Earnings Call: Q2 2026

Aug 6, 2026

Summary

Q2 sales and EBITDA margin exceeded forecasts, but profits remained negative, with strong AI-driven demand for 300mm wafers pushing shipments to record highs. Full capacity utilization is expected in Q4, and future investments hinge on significant price increases to offset rising costs.

Thank you for your participation today. This is the results briefing for the second quarter of the fiscal year ending December 2026. Before starting the presentation, allow me to confirm today's materials, which consist of three items: the consolidated financial results for the six months ended June 30, 2026, the presentation deck entitled Results for Q2 Fiscal 2026, and the announcement regarding interim dividend. 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 and President Jiro Ryuta, and Representative Director and Vice President CFO Shinichi Kubozoe. President Ryuta 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 President Ryuta. I am President Ryuta. I will start with an overview of the results. In Q2 2026, SUMCO exceeded its forecast for sales of JPY 112 billion by JPY 1.6 billion, reporting JPY 113.6 billion. In line with this, we also beat our forecast for OP, ordinary profit, and net profit attributable to owners of the parent. The EBITDA margin came in at 26.9%, overshooting our Q2 forecast of 26% by 0.9 percentage points. Forex for the quarter was in line with our assumption of JPY 160 to the dollar at JPY 159.9 to the dollar. For Q3, we project Q3 sales of JPY 118 billion, a Q-on-Q improvement of JPY 4.4 billion versus Q2 sales of JPY 113.6 billion. For each of operating profit, ordinary profit, and net profit, we similarly forecast Q-on-Q gains of JPY 1.1 billion, JPY 5.2 billion, and JPY 4.4 billion, respectively. We forecast our EBITDA margin to be 28.6%, up 1.7 percentage points from the 26.9% of Q2. Our Forex assumption for Q3 is largely unchanged at JPY 160 to the dollar versus the actual for Q2. Next page, please. This slide shows shareholder returns. We have set the interim dividend at JPY 10. The fiscal year-end dividend is to be decided at this time. With regard to dividends per share, we take a comprehensive view, factoring in the level of expected profits for the fiscal year, the outlook for the next fiscal year and beyond, cash needs, such as for CapEx, free cash flow, EBITDA, and the level of funds available for dividends in determining our guidance. Next page, please. This shows the trends for wafer shipments. The current fiscal year is shown in the solid red dots. Already in Q1, 300 millimeter was at a high level, just slightly below 8 million wafers per month. We were monitoring Q2 levels closely, but shipments exceeded the 8 million wafers per month, hitting a record high. Current demand growth suggests that there will be further sequential growth in Q3 and Q4. 200 millimeter is significantly lower than it has been in the past. However, wafer shipments in Q2 rose slightly from the Q1 level of just below 4 million wafers per month. Conditions within 200 millimeter vary by type, but overall, we are seeing a gradual recovery. The main driver is AI demand, which is boosting demand for devices such as power MOSFETs and others. We will continue to monitor the market, but I believe that we will see further improvements. Next page, please. This is the estimated customer wafer inventory for 300 millimeter wafers. Recently, in Q1 and Q2 of this year, inventory months have been falling. Customers had been adjusting inventory levels, but we will need to monitor the situation further to determine whether customer inventory adjustments will continue. Next page, please. If we split the trend for inventory into logic and memory, it provides a clearer picture. Logic inventory levels have been falling since the second half of last year. Wafer inputs had continued to exceed wafer purchase volumes in Q1 and Q2 of this year. That said, if we look specifically at leading-edge wafers, I believe there are signs that customers now have more of a sense of urgency about tightening market conditions and potential shortages rather than continuing to adjust inventory. Similarly, with regard to memory, there was a sudden increase in demand for memory use wafers in the second half of last year, driven by demand for memory for AI data centers, sparking a recognition that there was a shortage of wafer supply relative to rising demand and pushing up prices for both DRAM and NAND. Customer inventories are still sizable, but there are some customers who are concerned about tightness in wafers, which might mean they may not be able to lock in sufficient wafer supply for next year and the year after. As a result, the market picture is mixed. There are customers that are adjusting inventory, but also those that are considering increasing inventory. Next page, please. To sum up what I have discussed so far, in Q2, 300-millimeter volumes increased on the back of strong demand for leading-edge logic and memory for AI. 200-millimeter wafers also reported increased shipment volumes on demand related to AI and data centers. LTA prices continued to be respected in Q2, but from the middle of Q2, there were emerging moves to revisit and negotiate spot prices by some customers. The price negotiations we observed were not for lower prices, but for higher prices or increased volumes. Our view for Q3 is as follows. For 300-millimeter volume, we expect strong demand for leading-edge logic and memory to continue, driven by AI. Customers have made progress on adjusting wafer inventories, putting the market solidly on track for a recovery. For 200-millimeter volumes, while the picture is mixed both by customer and product, overall, we expect to see a gradual recovery. In terms of prices, LTA prices should continue to be respected, and similar to 300 millimeter, there are some customers that are starting to revisit prices for spot. Looking out further, for the semiconductor market, we expect to see continued strong growth driven by AI. For the silicon wafer market, we expect a continuation of strong, increasing 300-millimeter wafer demand for AI and data center-related leading-edge logic and memory. For 200 millimeter and smaller diameters, we should see a gradual recovery in demand in line with customer progress on inventory adjustments. Next page, please. The topic I would like to highlight this time is the expanding role of CPUs driven by the ongoing evolution of AI. As you already know, we had been seeing dramatic growth in demand forecasts as of last year, driven by generative AI, mainly focused on HBMs and GPUs. More recently, the expectations for demand growth have stepped up to the next level. One reason for this is the factoring in of expectations for agentic AI, either as a parallel driver of similar levels of demand to generative AI or potentially as a driver of even more demand. As AI evolves and advances, the level of required chips and devices is expected to expand further. As a result of these advances, the volume of data to be processed and the complexity of processing is projected to increase, expanding the role of CPUs to control the process. Up to now, it has been GPUs and HBMs that have been associated with AI, with the assumption being that having GPUs and HBMs that are capable of high-speed data processing was sufficient. Now, on top of this, there is a need for DRAM and NAND flash, and memory density required for CPUs is also expected to grow. We also show physical AI on this slide. I hope to be able to talk about this at the next opportunity, but we are seeing rapid changes in the evolution of AI. One year ago, I think very few people would have anticipated that not only would we see the establishment of the AI era, but that the pace of evolution would come this far in this timeframe. On this slide, I cover the transition from generative AI to agentic AI. In terms of capabilities, as you know, generative AI is chat-based, like ChatGPT, responding to queries in natural language or text. It responds quickly to all queries and is being widely used now around the world. In contrast, agentic AI responds to a specific objective and timeframe that is set by the user, autonomously developing and executing a plan. As it executes, it also constantly reevaluates and adjusts the plan as it works towards its objective. As an example, if you ask agentic AI to come up with a plan for a trip to L.A. during a specific timeframe, including identifying the optimal flights and hotel, it will not only provide you with that information, but also make specific arrangements for the trip, including booking the flights and hotel. Furthermore, it will provide updated information if it is able to subsequently get a better room rate at the hotel that is originally booked, or information about competing hotel rooms and services that can be had for the same price. If you choose to make a change, it will rebook in line with your preferences. Agentic AI offers a higher degree of convenience for the user. Similar to DX, it does not require the intervention of a human being, but instead opens the door to a world where the tool is capable of operating autonomously with AI managing the AI process. This has the potential to expand the range of applications for AI exponentially. In terms of application, generative AI can support certain tasks such as document creation and summarization. Generative AI, as you know, responds very rapidly to queries, pulling together an answer from a much wider range of information than is accessible to an individual. Agentic AI, on the other hand, has the ability to fully take on an entire task from the initial research and analysis, to the creation of a report, as well as taking on all of the logistics and other arrangements. In some instances, it can also handle the processing of expenses as well, making it highly convenient. If we look at the role of the CPU, generative AI uses the AI model or GPU for data input and output, primarily leveraging HBMs. The CPU for agentic AI not only does the data input for the AI model, but also actively manages plan tasks, collaborates with tools, and compiles and organizes information. As a part of the iterative process of data input and output, it accesses the AI model repeatedly, but also manages the process of referring back to the original plan to assess its progress. In order to retain the information on the original plan, it requires a large volume of NAND flash memory. It also requires significant volume of DRAM to organize information swiftly. As such, the advent of agentic AI will lead to an increase in data processing volume for the CPU, which is expected to drive a significant increase in required memory density. Semiconductor demand driven by AI will lead to not only growth for GPUs and HBMs, but for CPUs, DRAM, and NAND as well. It will also lead to further demand beyond this. Next slide, please. Looking at the change in the server demand forecast, the increased convenience of agentic AI is likely to drive an increase in users, pushing up the demand forecast for server volumes. Next slide, please. Agentic AI is likely to drive increases in not only server unit volumes, but changes in server memory demand as well, boosting demand for not only HBMs, but DRAM and NAND flash in large volumes too. The charts show projections in gigabytes. You can see that demand is expected to jump up solidly. Next slide, please. This is the demand outlook for silicon wafers for servers. On the back of an upwardly revised forecast for server demand volumes, demand for leading-edge logic and memory is expected to increase significantly as well. SUMCO is focused on further enhancing its ability to respond to rising demand for leading-edge wafers by upgrading its existing manufacturing facilities in response to continuing strong growth. We have been saying this since last year, but continue to view this as a priority. This completes my section of the presentation. I will hand over to CFO Shinichi Kubozoe to talk about details of our Q2 earnings. I, Shinichi Kubozoe, will present an overview of the results and forecasts. As touched upon at the outset, Q2 sales were JPY 113.6 billion. We posted losses at all levels with operating profit at -JPY 1.1 billion, ordinary profit at -JPY 4.2 billion, and profit attributable to owners of the parent at -JPY 4.4 billion. First-half sales is to the right with sales of JPY 215 billion. Operating profit was -JPY 6.3 billion, and profit attributable to owners of the parent was -JPY 12.8 billion. Lower down in the table, we show first-half CapEx on an acceptance basis of JPY 19.4 billion. This represents a level of less than half of the last year's first-half level of JPY 51.9 billion, reflecting the fact that greenfield plant investments have been largely completed. In contrast, reflecting the ramp-up of new facilities, total depreciation was JPY 64.4 billion, of which JPY 60.3 billion was above the line. Depreciation as a whole increased a significant JPY 15 billion. Other key financial metrics are as stated in the bottom half of the table. Next page, please. This is the analysis of change to operating profit. Starting on the left, in the analysis of sequential changes to quarterly operating profit, Q2 sales grew JPY 12.2 billion quarter-on-quarter, and the operating loss narrowed by JPY 4.1 billion from -JPY 5.2 billion in Q1 to -JPY 1.1 billion in Q2. The yen weakened by JPY 4.5 quarter-on-quarter from Q1 to Q2. As you can see from the chart below showing the components of OP change, costs increased only slightly, but depreciation rose JPY 2.9 billion quarter-on-quarter. But this was offset by strong sales growth for a solid positive impact from sales variance of JPY 6 billion. Sales increased by JPY 12.2 billion, of which JPY 1.6 billion is attributed to forex. After excluding the forex impact, sales variance accounted for JPY 6 billion of the roughly JPY 10 billion profit impact resulting from the increase in sales. Increased sales contributed meaningfully to sales variance. We show the year-on-year change for first half on the right. Sales rose JPY 9.7 billion year-on-year, while OP fell JPY 13.7 billion from JPY 7.4 billion to minus JPY 6.3 billion. The change in forex rate was JPY 8 year-on-year, with the dollar/yen rate for the period going from JPY 149.5 to JPY 157.7 to the dollar. The waterfall chart below shows an increase in cost of JPY 3.5 billion year-on-year. This reflects the impact of increased volumes, which pushed up labor cost, as well as an increase in maintenance costs related to the ramping up of the new plant. Energy costs were also up on the impact of net zero-related costs and higher unit prices. The increase in depreciation above the line was JPY 11.6 billion. For sales variance, in Q1, there was an impact from inventory adjustments in non-leading-edge epi wafers, which led to a change in product mix, hence the negative sales variance despite the increase in sales. Forex impact was a positive JPY 5.1 billion year-on-year. On a year-on-year basis, the increase in depreciation had a significant impact in depressing first half OP. Next page, please. On this slide, I will cover the balance sheet and cash flow. Looking at the middle of the balance sheet on the left, total assets as of the end of June were largely unchanged compared to the end of December 2025. In terms of major changes, cash and deposits increased by JPY 36.5 billion to JPY 111.7 billion as of the end of June. I will discuss the change in cash and deposits in covering cash flow on the right in a moment. There was a significant change in tangible and intangible assets further down on the page, which fell JPY 40 billion as of the end of June from JPY 670.6 billion to JPY 630.6 billion, with depreciation significantly outweighing CapEx acceptance in first half. Interest-bearing debt increased JPY 13.6 billion. Given that interest rates are expected to rise in Japan going forward, we chose to front-load the refinancing of some of our cash needs for this year in March, pushing up outstanding borrowings for the end of Q2. There will be repayments going forward, which should lower the level of interest-bearing debt. Overall, interest-bearing debt should decline toward the level as of the end of last year or potentially lower over the course of the remainder of the year. Under net assets, I highlight retained earnings. As a result of the net loss, there was a slight decline in retained earnings, as shown here. The equity to asset ratio was 50%, and the D/E ratio on a gross basis was 0.65 times as of the end of June. This represents a slight deterioration from levels as of the end of December 2025. On the right, we show cash flow. Operating cash flow was a positive JPY 46.5 billion. Cash flow for investment activities was an outflow of JPY 25.5 billion. As a result, free cash flow was a positive JPY 21 billion as of the end of June. Since Q3 of the previous fiscal year, we have been able to generate solid sequential improvements in achieving positive free cash flow. In terms of the use of cash, there was an outflow of cash for dividends paid. Reflecting the front-loading of financing, debt also increased for a net increase in cash and deposits of JPY 36.5 billion. We show our Q3 earnings forecast here. As touched upon earlier, we project sales of JPY 118 billion and break even for operating income. We project ordinary income of JPY 1 billion, but expect to break even for net loss attributable to owners of the parent. Q3 depreciation, shown further down the table, is projected to rise to JPY 35.4 billion. Nine-month cumulative depreciation is forecast to be around JPY 100 billion. We anticipate that quarterly depreciation in Q4 is likely to increase slightly Q-on-Q as well. As we have stated thus far, we expect 2026 to be the peak for depreciation. The EBITDA margin, as mentioned earlier, is projected to gradually improve, rising to 28.6% for Q3. I will highlight one point. Non-operating income, which is the third line down on the table, is projected to be JPY 1 billion. This reflects the expected impact of the receipt of subsidies from the national government. Next slide, please. On this next slide, we show the analysis of change in operating income. On the left, we show the sequential changes. Q3 sales are projected to rise JPY 4.4 billion Q-on-Q. Operating profit is projected to improve by JPY 1.1 billion. If you look at the waterfall chart below, you can see that we expect depreciation to increase by JPY 2.1 billion Q-on-Q. Costs are expected to rise JPY 3.9 billion Q-on-Q. This includes the expected seasonal increase in electricity costs reflecting higher summer rates. The increase in production volume and higher utilization rate is also expected to push up labor costs, such as overtime. There is also some expected impact from the conflict in the Middle East, which should boost electricity unit prices, as well as a slight impact on prices of some materials. With regard to sales variance, the contribution of JPY 7.1 billion outweighs the expected increase in sales of JPY 4.4 billion. There will be regular maintenance at our mainstay plant in fourth quarter, so we have chosen to build some inventory in advance, which will boost profits related to manufacturing. All of these factors combined are expected to lead to a JPY 1.1 billion Q-on-Q improvement in OP. On the right, the year-on-year changes for the first 9 months are as shown here. Trends here are similar to those for the first half year-on-year change. On the back of increases in volume, we expect a gradual shift into positive territory for sales variance. This completes my section of the presentation. Thank you. Mr. Komori. Thank you. We will now open the floor to questions. We will start with Mr. Umebayashi. Mr. Umebayashi. I am Umebayashi of Daiwa Securities. Thank you. I would like to ask about the waterfall chart for the sequential change in operating profit from Q2 to Q3 in the final slide presented, where you indicated you expect a positive contribution from sales variance of JPY 7.1 billion in excess of the sequential improvement at the top line. First, I estimate that the split of sales variance between the profit contribution from sales during the quarter to be slightly more than JPY 2 billion, versus a profit contribution from inventory build of slightly more than JPY 5 billion. Is that roughly correct? I estimate that the JPY 5 billion profit contribution from inventory build would be equivalent to around JPY 10 billion in sales. Does this mean that your utilization rate will be close to 100% at the end of Q3 as you head into Q4? Also, do you have room to further increase production beyond this after production rises to these levels? The reason why I ask is because your competitor indicated that their production volumes in June rose significantly and that they believe that the rest of the industry is operating at full capacity as well. President Ryuta. I will respond on production volume. With regard to production for 300 millimeter, we are very close to full capacity utilization. Furthermore, every month there are customers that are asking for more volume. We will truly be at full capacity utilization in Q4 in my view. Mr. Umebayashi. Understood. One quick question. Are you saying that you are already at full capacity and you expect to be at full capacity in Q4 after factoring in the impact of regular maintenance? Is the image that you will be running at full capacity into subsequent quarter as well? President Ryuta. The timing of the ramp-up capacity at our new plant is very fortuitous. Relative to our target of getting to full capacity for this plant by the end of the year, the ramp-up is going well. For this reason, we do expect to be able to remain at full capacity going forward. Mr. Umebayashi. Understood. Thank you. Mr. Komori. Thank you. Next is Mr. Miyamoto. Mr. Miyamoto. I am Miyamoto of SMBC Nikko Securities. Thank you for the presentation. I would like to ask about negotiations for LTAs and how you are thinking about the next round of capacity investments. Last time, you indicated that you had yet to start negotiations for the next round of LTAs. Recently, however, my impression is, based on comments from your peer, that LTA negotiations are now starting to pick up. Where does SUMCO stand today? Also, I believe you will need to raise prices in order to undertake the next round of investments. In terms of the magnitude of price hike you would require, would you be looking for the kind of increase as seen in the last cycle in 2022 and 2023, or more like the larger price increases seen in 2017 through 2019? Please comment on how you are thinking about the next round of investments to expand capacity. President Ryuta: In terms of LTAs, they are rolling off over time. There are some LTAs that will run until 2028. If your question is, will we wait until the current LTAs expire to negotiate or will we begin negotiations earlier? At this point, SUMCO is not considering accelerating the start of negotiations. We are getting requests from customers to increase shipment volumes, but in terms of responding to such requests, if we were to put on LTAs, it would mean that we are committing to guaranteeing specific volumes. To live up to such a commitment, we would require capacity, especially for leading-edge wafers. Installing new equipment for leading-edge wafers or investing to expand overall capacity would require significant CapEx. In an environment where semiconductors are booming, prices for everything in the value chain are rising across the board, including materials, gases and chemicals, electricity, transportation, and most significantly, production equipment. If we build a new plant, we would incur construction costs as well. We will need to take all of the above in consideration in discussing whether to invest in expanding capacity. Currently, we are upgrading our existing facilities, and any discussions we have with customers at this point are based on the capacity that we believe is possible as a result of such upgrades. There are some customers who say that this is completely insufficient, but we have been through cycles in the past where our customers have said similar things, but demand collapsed after only 3 years. We would need to be confident that we would be supporting our customers by supplying leading-edge wafers over the long term in investing to expand capacity. In the absence of this, LTAs would be difficult, as is responding to requests for more volume at current prices. Does this answer your question? Mr. Miyamoto: Understood. Thank you. Mr. Komori: Thank you. Next is Mr. Enomoto. Mr. Enomoto: I am Enomoto of BofA Securities. My question is on how to view the Q3 forecast that you show on page 22. Do the figures still include FST as a consolidated entity? Also, does the sales variance contribution include the impact of the price reviews you mentioned in the context of spot prices? CFO Kubo Zoe: I, Kubo Zoe, will respond. For the purposes of the Q3 forecast, FST is still being consolidated. It is not clear yet when it will transition to becoming an equity method affiliate, as we are gauging the timing of share sales when possible. On spot prices, there are some customers where we are reaching agreements on higher prices, so there is some impact on sales variance projections, but it is not huge in value terms. Mr. Enomoto: So basically, the biggest contributor to the JPY 7.1 billion is production-related gains. Is that correct? CFO Kubo Zoe: The split between the contribution from sales in the current quarter versus inventory build is not JPY 2 billion to JPY 5 billion as suggested earlier. The contribution from sales in the current quarter may be slightly larger, but in any case, gains related to production is the big contributor to sales variance. Mr. Enomoto: Understood. Thank you. Mr. Komori: Next is Mr. Ikeda. Mr. Ikeda: I am Ikeda of Goldman Sachs. In looking at the silicon wafer projections for servers you have outlined for 2027 and 2028 on page 15, the numbers appear very conservative if you take into account the expected increase in chips per end product or CBA. Have you baked in some supply risk into your forecasts? For 300 millimeter, I expect 2027 to be very tight and expect to see shortages in 2028. Despite this backdrop, I understand that you are hesitant to commit to greenfield investments, as you alluded to earlier. If you use a hurdle rate that is close to your weighted average cost of capital, you would need to see a doubling of wafer prices to enable new investments, or the economics would not work at all. It means that prices need to be significantly different from current levels. With the profitability improving significantly at your customers and structural demand growing, what sort of negotiations are you currently conducting? Are there measures you can put into place, such as requiring upfront payments or extending the tenure of the LTAs to ensure that you can sustain profit growth over the longer term? Please comment to the extent you can, including lesson learned in hindsight. President Ryuta: There are many things that I regret in hindsight in many different fields, which makes it difficult to talk about in short. It is true that the forecast for server use silicon wafer demand takes into account the fact that the supply side is not in a position to increase supply at this time. Actual demand levels are higher, but at this stage, it is not possible to grow supply. Demand is likely to exceed supply for more than three years, in my view, because of such supply constraints. In terms of the price level that we would need to increase supply capacity and invest to expand supply, we have made no specific statement at this stage, but what I can say is my personal sense is that prices would need to be dramatically higher. To your point, I think an increase of 5% or 10% would not have a meaningful impact. That's my impression. Mr. Ikeda: Do you think that incorporating tougher clauses in LTAs would help SUMCO achieve more stable earnings at high levels over the longer term? Could fixed volume contracts work? How would you negotiate such a contract? To the extent that you can, please comment or talk about your aspirations. President Ryuta: In the past, SUMCO has tried many different contracts, which I have seen in a previous role when I was responsible for international sales. If you look at past silicon cycles, when prices collapse, customers can't buy, and they can't take delivery of shipments. So even if you try to lock customers in with tighter contracts, what ends up happening is similar to the inventory adjustments we have seen recently. In other words, customers are buying wafers that they are not going to use and which pushes up their inventory. This then leads to the inventory emerging somewhere else, driving down prices. Things you suggested like down payments or very long-term contracts where the prices are tied to a benchmark are possible choices, but for now, we are able to improve mix within our existing capacity by upgrading existing facilities. Our current stance is to continue to upgrade our facilities while monitoring the market and peers closely. Mr. Ikeda. Understood. I have high expectations for SUMCO. Thank you. Mr. Komori. Thank you. Next is Mr. Watabe. Mr. Watabe. I am Watabe of Morgan Stanley. With regard to the spot market, you said you are renegotiating prices. Please talk about how much spot prices had fallen from last year into this year, and how much of a recovery you have seen so far in percentage terms. Also, what percent of your business is spot? CFO Shinichi Kubozoe. SUMCO has a very high proportion of LTAs. Spot is the remaining small portion, but the majority is FST. I can't discuss the change in spot prices in percentage terms, but directionally for some specific products where demand was low, including smaller diameters, spot prices have been falling, but recently we have seen a reversal in the price declines in these areas. On top of this, in market segments that are tight, we have been able to renegotiate prices with some customers and are in the midst of discussion with others, but can't talk about the numbers. Mr. Watabe. On page 18, you show year-on-year sales variance of minus JPY 3.7 billion for the first half. Is the impact of weaker spot prices a significant contributor to this sales variance? CFO Shinichi Kubozoe. It does include a negative contribution. Mr. Watabe. But it's not big? CFO Shinichi Kubozoe. You're talking about what we show on the right. There was an inventory adjustment in Q1 related to logic, which impacted mix. Also, it was early days in the ramp-up of the greenfield plant, so there were areas where fixed costs were relatively high. These elements were relatively sizable, so if your question is: "Was there a major negative impact from price on this?" Then that is not the case. Mr. Watabe. Understood. Thank you. Mr. Komori. Thank you. Next is Mr. Omura. Mr. Omura. I am Omura of UBS Securities. I would like to ask about CapEx in the future. You were quoted in a publication as saying that in this industry, you don't have visibility even for 3 months out. How are you now thinking about the Yoshinogari project, which was postponed in March? Hypothetically, if you were to restart the project, how would the subsidies work? President Ryuta. I will respond with the exception of the point about subsidies. At this time, we have no plans whatsoever to restart this project. At current price levels, it is not viable. All costs are rising now, and with Japan potentially headed for more inflation, costs are likely to rise further. In considering this project in comparison to others, we must take both of these points into account. We must be able to demonstrate that the project can generate an appropriate return over an appropriate timeframe. Additionally, it is not enough to show an appropriate return at the time of the investment decision. We must be confident that the project will continue to generate appropriate returns over time based on appropriate prices. We have not fully completed the land work at Yoshinogari yet, but I am informed that it will be completed next spring. CFO Shinichi Kubozoe. With regard to subsidies, initially, we received subsidies from the national government for the purpose of expanding capacity at Yoshinogari. But in March of this year, we submitted an application to change the target of the subsidies, allocating them instead toward the upgrading of existing facilities. This has been approved. As a result, the subsidies are no longer specifically for Yoshinogari, but instead have been improved for upgrades to existing facilities, renovations, and modernization. As mentioned earlier, we have no plans for new development at Yoshinogari, but if, hypothetically, we were to restart this project, the potential for subsidies would be subject to the programs available at that time. At this point in time, therefore, no plans on the subsidies either. Mr. Omura. Understood. Thank you. Mr. Komori. Thank you. Next is Mr. Nishiyama. Mr. Nishiyama. I am Nishiyama of Citigroup Securities. With regard to the next round of LTAs, the magnitude of the required price increase will depend on the magnitude of the CapEx required. For instance, the greenfield investment you announced in 2021 was JPY 230 billion. I calculate that total CapEx for the 5-year period between 2021 and 2025 was around JPY 800 billion plus, including maintenance CapEx. Going forward, you have suggested that CapEx will be around JPY 50 billion to JPY 60 billion a year for the next few years. But could this change based on recent communication with customers, including their forecasts? How much of future demand can be covered by upgrades to existing facilities? President Vyuta. Of course, it does hinge on where wafer prices are, but at this stage, we don't know how much CapEx would be required. 2 years ago, who would have predicted the inflationary environment that Japan is currently experiencing? Equally, it is difficult to predict how high inflation could get in 2 years' time. What we can say is that when it truly becomes necessary to invest to support our customers, it will need to be at an appropriate wafer price level. It might be possible to do LTAs for CapEx, but if inflation continues to have an impact, it becomes a moving target. We would therefore need to seriously evaluate LTAs before signing to ensure that they don't come back to haunt us later. I'm sorry, but just like we don't have visibility for even 3 months from now, it is tough to read the situation for next year and beyond at this point. We will continue to monitor the situation, including our peers, to make necessary changes as appropriate. This applies to not only investments, but many other areas where we must make changes. CFO Shinichi Kubozoe. You also asked about how much we can do based on annual CapEx of JPY 50 billion to JPY 60 billion. I can't specify a timeframe, but we should be able to support our customers to a certain extent by upgrading existing facilities. If customer demand exceeds this, then I think we would need to have wafers priced at a level that appropriately covers rising costs. To the extent that a gap develops between our capacity and rising customer demand, we recognize we must engage with our customers as necessary to achieve appropriate wafer prices. This engagement with customers is something we recognize will be an ongoing initiative. Mr. Nishiyama. Understood. Thank you. Mr. Komori. Thank you. Next is Mr. Nishihira. Mr. Nishihira. I am Nishihira of Okasan Securities. I was looking at slide 8. What do you consider to be a healthy level of customer inventory for your customers? President Vyuta. We don't really have a good sense for what is a healthy level of inventory for our customers. That is really something that is under the control of our customers. I think it varies from customer to customer. If the customer feels that it may become difficult to procure wafers in the future, they may choose to carry a higher level of wafer inventory. Equally, a customer could feel that sub 2 months of inventory is sufficient for their needs. Our role is to ensure that, to the extent possible, our manufacturing capability is well-positioned to meet customer needs as appropriate. Mr. Nishihira. Customer inventory peaked in Q4 of the previous fiscal year and has since been declining. In the near future, would you expect customer inventories to get down to the year 2000 level, or do you think that inventories settle for a time at a level above this? Do you have a sense of direction for customer inventories? President Vyuta. My personal expectation is that customer inventories remain slightly elevated over the next 3-5 years. There has been a lot of talk about shortages in memory. Memory makers are the biggest consumer of wafers on a volume basis, so my personal assumption is that memory makers will be scrambling and could potentially choose to maintain elevated levels of wafers. Mr. Nishihira. Understood. So directionally, rather than inventory levels dropping back to the year 2000 levels, customers are likely to carry slightly higher levels of inventory. Is that correct? President Vyuta. For customer procurement departments in particular, running out of wafers can lead to serious issues of accountability. Based on past experience as well, when the market tightens, customers tend to carry elevated levels of wafer inventory. Mr. Nishihira. Understood. Thank you. Mr. Komori. Thank you. Next is Mr. Yamada. Mr. Yamada. I am Yamada of Mizuho Securities. I would like to ask about inventory and depreciation. The third quarter guidance projects some depreciation that is below the line, with total depreciation below the line of JPY 5.8 billion for 9-month cumulative. Am I correct to assume that if you are operating at full capacity utilization, this depreciation moves above the line once we get into the next fiscal year? Also, if we assume that annual depreciation for the full year of more than JPY 140 billion, as previously suggested, it implies Q4 depreciation of more than JPY 40 billion. If we extrapolate expected depreciation above the line for 2027, it suggests it should be around JPY 120 billion or JPY 130 billion on a full year basis. Is this in the ballpark as an image? Also, although depreciation is a non-cash item, in considering the next round of investments, would your wafer price assumption be based solely on the cash outflow for CapEx, or would you want to also factor in anticipated depreciation when determining the appropriate wafer price level? Or is nothing decided at this point? How are you thinking about this? President Ryuta. At a high level, unless prices rise to a level that can justify the investments, we cannot invest. This is true for the industry generally as well. At current prices and the current pace of cost increases, it is not economically viable to expand capacity. For us, costs are rising, and it's not just depreciation, but many other costs such as materials and others. There is a need for prices to rise to a level that covers total costs. What was your other question? Mr. Yamada. I also wanted to know about depreciation, given that the construction in progress stands at JPY 70 billion. If Q4 2026 depreciation is really JPY 40 billion-plus, I project next year's depreciation to be at around JPY 120 billion to JPY 130 billion. Also, will all of this be above the line? CFO Shinichi Kubozoe. To your point about depreciation below the line, you are correct that since utilization levels are rising, this will shift above the line over time. As such, depreciation below the line should decline sequentially. With regard to full year depreciation for this fiscal year, we have disclosed the quarterly figures and forecasts for each of the quarters up to Q3. If you look at the pace of sequential increase, that should give you a sense for what Q4 is likely to look like. I leave it to you to do the math. In terms of the proportion of depreciation that will be above the line, it will generally be in line with the levels you suggested earlier. In terms of the timing of payments or locking in funds, fundamentally, we haven't made any decisions on new investments at all. But as a general statement, in terms of timing of payments, CapEx acceptance tends to be a little further out, so there is a need to have the cash on hand for payment at around that time. So timing of payments is likely to come sometime between the initial investment decision and the start of depreciation, and we would typically track CapEx acceptance in terms of securing cash for payments as a general rule of thumb. Mr. Yamada. If so, depreciation is growing at a pace of around JPY 2 billion each quarter, which means that the balance under construction in progress is likely to remain high. Am I correct in assuming that this is the right image for the ramping up of facilities? CFO Shinichi Kubozoe. Basically, there is a certain proportion of depreciation tied to non-greenfield investments, which will mean that construction in progress will not go to zero. In terms of the timing of ramping up new facilities, we are still in the midst of ramping up greenfield investments, but this will progress in line with plan. It is this piece that will gradually decline over time in tandem with an increase in depreciation. Mr. Yamada. Understood. Thank you. CFO Shinichi Kubozoe. One more point to note is that with the start of the next fiscal year in January, there will be a resetting of depreciation, so you will need to net that out against new depreciation that kicks in. Mr. Yamada. Understood. Thank you. Mr. Kobori. We will end the meeting here. Thank you to everyone for joining the Q2 fiscal 2026 results briefing. We are grateful for your participation today.