Sumco Corp
TSE:3436
Sumco Corp
Sumco Corp., a notable player in the semiconductor industry, emerged as a crucial figure in the high-stakes game of enabling technological advancement. The company specializes in the production of silicon wafers, the foundational material from which the intricate pathways of microchips are carved. These wafers represent a critical component for semiconductor manufacturers, who fashion them into the devices that power modern electronics. Sumco's operations are deeply rooted in innovation and precision, engineering its wafers to suit a diverse range of applications from consumer electronics to automotive technologies. As a key supplier, Sumco is instrumental in the tech supply chain, where its products are highly demanded by companies crafting everything from smartphones to smart appliances.
What sets Sumco apart is its relentless pursuit of quality and efficiency, carving a niche in an industry dictated by high barriers to entry and constant technological evolution. The company's revenue streams heavily rely on their ability to maintain relationships with large semiconductor manufacturers, providing them with high-quality, customized wafers that meet the stringent requirements of cutting-edge chip designs. By leveraging advanced manufacturing technologies and maintaining a robust R&D framework, Sumco not only aligns itself with the rapid pace of technological development but also secures its financial footing. As these industries continue to expand with the rise of artificial intelligence, electric vehicles, and the Internet of Things, Sumco positions itself at the forefront, ensuring that it remains a vital cog in the machinery of digital innovation.
Earnings Calls
In Q4 of fiscal 2024, Sumco exceeded forecasts with JPY 2 billion in additional sales, largely from strong demand for leading-edge wafers. However, they expect Q1 revenues to increase by JPY 2 billion but project an operational profit decline of JPY 2.5 billion due to rising depreciation costs, which will surge by JPY 2.3 billion. For the year, they forecast a full-year dividend of JPY 21, maintaining a payout ratio of 39.7%. Despite a competitive environment, particularly with Chinese manufacturers, Sumco aims to strengthen its focus on leading-edge products.
You don't have any saved screeners yet
Thank you for your participation today. This is the results briefing for the fourth quarter of the fiscal year ending December 2024. Before starting the presentation, allow me to confirm today's materials, which consist of two items, the brief statement on consolidated financials for Q4 fiscal 2024 and the presentation deck entitled Results for Q4 fiscal 2024. In addition, separate to the materials on earnings, also posted a press release regarding the announcement on reorganization of small-diameter silicon wafer production, which is also available on our website. Please refer to this as well.
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 Representative Director 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.
I am Chairman Hashimoto. I will start the review on Slide 5. This is a summary of the results for the fourth quarter of fiscal 2024. We were able to once again overshoot our fourth quarter forecast by around JPY 2 billion. Most of the overshoot is the result of higher-than-expected volumes. There was also a slight contribution from mix. Strong sales of leading-edge wafers, which command high price points contributed to the slight overshoot.
Turning to the earnings forecast for the first quarter of 2025. While we expect sales to grow JPY 2 billion Q-on-Q, we project a JPY 2.5 billion Q-on-Q drop in operating profit. The vast majority of the decline in profits is the result of higher depreciation expenses. There are other contributing factors, including ForEx and expected cost increases, but the single biggest element is the JPY 2.3 billion increase in depreciation which will depress profits.
Next slide, please. This slide shows shareholder returns, our dividend guidance. The fiscal year-end dividend will be JPY 6, which will bring the full year DPS to JPY 21 for a dividend payout ratio of 39.7%. Next page, please. This is the trend for 200-millimeter wafers. As you can see, after hitting peak levels in 2022, 200-millimeter wafers have subsequently fallen to lows that have not been experienced in recent years. The weakness in EVs and other consumer-driven markets was a major factor, but the intensification of competition with China was also significant. In particular, sales of 200-millimeter wafers to China have been relatively sizable, but now sales of 200-millimeter in China have virtually dried up. With the exception of extremely technically challenging wafers, which represent a very small portion of volume, Chinese wafer makers have developed the ability to manufacture 200-millimeter wafers. As a result, sales of 200-millimeter wafers in China have dropped dramatically. This is one reason for the weak trend. As a result, we no longer expect to see a significant recovery.
This is the same slide for 300-millimeter wafers. As you can see, 300-millimeter bottomed in first quarter 2024, but although there has been a recovery, the momentum of the rebound has been a little weak compared to previous upcycles. While semiconductor sales have continued to hit new record highs in value terms, the strength has been concentrated in generative AI. The recovery in sales trends for other applications is weak. The reason why semiconductor sales continue to surge is because the price point for AI chips is 10 to 20x higher than conventional chips. So even if volumes are not that high, the impact on the value of sales is 10x higher than conventional chips. So while the market is growing in value terms, growth in volume terms is muted, although the content has changed.
Next slide, please. Looking at market conditions during fourth quarter, we saw a steady but slow continued recovery in 300-millimeter wafers. 200-millimeter and smaller-diameter wafers remained weak. Fortunately, customers continue to respect LTA prices for 300-millimeter. However, for 200-millimeter, there aren't many LTAs to begin with, so very simple 200-millimeter and smaller wafers saw significant price decline. On the outlook for 300-millimeter wafers in first quarter, demand for leading-edge applications will continue to be strong, but the recovery in legacy applications is expected to remain weak. We are continuing to see incremental improvements, but we are not getting the strong recovery we had expected.
As well, recently, there has been a correction in memory. While logic continues to grow, led by leading-edge, the recovery in NAND memory has become sluggish. I have already commented on the situation for 200-millimeter. Our expectation is that the current conditions will persist. In terms of prices, 300-millimeter LTA prices continue to be respected by customers. However, while there are differences by geographies and applications, spot prices, especially for less technically challenging products for 200-millimeter and also 300-millimeter as well have fallen sharply. Prices for more demanding applications have remained relatively firm.
On the outlook for semiconductor demand, AI data center applications will continue to be a strong driver. However, demand for non-AI-related applications is not very good. The market has become polarized. That said, more recently, we have seen slight signs of a recovery in PCs. We hope to see a recovery in conventional applications, but conditions are still very weak at this time. As a result of all of the above, the overall trend for 300-millimeter as a whole is a gradual recovery. For 200-millimeter, while the market is weak, competition with Chinese wafer players is also intensifying. Chinese players now dominate in terms of demand for consumer-related applications in China, making it increasingly challenging for us to access this market.
Although this had already been a trend in the China market for some time, since the U.S. has stepped up its efforts to decouple from China, the Chinese government has been very clear in guiding chip makers to buy wafers made in China. Next page, please.
We have announced that we will be shutting the Miyazaki plant. This slide looks at the market environment for 100-millimeter and smaller wafers. It is certainly true that the smaller diameters are winding down and have been in a prolonged decline for some time. In terms of 150-millimeter and 200-millimeter wafers, much of what is done with 150-millimeters can be done on 200-millimeter wafers. Given the economics for 200-millimeter are better, those customers with relative sound finances have been and are transitioning to 200-millimeter on an ongoing basis.
Automotive and consumer applications have been particularly weak and Chinese wafer makers have increasingly become a force in the market. Although this had already been the case, Chinese players are fully able to manufacture 150-millimeter wafers, competing directly with the Chinese means cost competition, so producing in Japan is very difficult and profitability is challenging. Next page, please.
We have been focused on reforming our business structure to strengthen our capabilities in leading-edge. What we have seen is the evolution of technology has accelerated significantly. So customers with LTAs have virtually no inventory of leading-edge wafers, but customer inventory of non-leading-edge logic wafers at 5-nanometers or higher is at high levels. Because customers have LTAs in place, if they were able to take leading-edge wafers, that would be one thing, but not all customers can do leading-edge applications. So customers that focus on conventional applications who have LTAs are receiving wafers in spite of weak demand for legacy products, which is boosting inventory levels at a time when inventory turnover is low.
Fortunately, Sumco has top market share in leading-edge, so we are focusing on replacing wafers produced for conventional applications by increasing the proportion of leading-edge wafers. Our immediate priority is to ramp up the new plant as quickly as possible. We will also upgrade existing plants to enable the production of leading-edge wafers. We are accelerating our structural reform efforts. As a part of this, the production capacity for 150-millimeter wafers where the market is very slow, technological differentiation is increasingly difficult, and unfortunately, profitability is very poor. I believe it is better to reallocate management resources to new areas. In any case, we will require human resources to ramp up production at the new plant, so staff from Miyazaki will be reassigned to the new plant. Our biggest objective is to make more effective use of our management resources. So as a part of our business structure reforms, we will shut down the Miyazaki plant.
200-millimeter production at Miyazaki will be transferred to the Nagasaki and Imari plants. 150-millimeter production is to be transferred to the highly cost competitive Indonesia plant. For those 150-millimeter products that have limited future potential, many of the customers are also buyers of our 200-millimeter and 300-millimeter wafers. So it won't be possible to simply pull the plug on production. The plan is to ask these customers to agree to an appropriate grace period, and we will also build inventory before ceasing production. However, the monocrystalline production at Miyazaki is a special type FZ or floated zone, so this operation will be maintained. As a result of all of this, we incurred an extraordinary loss of JPY 5.8 billion in 2024. Next slide, please.
In the next few slides, we explore the outlook for technology for leading-edge devices using wafer bonding. Recently, we have seen a sharp acceleration in the pace of innovation. In response to the waves of new technology, Sumco is stepping up its own R&D efforts. I will highlight some of the key trends.
Next slide, please. The key word going forward is the transition to 3D structures. For example, Logic at 2-nanometer will require a structure different from what is currently used. Of course, GAA is an example of this. But going forward, we are likely to see bonding of 2 or even 3 wafers. The reason for adopting bonding is because there is a need to avoid increasing interconnect resistance from scaling by separating the power interconnect and signal interconnect. Up to now, the signal interconnect and the power interconnect were located together. The resulting proximity led to power losses as well as generating heat. More recently, there is a move towards bonding, which is referred to as BSPDN, which is backside power delivery network, to address this. There is a lot of R&D being done in this area. Sumco is involved in this R&D. However, while bonding wafers may appear to be easy at first glance, in fact, it is very challenging. It has also led to the emergence of new specifications as well. So Sumco is involved in many development initiatives with our customers.
Next is DRAM. This is a DRAM that is being used in generative AI. By the way, the Logic we just talked about is also being used for generative AI. DRAMs are increasingly being stacked, so there is a need for an interposer at the bottom. This is also not that easy to achieve. The structure requires polished wafers. We are also seeing an increasing trend towards 3D structures in NAND as well. There is a customer that is already in mass production with this structure. The structure is referred to as CBA or CMOS directly bonded to array in which a logic die is positioned at the bottom of a memory stack. This also requires wafer bonding. Although volumes compared to the 3 previous examples are not that significant, in CMOS image sensors, the trend is for increasingly larger photodiodes.
The logic circuitry had traditionally been positioned next to the photodiode, but the trend is shifting towards vertical stacking. And increasingly, we are seeing adoption of this structure. At this stage, it is still too early to quantify the volume of wafers that might be required. What we can say is that the structures will definitely change.
Next slide, please. This slide shows a simplified depiction of the various structures. So for example, in bonding 2 wafers face-to-face, one of the wafers is flipped upside down, after which the upper wafer with the power interconnect is ground down. The combination is used as an integrated structure. Approaches vary from customer to customer. There are also customers that use support wafers as shown in the back-to-face wafer bonding diagram. They bond the support wafer, then flip it over to grind down the first semiconductor wafer. They next bond this to a second semiconductor wafer before removing the support wafer. There are bonding structures that are used for logic and memory. There are many customers that are using support wafer technologies in many different ways.
The next example is also for a support wafer. In this case, the bonding is not for 2 semiconductor wafers, but instead the bonding of a power interconnect to a wafer. The support wafer is to create a structure similar to what I described earlier in talking about logic with the support wafer removed at the end of the process. Apparently, the removed sport wafer can be reused. In any case, there are many different types of polished wafers that are likely to become necessary. Customers are actively conducting R&D and Sumco is working alongside its customers in the development process.
Next slide, please. So what is the outlook for DRAM going forward? DRAM is expected to show strong growth on the back of strong demand for HBM used for AI data center applications. We project a bit growth CAGR of 19%. Bit growth from scaling is estimated to be around 10%, so the remainder represents wafer volume growth.
Next slide, please. This shows 300-millimeter demand forecast for DRAM. We project a CAGR of around 7%. HBM is likely to be the key driver of growth, while conventional DRAM is unlikely to grow much. That said, within this, conventional DRAM using EUV will account for an increasing proportion. Increasingly, there will be a need to transition to using EUV. This is an area where Chinese players do not have capability.
Next slide, please. This is a slide for NAND to bit growth. We are projecting a bit growth CAGR of 21%. Similar to DRAM, the contribution from scaling is around 10%, although there are instances where it is close to 20%, so the potential for wafer volume growth is not that high. In terms of wafer volume growth, it is more likely to come from the trend towards 3D structures. So NAND is unlikely to be a big driver of wafer volumes in my view. The number of layers is likely to continue to rise significantly. This is an area where the Chinese have significantly elevated their technological expertise. The key Chinese player in this space is producing very good chips.
Next slide, please. As a result, leading-edge NAND is likely to use wafer bonding. For instance, to bond control circuitry, I expect there will be an increase in the use of the wafer bonding process. But in terms of wafer volume growth, I would expect a mid-single-digit CAGR at most.
Next slide, please. This slide shows customer wafer inventory trends. Inventory had come down, but more recently, we have seen inventory rising again.
Next slide, please. If we look at the trends for logic and memory separately. It may be a little surprising, but inventory in logic has increased, as has inventory in memory. Within memory, NAND is clearly in a correction phase. For logic, as mentioned earlier, wafer input momentum for leading edge is very strong, but the trend for conventional applications is not very strong. I think that the sluggish trend has meant that customer inventory turnover is very slow. So for instance, customer volumes for 28-nanometer or 16-nanometer chips have declined from previous levels and production levels are low. But because customers have LTAs in place, they are still buying wafers even if they have delayed taking some of the scheduled deliveries. Not all wafer manufacturers are capable of producing at leading-edge design rules, but wafer demand leading-edge is surging, so I suspect they are prioritizing leading-edge, which exacerbates the poor conditions for inventory in conventional products.
This completes my section of the presentation. I will hand over to CFO, Kubozoe, to talk about the details of our Q4 earnings.
I, Kubozoe, will present the earnings for this quarter. Please turn to Slide 22. First, the results for fiscal 2024. Please look at the second column from the right. On a full year basis, sales were JPY 396.6 billion, and operating income was JPY 36.9 billion for declines of JPY 29.3 billion and JPY 36.1 billion, respectively. Non-operating income was a positive JPY 0.5 billion with some lumpiness in terms of the quarterly sequential progression into fourth quarter. This was because in terms of ForEx translation, the yen trended stronger in third quarter, but weaker in fourth quarter. As well, we received a small amount in subsidies in fourth quarter, resulting in a positive. As a result, on a full year basis, ordinary profit was JPY 37.4 billion, largely in line with operating income.
As discussed earlier, we took an extraordinary loss of JPY 5.8 billion related to the reorganization of the Miyazaki plant in fiscal 2024. In fiscal 2023, we reported extraordinary gains related to the acquisition of the polysilicon business from Mitsubishi Materials, which has led to the year-on-year deterioration of JPY 25.9 billion. Net profit attributable to owners of the parent was JPY 19.8 billion, down JPY 44 billion year-on-year.
CapEx on an acceptance basis was JPY 214.9 billion, down roughly JPY 100 billion from the peak of JPY 315.4 billion of fiscal 2023. This reflects a slight decline of greenfield investments on an acceptance basis. Although CapEx exceeded JPY 200 billion, it is down from the 2023 level. In contrast, depreciation expenses have been gradually increasing. It was JPY 78.9 billion on a full year basis in 2024, up JPY 7.5 billion versus 2023. Within depreciation, operating depreciation rose JPY 5.7 billion year-on-year to JPY 77.1 billion. As you can see in the lower part of the table, OPM was 9.3% and EBITDA margin was 28.8%.
Next slide, please. In the analysis of sequential changes to quarterly operating profit, sales increased by JPY 1.6 billion from third quarter, while operating profit fell JPY 2.1 billion. If you look at the waterfall chart below, the increase in depreciation expenses accounted for the majority of the decline in profit. While there was a negative impact from ForEx, this was largely offset by solid cost trends and a positive from sales-related variance, effectively leaving depreciation as a key factor for sequential change into fourth quarter.
On a full year year-on-year basis, as noted earlier, sales and operating income fell JPY 29.3 billion and JPY 36.1 billion, respectively. The key contributor was sales-related variance, reflecting the significant impact of a decline in volumes in the first half of 2024. In addition, small diameters were also a drag on profits continuing to perform poorly, resulting in a JPY 56.8 billion negative year-on-year from overall sales variance. Depreciation expenses also increased JPY 5.7 billion year-on-year. However, reflecting the JPY 11 weakening of the yen versus the dollar, ForEx was a tailwind. Cost was also a positive factor as a result of lower electric power unit prices and lower labor costs following a review of labor expense levels on top of relatively steady production conditions. Despite this, overall profits fell JPY 36.1 billion year-on-year, reflecting the substantial negative from sales-related variance.
On this slide, I will cover balance sheet and cash flow. Looking at the balance sheet, total assets were JPY 1,172.6 billion, up roughly JPY 100 billion compared to the end of fiscal 2023. The significant increase is the result of a JPY 140 billion rise in tangible and intangible assets from the end of fiscal 2023 to JPY 699.6 billion, as shown in the middle of the table. This reflects the increase in buildings and structures. In addition, equipment and facilities will commence operations from this year onward, which has boosted the amount under construction progress significantly.
However, there was also a significant change in cash and deposits as shown at the top of the table, which fell JPY 60.7 billion to JPY 95.6 billion related to the acquisition of tangible assets. Total liabilities were JPY 515.4 billion, of which interest-bearing debt was JPY 353.9 billion, up JPY 129.5 billion. On top of the increase in borrowing in Japan in 2024, including this year's demand for funds, our Taiwanese subsidiary borrowed funds for CapEx. The combined total borrowings were slightly less than JPY 130 billion, resulting in the increase in interest-bearing debt. Net assets increased JPY 21.7 billion on the back of increases in retained earnings and others. Based on this, the equity-to-asset ratio was 50.5% and the D/E ratio on a gross basis was 1.6x, deteriorating year-on-year by 2.8 percentage points and around 0.2x, respectively.
On the right, we show cash flow. Operating cash flow was JPY 69.6 billion. However, the outflow of cash flow for investment activities was JPY 247.8 billion, reflecting higher CapEx expenditures on an acceptance basis, and net facilities-related payables under others. The resulting free cash flow was a negative JPY 178 billion. The negative free cash flow was covered by the increase in interest-bearing debt, as discussed earlier, in cash and deposits.
Jumping forward to Page 26, I will now discuss our earnings forecast. The projections for first quarter 2025 are highlighted in the middle of the table. As shown in the lower part of the table, our ForEx assumption is JPY 155 to the dollar. We are guiding for sales of JPY 102 billion, operating income of JPY 4.5 billion, ordinary income of JPY 4 billion and net profit attributable to owners of the parent of JPY 1.5 billion. Depreciation is projected to increase slightly less than JPY 2 billion sequentially to JPY 24.6 billion. Depreciation expenses for greenfield investments are gradually starting to kick in.
On the next slide, we show the analysis of changes in operating income. On a Q-on-Q basis, sales are expected to rise JPY 2 billion, but operating income to fall JPY 2.5 billion. The ForEx assumption will change to JPY 155 to the dollar from JPY 149 in fourth quarter for a positive impact of JPY 2.1 billion. However, depreciation expenses, as noted earlier, will rise. Costs are also projected to rise a modest JPY 1.5 billion Q-on-Q, reflecting higher unit prices for materials and the start of a new fiscal year. In addition, we expect to incur maintenance expenses and to be impacted by the timing of expense payments as well.
On sales-related variance, reflecting the impact of the ongoing weakness of smaller-diameter wafers, we project a negative of around JPY 0.8 billion. Ultimately, however, it is the increase in depreciation expense that is driving down operating income. On a year-on-year basis for first quarter, the increase in depreciation expense is the major contributor to lower OP in spite of the year-on-year positives from sales-related variance, with volumes up slightly and ForEx impact. In addition, I mentioned the impact of higher unit prices for materials.
In total, overall OP is expected to decline around JPY 4 billion year-on-year in first quarter.
Next, on Page 29 in the reference section, we show historical trends for earnings. The dark blue bar is EBITDA on a value basis. As you can see, we have been able to maintain a level of around JPY 29 billion per quarter. Please review the other items at your leisure. This completes my section of the presentation.
Mr. Komori. Thank you. We will now open the floor to questions. We will start with Mr. Watabe of Morgan Stanley MUFG Securities. Mr. Watabe?
In an environment where customer inventory levels remain persistently high, I noticed that you have not included the 300-millimeter wafer global capacity and demand forecast slide in the presentation this time. Can you elaborate on the view you currently have on demand volume and the supply-demand balance at this time.
As you know, forecasting is notoriously difficult. But when you then add in the risk of political constraints, all bets are off. At this time, based on the information I have been able to gather, there were still shortages in 300-millimeter leading-edge wafers. Up to this point, given that we were in an environment of rising demand, we had been focused on showing volume with China accounting for a significant portion of that rising demand. Now that we have a complete decoupling, China chipmakers are being strongly instructed by their government to always have Chinese-made wafers account for a certain proportion of their procurement.
Regardless of how bad the yields are, they are to buy these wafers to produce chips. The current situation is that the government covers the full amount of losses incurred as a result of this policy. So with the exception of leading-edge products that can only be produced by us, competition in the China market is extremely intense. To quantify this, for instance, in Sumco's case, China represents about 10% of our sales. Within this, there are products that the Chinese wafers are not able to produce at all. So we can do business in wafers that the Chinese absolutely want to have, but I believe that it won't be possible for us to sell wafers for conventional applications in China given the current political situation. The potential for growth if we are limited to competing in the free world would not be that strong.
So I think that we need to focus on quality at this juncture rather than volume. As such, we must swiftly implement structural reforms to our business. Up to this point, we might have chosen to wait a little more in making a decision on the Miyazaki plant, but this is not a time for being complacent. What's more, as we ramp up the leading-edge wafer plant, we will need to hire. So on the one hand, we had a plant that was loss-making where sales were in decline, while on the other hand, we have a need to hire. This situation cannot be justified from an economic perspective. This is why we are accelerating structural reforms. What's more, going forward, we will accelerate the modernization and upgrading of existing plants rather than capacity expansion. Unfortunately, we cannot produce leading-edge wafers with older facilities. So the key going forward will be determining how best to modernize and at what pace, in my view.
Next is Mr. Enomoto of BofA Securities.
With regard to small-diameter wafers, you are closing the Miyazaki plant. Please talk about the impact you expect the reorganization to have. Also, if current trends persist, I think you will need to take some sort of action for the other small-diameter plants, for example, Chitose or Phoenix. You have a number of small-diameter plants. What sort of measures are you considering?
First, the small-diameter wafers produced Chitose are highly specialized wafers that cannot be produced by wafer makers that focus on conventional wafers. This is because we are not just simply selling wafers, but actually depositing an epi layer on the wafer and then layering an oxidation film on top. The degree of processing difficulty is very high. Therefore, I think that Chitose will continue to operate going forward given that it has customers in Japan.
On the U.S. plant, there is surprisingly solid demand for 200-millimeter in the U.S. And if the U.S. imposes tariffs on Chinese wafers to keep them out of the U.S. market, there will be a need to produce conventional small-diameter wafers in the U.S. So there may be a positive Trump impact on this plant. We will wait to see how this plays out before making any decision. So we will need to take some time to monitor the situation. That said, I don't think 200-meter wafers will completely go away. The question will be the level at which the market settles.
Also, there is a question of how big the geopolitical impact might be. This is difficult to quantify and is something we are monitoring closely. With regard to the 150-millimeter wafer production to be transferred to Indonesia, I expect that Indonesia will be very busy, but fortunately, labor costs in Indonesia are cheaper, so production costs will be significantly lower than Miyazaki. Given that Indonesia would be able to produce at a cost level where we could compete in China to a certain extent, I would expect 150-millimeter wafer production in Indonesia can be maintained for some time. What is important about the structural reforms this time is the human resources. If we left Miyazaki as is, it is a standalone plan, so ultimately, we would need to do layoffs or offer early retirement. However, as we are ramping up the new plant, we need to hire several hundreds of staff. So it is possible to reassign the human resources of Miyazaki.
Our objective this time is to make the most effective use of the management resources we have as we make structural reforms to shift our focus on producing more leading-edge products. This is the first stage of the process. Going forward, and this is true for Nagasaki and Imari, we need to replace equipment to enable plants that are doing 28- or 40- nano for 5-nano and finer design rules. If we don't change the content of what we do, we won't be able to produce competitive products in Japan in my view. This is why we are stepping up our reforms.
On the magnitude of the impact, was Miyazaki in the black?
Miyazaki was loss-making. It was a drag on overall profitability. So first of all, those losses will go away because the plant will be shut down. However, we do have customer relationships, which prevent us from shutting down immediately. This can't be helped. These are customers that also buy 200-millimeter and 300-millimeter wafers from us. In terms of the magnitude of the impact, at a minimum, I expect that we can recoup the 2024 extra orders over the next few years.
Next is Mr. Ikeda of Goldman Sachs Securities.
Thank you for the detailed technology discussion of wafer bonding at this time. With regard to Page 18 and NAND, I would like to confirm that the wafer bonding portion that you highlight does not include the logic portion. You provided information on memory demand, but could you provide a sense for the total CAGR for 2024 to 2027 for incremental 300-millimeter wafers for logic, interposers, et cetera? I know that you said predicting was difficult, but...
I don't know. It's still very early days.
Yes, I understand. Including this impact, how big of an impact do you think there will be from the emergence of local Chinese wafer makers outside of China going forward? In particular, will there be a big impact on LTAs from 2027 onwards. At that point, there should be more of a shift to leading-edge, but can you talk about the long-term business?
If we talk just about Sumco, then the proportion of leading-edge is likely to be very high in 2027. From that perspective, I don't think there will be that much of an impact from Chinese players. They can't do leading-edge and getting into leading-edge would be very challenging. That said, if we look at the customers of China wafer makers, many are Western customers, although they appear to be only sampling the product. The customers appear to be working to avoid becoming overly reliant by not buying Chinese wafers in large volumes. I don't think there will be much impact from China-made wafers outside of China.
Would I be correct in assuming that wafer bonding, for example, would also be very difficult?
Actually, there is easy wafer bonding and difficult wafer bonding. Support wafers are easy. So this could be done with China-made wafers. However, while customers are sampling China-made wafers, there is a question of how much they would choose to procure. So for example, even if they held beauty patents to drive competition for support wafers to select a supplier, there seems to be a trend that suggests that they might well choose not to use Chinese suppliers.
Also, on the previous point about NAND, the wafer bonding of control interconnects is being done in Japan using CBA. But there is a major Chinese player that is doing wafer bonding with logic. So while there are different types, I think in NAND, CBA is likely to grow. This is controlled directly bonded to array. I imagine that this will require wafers for the control portion.
Next is Mr. Nishiyama of Citigroup Securities.
I have a question about the quarterly forecast. You indicated that you expect a slight Q-on-Q increase in March quarter sales, but a decline in profits. Can you comment on your expectations for sales and profit trends for the June quarter onward? You have said that for sales, you expect a gradual recovery in 300-millimeter, but your peers read of the situation is that conditions will be very tough to the end of the first half. What is your view? On profit, it is necessary to take into account the impact of depreciation expenses, but please talk about the directional trend factoring in your expectations for the trend in depreciation expenses?
Depreciation expenses will rise sequentially, so profits will hinge on the demand outlook going forward. With the most advanced leading-edge production very busy, if production is solid, it would then be a question of how much this could offset depreciation expenses. Also, ForEx rates have recently been very volatile. So this complicates the picture. Given the range of volatility is large, I think trying to read this trend at the moment is very tough.
So volume will go gradually and mix will improve, but profits will hinge on depreciation. The forecast for first quarter is an increase of JPY 2.3 billion. Given this is the first quarter of the year, depreciation resets at a lower level, so should we assume that the pace of incremental increases will step up in second quarter and beyond? Also, what are you budgeting for on a full year basis?
Last year, I once mentioned a figure of JPY 150 billion. Subsequently, on the back of changes in the demand environment, we are in the process of reexamining depreciation amount and the timing of the start of depreciation. In addition, recently, given the strong demand for the most advanced leading-edge wafers, we are also currently considering a number of measures to respond to various scenarios in terms of how we ramp up the greenfield facilities. As we are in the midst of considering specifics about individual facilities and when we might start operations, I can't provide a specific number at this time or comment quantitatively. What I will say is that I think level for this year as a whole is likely to be several tens of billions of yen lower than the figure mentioned last year, although we're still in the midst of deliberating when operations will start.
So the image is several tens of billions of yen less than JPY 150 billion. Is that correct?
Yes.
I see. Understood.
That said, in terms of what the figure for second quarter is likely to be, we have only just disclosed our projection for first quarter, so we will continue to review figures for a while longer.
Next is Mr. Miyamoto of SMBC Nikko Securities.
I want to ask about LTAs. You said the customers are respecting LTA prices. But can you comment on whether you are seeing an increase in the number of contracts being renegotiated? The reason why I ask is because downstream device makers have been saying that they are reviewing LTAs for materials. Also, there are some device makers where penalty payments have started turning up on the P&Ls. When I look at your P&L, I don't see any signs that you have received penalty payment, but are you seeing an increase in the number of LTA renegotiations? If so, how are you planning to respond?
In terms of renegotiations or negotiations, in instances where the customer's capacity utilization has dropped significantly, and there is no way that the customer can use all of its wafers and their warehouses are full, we are accommodating requests for pushouts to delivery. What we have seen is the term of our LTAs has been getting longer with some now running to 2028. We have been entertaining these kind of negotiations with everyone. However, we have not had any cases at all where the customer has paid a penalty. I haven't heard much talk about customers offering to pay a penalty to cancel an LTA. Is this actually happening?
Downstream device makers has commented that it is reviewing its LTAs for materials. There was no indication that they were talking about silicon wafers. But given the scale of the impact, I suspected it was silicon wafers. But you are saying that you haven't heard any talk about this happening.
We have not received any penalty payments at all.
Next is Mr. Yoshida of CLSA Securities.
Earlier, you indicated that you plan to update and modernize existing plants for 300-millimeter as well. Can we look forward to more efficient CapEx through the migration from mature to leading-edge nodes rather than simple capacity expansion going forward? In 2023 and 2024, the investments made by Sumco were around JPY 300 billion and JPY 200 billion, respectively. What should we expect in terms of the level of CapEx investments for this year and next?
We would obviously go under if we undertook capital expenditures of JPY 300 billion or JPY 400 billion every year. So CapEx won't be at these levels. Recently, it has become very expensive to build physical structures and utilities. In hindsight, I am glad that we moved quickly. Going forward, investments will consist of replacing facilities and equipment. Generally speaking, the scale of equipment investment tends to be 1 decimal point lower. If we build a single physical structure with utilities, it takes JPY 100 billion to JPY 200 billion, which significantly increases the CapEx amount. Investment levels going forward, given that we expect to be cash flow positive next year, are likely to be at a level that can be covered by free cash flow, in my view.
I think the total investment scale for 2018 to 2020 was around JPY 50 billion or JPY 60 billion. Is that a useful benchmark to use?
Yes.
I see. So you expect to be free cash flow positive next year?
Yes.
Next is Mr. Nakahara of Tokyo Tokai Intelligence Lab.
If we look at the 300-millimeter wafer market, in volume terms, there has been a solid increase from 6.4 million wafers per month in first quarter to 7.6 million wafers per month in third quarter and fourth quarter. However, Sumco's sales have not increased proportionately. Can you explain why by talking about what happened with unit prices and volumes?
I can't discuss it in too much detail since I would need to talk about individual customers. But the reason why we look like we lost ground on sales in volume terms is because 2 of our customers where we are the top supplier, were struggling. One of the two, in particular, was very challenged. This had a significant impact on Sumco. We are also the top supplier to a customer that is extremely busy, but we would certainly hope to see our ailing customer get better soon. At the moment, things are very tough for this customer. This had a significant impact on Sumco.
I think the price points for the customer that is extremely busy are very high?
Not extremely high, but relatively high.
I had thought that you would be able to offset the impact of lower volumes at the ailing customer with sales to the very busy customer. But are you saying the price points for that customer are not that high?
So the proportion of leading-edge wafers we sell to the customer that is very busy is rising rapidly. But the volume we are shipping is not enough to offset the declines at the weak customer. That's the situation. So for instance, say, the busy customer buys 100 wafers, the proportion of leading-edge wafers has increased from 20% to 50%, and the absolute number of wafers is also rising. That is not enough to offset the impact of a customer where volume has gone from 50 to 10. In particular, with epi wafers, we were a top supplier, but this customer is struggling. We are doing our best to support them on the technology side. We hope they will do better, but it is taking time.
One related follow-up. In fiscal 2024, customers' reviews of LTAs led to a pushout in volume, which resulted in deliveries at price levels of previous years, capping price improvement. In 2025, I think the impact from this should drop out, suggesting that prices should rise. How are you thinking about this?
In 2025, we will return to normal conditions for leading-edge. It will be 100% in line with their commitment. With regard to delayed deliveries and the impact on price points, the pushouts are ongoing, so it will take a while longer before the impact drops out.
Next is Mr. Yamada of Mizuho Securities.
With regard to your cost for first quarter, costs have been highlighted as a Q-on-Q negative for profits. Can you talk about the background? Also on sales-related variance, there is a negative impact from the weakness in small diameters, but given that sales are being inflated by ForEx, are volumes for smaller diameters down? Also on 300-millimeter, should we feel comfortable that while volumes are likely to be flattish, the proportion of leading-edge will rise. If this is the case, then when I look at the EBITDA margin on Page 29, while small-diameter wafer volumes are falling in first quarter, I don't understand why the EBITDA margin would fall? Please explain.
First, with regard to increased costs, while pressure to revise costs from suppliers has eased, it hasn't gone away. Also, our contracts reset with the start of a new year, so there is some impact from this. Also, and this relates to the second part of your question, maintenance costs include regular maintenance at each plant. These expenses are relatively sizable, so the occurrence of regular maintenance is included in the increased cost.
On the second part of the question, I did say that small diameters are expected to be weak, but there is another factor, which is the number of operational days during the March quarter. February only has 28 days, so the number of operational days is shorter. If we compare first quarter to fourth quarter, we do have plans for a series of relatively large-scale regular maintenance work. In addition, this is the start of a new year. So on an overall basis, on top of small diameters being weak, the lower number of operational days means the production volume will be lower. Given this, while we can make sales from inventory, the Q-on-Q change in production levels will have a negative impact.
The framework for 300-millimeter and 200-millimeter you suggest is correct.
In other words, the EBITDA margin is under pressure in the short term because of regular maintenance and other factors. But if we look at just 300-millimeter, the EBITDA margins remain at the level previously stated by Chairman Hashimoto, where Sumco can stay above 30% even in poor conditions. So there is no need to worry. Is that correct?
You are correct on 300-millimeter. 200-millimeter, unfortunately, is weak.
Understood. I have high expectations for growth of and an increase in the proportion of leading-edge.
Thank you, Mr. Yamada. We will end the meeting here. Thank you to everyone for joining the Q4 fiscal 2024 results briefing. We are grateful for your participation today.
[Statements in English on this transcript were spoken by an interpreter present on the live call.]