Yes.

[Shigeki Okazaki, analyst]

Do you think that LTA pricing will revert to the originally planned prices in 2025 and 2026? Will we revert to the expected prices for the March 2024 quarter in the second half of 2025?

[Mayuki Hashimoto, executive]

Yes, I guess that's fair.

[Takayuki Komori, executive]

Next is Mr. Miyamoto of SMBC Nikko.

[Go Miyamoto, analyst]

I want to ask a little more about the Q1 Q-on-Q sales-related variance covered in the waterfall chart on Page 20. 3 months ago, you indicated that you were assuming price declines of around JPY 2 billion. Once the actual price impact in line with your assumptions? Is the remaining JPY 4 billion, the result of volume decline. Similarly, on Page 24, the Q-on-Q sales variance impact for the second quarter forecast is a positive JPY 1.8 billion. Is this all volume with no impact from price?

[Shinichi Kubozoe, executive]

With regard to the sequential JPY 2 billion price impact expected for Q1 discussed 3 months ago, the impact was actually slightly smaller. This is a function of customer volumes but price impact was smaller than the expected JPY 2 billion.

[Go Miyamoto, analyst]

Understood. So you were able to limit unit price declines to less than JPY 2 billion in Q1.

[Shinichi Kubozoe, executive]

Yes, with regard to sequential sales variance impact on Page 24 for Q2, we have factored in a very small negative impact of several hundreds of millions of yen from spot prices, but basically, the remainder is related to volume. So there is almost no impact expected from prices.

[Go Miyamoto, analyst]

Understood. On spot prices, do you expect prices to stop falling in the June quarter? Or will there continue to be price declines in the September quarter.

[Mayuki Hashimoto, executive]

I think 300-millimeter prices will be unchanged, but for 200 millimeter, we have generally been able to agree with customers for the most part. So I think it is unlikely that we would see further price declines in the second half of the year. I think prices have dropped a lot already. There is a wide variety of product within 200-millimeter from very cheap to expensive prices, but my sense is that we should start to see a bottoming. Unfortunately, the proportion of spot is high for 200-millimeter and it is a little difficult to read future trends. But in recent price negotiations, I haven't seen big price declines which suggests things are starting to stabilize in my view.

[Takayuki Komori, executive]

Next is Mr. Nishiyama of Citigroup Securities.

[Yuta Nishiyama, analyst]

I believe that there is a polarization in demand with AI-related demand very strong, but a slow recovery in demand from other applications. Sumco has a high market share in leading edge, so I expect that you will be amongst the winners within the industry. This is reflected in the overshoot of your forecast in Q1 and and the expectation for earnings growth in Q2. You appear to be pulling ahead of your peers. At the same time, Sumco's margins are still low versus peers. Although you have strength in high value-added segments of the market, it does not appear to be reflected in your margins compared to peers. You are not guiding for the Q2 EBITDA margin to recover. How should we think about your strength at the top line and your margin compared to peers?

[Mayuki Hashimoto, executive]

Shin-Etsu does not provide segment breakout, so I can't comment. Global wafers recently reported earnings, but there were things that we couldn't make sense of. The EBITDA margin was quite good. However, our only key competitor in leading-edge wafers is really Shin-Etsu. So it's never been that clear to me. We do a lot of analysis, but we don't really understand global wafers earnings. Siltronic is similarly structured to us, so I have a good sense for how their earnings are likely to trend. I don't have a lot of clarity around other peers. Are you doing peer comparisons? What do you look for? We are doing detailed analysis, but there are things that don't make sense to us.

[Yuta Nishiyama, analyst]

I see, thank you. You say analyzing margins is tough, but with regard to the outlook for demand in Q2 and beyond, as covered in the first question, you indicated that the gap between the successful players and the laggards is widening. Do you believe that you are doing well in terms of the outlook for demand because of your focus on leading edge?

[Mayuki Hashimoto, executive]

Our basic policy is to only produce what we can sell so we don't produce to build inventory. If you want to make profitability look better, you can actually generate better margins when building inventory to a certain extent, but that's not how we operate. In fact, we had actually been working down inventory slightly. This doesn't make sense to me either. Margins improve on paper if you are building inventory.

[Takayuki Komori, executive]

Next is Mr. Yoshida of CLSA.

[Yu Yoshida, analyst]

When do you think you will start negotiating the next round of LTAs? What is your sense of customer response to the next round of LTAs? Near-term demand has been softer, but do you think you can raise prices further from current levels or maintain price levels in the next round? Any color you can provide on the next round of LTAs would be greatly appreciated.

[Mayuki Hashimoto, executive]

The next round of LTAs would be after 2026, but since there have been deferrals to the current LTAs, actually, we are covered out to 2027. With regard to the next round of LTAs, probably we will see the next cycle kick in from 2026 with shortages starting to emerge at that time. This suggests that the timing is likely to be favorable.

[Yu Yoshida, analyst]

I see, so the current contracts would have started to expire in 2026, but because of the pushouts of deliveries you are covered up to 2027. So you would begin negotiating in 2026?

[Mayuki Hashimoto, executive]

Yes, I think so.

[Yu Yoshida, analyst]

I see, but you haven't started specific negotiations yet. Is that correct?

[Mayuki Hashimoto, executive]

We aren't even close to that stage yet. The negotiations now are still about customers wanting to defer deliveries.

[Takayuki Komori, executive]

Next is Mr. Nakada of JPMorgan.

[Yasuhiro Nakada, analyst]

I would like to ask about the outlook for depreciation First, can I confirm that the annual depreciation guidance for 2024 is unchanged from JPY 90 billion. If that is the case, then it implies a half-on-half increase of JPY 18 billion, in which quarter do you expect to see an increase. Also, the outstanding balance of construction in progress is up quite significantly. When do you expect depreciation to hit a peak?

[Shinichi Kubozoe, executive]

Previously, we had guided for JPY 90 billion, but it may be slightly lower due to timing of ramping up facilities, although I don't have specific figures for the second half yet. I do think it will be close. In terms of the half-on-half increase, it is likely that the increase will be more than JPY 10 billion. Peak depreciation is likely to be in the second half of 2025, although we are expanding in both Japan and Taiwan.

[Yasuhiro Nakada, analyst]

So you expect to hit a quarterly peak for depreciation in the second half of 2025. Is that correct?

[Shinichi Kubozoe, executive]

We expect depreciation to be at high peak levels next year and the following year.

[Yasuhiro Nakada, analyst]

Understood. So given that Chairman Hashimoto, has said that you will not be producing in excess of demand, Sumco will not be ramping up facilities before demand recovers and therefore, depreciation will not increase either. Is that right?

[Mayuki Hashimoto, executive]

Yes, that's right. What may be challenging is an increase in depreciation on unused assets. It may be tough from an earnings perspective if the assets haven't been ramped up yet. Still, you could choose to look at it as being similar to accelerated depreciation. By doing this, the depreciation burden will be lighter when volumes pick up.

[Yasuhiro Nakada, analyst]

So in other words, if demand tightens in 2026 and 2027, but depreciation has already peaked out, this would be favorable from a margin perspective?

[Mayuki Hashimoto, executive]

Yes, that's it. I think this is the time to be hunkering down.

[Takayuki Komori, executive]

Next is Mr. Yamada of Mizuho Securities.

[Mikiya Yamada, analyst]

With regard to the previous question about depreciation, basically, you will not be ramping up new facilities in the absence of demand. But for a leading-edge plant that will work off of next-generation designs, I believe there is a project underway where you have received a JPY 75 billion subsidy. Will the timing for this also be pushed out. Also, am I correct in assuming that all of your customers use leading edge, you are the dominant player in leading edge. So any customer that has contracts with Sumco basically needs leading edge. Is that correct? Even when they are not using leading-edge wafers, the fact that they have need for leading edge means that there is no incentive for Sumco's customers to breach their LTAs with Selco. Is that right? That is my understanding, but is this correct? If so, even if the construction and progress account increases, you can still be comfortable in investing in the next round in my view. I would like to confirm that my assumption is correct.

[Mayuki Hashimoto, executive]

It is difficult for me to comment on the project where we would receive a subsidy as it is still under consideration. One thing that I can say is that construction demand in Q2 is very high and construction lead times are getting longer. We are looking at many things, but that is the situation. The semiconductor industry is a growth industry. So abandoning the project is not a choice we would make but it is very challenging given that lead times continue to get longer. Contractors are very busy. The new facilities will all be for leading edge. Obviously, over time, leading edge becomes commoditized but the current cutting edge is 2 nanometer, I guess.

[Mikiya Yamada, analyst]

Yes.

[Mayuki Hashimoto, executive]

Chairman Hashimoto. There are 2 types of GAA or gate all around, 110 and 100. 100 has already been achieved, and we expect to see very strong growth, but everyone is really struggling with 110 GAA, as are we. The customer has said that it is very challenging. So we don't know when this will come, but 100 is already starting. Sumco is expanding capacity in leading-edge epi, so I think we will have to gradually start up this capacity.

[Mikiya Yamada, analyst]

So with regard to 110 Crystalline alignment facilities, the leading-edge facilities from ingot pulling onward must be ramped up regardless of demand. There are customers that need this. So even if overall market supply/demand is disrupted, that would have no negative impact at all on Sumco's contract. Is that correct?

[Mayuki Hashimoto, executive]

Yes, that's correct.

[Takayuki Komori, executive]

Thank you for your questions. We will end the meeting here. Thank you to everyone for joining the Q1 fiscal 2024 results briefing. We are grateful for your participation today.

[Mayuki Hashimoto, executive]

Thank you.

[Shinichi Kubozoe, executive]

Thank you.