SUMCO projects flat 2024 amid high inventory, recovery hopes By biedexmarkets.com

SUMCO Corporation (ticker: 3436), a leading supplier of silicon wafers, held its fourth-quarter earnings call for the fiscal year ending December 2023, revealing mixed results and cautious optimism for the future. Chairman Hashimoto announced that earnings for the fourth quarter exceeded forecasts, but a decline in sales is expected for the first quarter of 2024. The company also raised its fiscal year-end dividend per share to JPY 13. Despite a challenging market, with high customer inventories and declining memory prices, SUMCO anticipates a flat shipment forecast for 2024 with a potential increase in the second half of the year, especially in the sectors of AI and smartphones.

Key Takeaways

• Fourth-quarter earnings surpassed expectations, but sales are projected to drop in Q1 2024.
• Fiscal year-end dividend per share increased to JPY 13.
• High customer and wafer inventory levels, especially for logic-use epi wafers.
• Market conditions for silicon wafers are in transition, with a tough first half expected in 2024.
• Prices for 2024 are projected to remain flat with a slight rise in the second half.
• SUMCO is hopeful for a recovery in epi wafer demand, driven by AI applications in the latter half of the year.

Company Outlook

• SUMCO predicts a significant sequential drop in sales and operating income for the first quarter of 2024.
• The industry’s 300-millimeter wafer volume may fall below 6 million per month due to inventory adjustments.
• Prices for delayed shipments in 2024 will be maintained at the original rate to uphold customer relationships.
• A sharp improvement in shipments is expected in the second half of the year, particularly for AI data centers and smartphones.

Bearish Highlights

• Current high inventory levels for logic and memory wafers may hinder short-term sales.
• Memory prices are affected by a decline in input volumes, with a slow recovery anticipated.
• Free cash flow was negative, resulting in new borrowings and reduced cash reserves.

Bullish Highlights

• Cost reductions have led to better-than-expected earnings in the fourth quarter.
• There are recent signs of a possible rebound in demand for logic-use epi wafers.
• The company expects an earlier recovery in epi and DRAM, driven by generative AI.
• More advanced wafers are fetching higher prices due to their increased technological complexity.

Misses

• The fiscal year 2023 showed a decline in sales and operating income.
• The forecast for the first quarter of 2024 indicates a significant drop in sales and operating income.

Q&A Highlights

• The company maintains an understanding with customers regarding pricing in case of delivery delays or early shipments.
• SUMCO is hopeful for an early normalization of inventory levels, which would positively impact shipment volumes and prices.
• The pricing for wafers depends on the level of technology, with more advanced wafers priced higher.

In conclusion, while SUMCO faces immediate challenges with high inventory levels and a soft market, the company is positioning itself for a potential rebound in demand, particularly from the AI and smartphone sectors, in the latter half of 2024. The company’s commitment to maintaining customer relationships through stable pricing, despite delivery delays, reflects a strategic approach to navigating the current transitional market conditions.

Full transcript – None (SUMCF) Q4 2023:

Takayuki Komori: Thank you for your participation today. This is the results briefing for the Fourth Quarter of the fiscal year ending December 2023. Before starting the presentation, allow me to confirm today’s materials, which consists of three items: the brief statement on consolidated financial results for fiscal 2023, the announcement regarding revision to dividend forecast and the presentation deck entitled Results for Fiscal Year 2023. This will be a 60-minute briefing, which will end at 5 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, Michiharu Takii. Hiroshi Ito, Executive Officer and General Manager of Accounting is also on hand. Chairman and CEO Hashimoto will discuss our forecast and operating environment to be followed by an explanation of the financial results by CFO Takii. We have set aside time for a Q&A session as well. I will now hand over to Chairman Hashimoto.

Mayuki Hashimoto: I am Chairman Hashimoto. I will start with the overview on Slide 5 of the presentation. This is a summary of the results for the fourth quarter of the fiscal year 2023. Earnings were well ahead of our forecast. The main factors were a positive contribution of around JPY1 billion from the weaker yen, with the remaining roughly JPY3 billion coming from better-than-expected cost reductions. As a result, fourth quarter operating income overshot our forecast by JPY4.7 billion, and ordinary income exceeded our forecast by JPY3.7 billion. Turning to the earnings forecast for the first quarter of 2024. We project a significant sequential decline in sales. On the back of this, operating income is projected to be JPY4.5 billion; ordinary income, JPY4 billion; and net income, JPY2 billion. We assume a dollar-yen rate of JPY145 to the dollar. The main reason for the sequential declines is the expected drop in sales. I will go into more detail later, but our customers are currently carrying inventory levels well above normal and are at a point where they are no longer in a position to accept more wafers due to a lack of available storage base, hence, the adjustment in terms of shipment volumes. Please turn to Page 6. This slide shows shareholder returns or dividends per share. Reflecting the better-than-expected results, we have raised our guidance for our fiscal year-end dividend per share from JPY10 to JPY13 for a full year total of JPY55. This represents a dividend payout ratio of 31%. Note that we have excluded subsidies received of JPY1.7 billion after tax from the calculation of the dividend as it is not appropriate to use prefectural subsidies to pay dividends. Next page, please. Looking at fourth quarter market conditions for silicon wafers, both logic and memory customers persisted with production adjustments, leading to a continuation of low shipment levels for 300-millimeter wafers. Similarly, for 200-millimeter and smaller diameter wafers, which are used in a variety of applications, reflecting the overall weakness of market conditions, customers continue to adjust production, depressing shipment levels. That said, customers respected LTA price levels for both 300- and 200-millimeter wafers. So SUMCO was able to achieve steady price increases in line with LTA prices. On the outlook for 300-millimeter volumes in first quarter 2024, while there are some customers that have seen slight recoveries, there are others that have seen a further deterioration, so the picture is mixed. On an overall basis, conditions have improved, but we think that fourth quarter into first quarter is likely to be the bottom with a gradual recovery to follow, although the momentum of the recovery is unlikely to be strong, in my view. This is because to date, customers have complied with the terms of the LTAs, continuing to purchase wafers, which has resulted in significant customer inventory levels. As such, we have no choice but to cooperate with customer requests to delay deliveries. The total volumes under the LTAs remain unchanged, but customers are asking to slightly push out the timing of deliveries. We have agreed to such request to a certain extent. For 200 millimeter and smaller diameters, customers continue to adjust production levels so shipments are expected to decline further. Unlike 300 millimeter, signs of a recovery have yet to emerge. However, if smartphone handsets start to recover, given that 200-millimeter wafers are used in many smartphone applications, then we would expect to see a recovery in 200-millimeter wafers as well. That said, the recovery in smartphones is still weak. In terms of prices, LTA prices continue to be respected. But prices for spot, especially for smaller diameters in certain regions for certain applications, are softer for commodity product. The exception is more technologically difficult applications where prices remain relatively firm. Looking at the outlook, investments are likely to remain strong for data centers for generative AI and firm for EV and energy-related applications. On top of this, the bottoming out of smartphone and PC demand is contributing to signs of light at the end of the tunnel. However, the recovery in wafers typically lags a recovery in end products. In particular, this time around, customers are carrying significant levels of inventory, which suggests it may take time before we see a recovery in wafers. I believe it is likely to take until the second half of this year. That said, as I have been saying at each of our recent meetings, 300-millimeter epitaxial wafers are used in made-to-order products while 300-millimeter polished wafers are used for memory, which is typically produced in anticipation of demand and, therefore, has a tendency to overshoot. So it is necessary to take into consideration not only wafer inventory levels but end product inventory levels as well, which appear to be high. As such, a recovery in polished wafers is likely to take more time than epi in my view. Page 8, please. This slide shows the wafer trend for 200 millimeter by quarter. Unfortunately, volumes have fallen below 2016 levels. We believe a recovery will take time. Page 9, please. This is the trend for 300 millimeter. I talked earlier about conditions for 300 millimeter. The first half of 2024 will probably be the bottom for wafers, but I think we should subsequently see a gradual improvement. Obviously, there will be a slight difference in trend between epi and PW. I expect it will take longer before we see a recovery in PW, but I imagine we should see a relatively early recovery in epi. Page 10, please. I would now like to talk about the future for silicon wafers. I will explain the four themes shown here in order. Slide 11, please. I will start with AI servers and how they are structured. You may already be familiar with this, but an AI server consists of a GPU and HBMs, which are DRAMs. Fabricating DRAM HBMs is technologically challenging. There is only one player that can fabricate HBMs in large volumes. Although there are other players that can do small volumes, the top player produces significantly higher volumes. The chips are combined on a silicon interposer as well. As such, AI servers will consume a huge volume of silicon, driving a significant increase in the area of silicon required. A single AI server is estimated to consume 1.8 wafers. Compared to a conventional server, an AI server consumes 3.4x as much silicon. Recently, smartphones have begun to introduce generative AI. This, combined with the emergence of 3D architecture, suggest silicon consumption is likely to grow substantially in future. Slide 12, please. This chart shows the projected demand growth for server volumes. Unit volumes are expected to increase as shown here. As there was a significant decline last year, we expect to see a slight recovery this year. However, we believe a full-scale recovery will only come in 2025. We believe that the proportion of AI servers will increase on the back of the increase in overall volumes. We project a CAGR of 26% for AI servers compared to conventional server volume growth of 5%. Slide 13, please. This slide looks at how advances in AI will drive demand for leading-edge process wafers. AI servers, of course, use leading-edge logic at 3 or 2 nanometers, so growth in AI servers will therefore drive demand for leading-edge process wafers. Also, AI servers consume 3.4x the silicon in surface area terms compared to conventional servers. Given this, we expect AI servers will rise to account for half of the demand for wafers used in servers in the near future. Slide 14, please. Turning now to smartphones. We are seeing a sudden increase in the adoption of AI in smartphone handsets from this year. We think it is reasonable to expect that, over time, AI-enabled smartphones will come to account for half of total volume. Assuming this is the case, the number of parameters or calculation volume, which is currently 10 billion, should double by 2026 and continue to increase subsequently. Given the need to process huge volumes of data at high speeds, this should drive demand for leading-edge wafers used in CPUs, GPUs and DRAM, effectively substantially boosting wafer consumption. Slide 15, please. This shows customer wafer inventory levels for 300-millimeter wafers. The white bars represent purchase volumes while the blue bars represent wafer input volumes and the green bars are inventory. Current customer inventory levels are at record highs. Even now purchase volumes continue to exceed input volumes, so customers are now at a point where they say they cannot buy any more wafers. At the same time, wafer purchases cannot go to 0 either. So we are engaged with our customers in thinking about how to return to more normal levels of inventory. The impact of customer inventory adjustments is a major factor behind our forecast for a decline in first quarter revenue to JPY87 billion. It is not an indication of a deterioration in market conditions. Instead, the market is bottoming out in fourth quarter into first quarter. We do not expect to see a further deterioration of market conditions. Slide 16, please. This shows the breakout of customer inventories between logic and memory. You can see that wafer inventory levels are similarly high. Slide 17, please. This chart shows trends for logic-use epi wafers. The orange line is inventory. Inventory is at record high levels. However, the gray line, which is price, has been rising steadily in line with the LTAs. That said, demand has fallen slightly, although recently, there are signs of a possible rebound after adjustments in first or second quarter. Slide 18, please. This is the same chart for PW. Although there has been talk that memory prices have been rising, this reflects a decline in input volumes, which has led to a slight tightening of the market rather than prices rising on the back of increased demand. Given this, it is likely it will take some time for a recovery to emerge. Slide 19, please. So in terms of how we see the market over the longer-term, we expect the first half of this year to be very tough, although we expect a rebound in the second half of the year. On a full year blended basis, we would hope to match 2023 levels, although it might be slightly weaker on a year-on-year basis. We are currently in a transition period, which makes it tough to read. What we can say is that the market has bottomed. The strength of the recovery will hinge on the strength of generative AI-related demand as well as what happens with China, given the sheer scale of purchasing power. The strength of replacement demand for smartphones and other end products in China will also be an important factor. This completes my section of the presentation. I will hand over to CFO Takii to talk about details of our earnings.

Michiharu Takii: I, Takii, will present the earnings in more detail. Please turn to Slide 21. This slide shows the results for fiscal 2023. On a year-on-year basis, sales declined but consolidated sales, include sales of polysilicon from second quarter onward which implies the declines in wafer sales, was actually larger. Operating income also fell year-on-year to JPY73 billion. Ordinary income was JPY72.6 billion. However, we recognized an extraordinary gain in first quarter related to the acquisition of the polysilicon business, so the magnitude of the profit decline is smaller at the net profit level at JPY63.8 billion. CapEx was up substantially year-on-year on continued full-scale Green Field investments in both Japan and Taiwan on an acceptance basis. Similarly, depreciation also rose on the back of investments. The main increase was in property. As a result, EBITDA was down on a year-on-year basis, but we have been able to maintain an EBITDA margin in the 30% range. On a full year basis, ROE was 11.6%. Please turn to Slide 22. This is the analysis of changes in operating income. First, the sequential analysis. As noted at the outset, fourth quarter earnings were much better than expected. As shown in the analysis, costs improved by JPY4.3 billion Q-on-Q. Initially, we had expected to see an improvement of JPY1 billion in costs in fourth quarter related to lower electric power costs, but in addition, we were able to achieve JPY3 billion in cost reductions through internal efforts to reduce expenses, lower materials costs and maintenance expenses and other initiatives. Depreciation was in-line with plan at minus JPY2.5 billion. Sales-related variance was minus JPY7 billion. The reason for the decline in sales-related variance, despite sales remaining largely unchanged, is a reflection of production rather than sales. I had previously mentioned our plans to undergo regular maintenance at our mainstay crystal ingot plants in October and November as well as at a processing plant. This depressed capacity utilization significantly. In addition, related to the regular maintenance, we also chose to reduce work-in-progress inventory significantly in anticipation of a substantial drop in shipments in first quarter. The weaker yen generated a positive of JPY1.3 billion. Looking at the year-on-year change for the full year results, as shown on the right, operating income fell JPY36.6 billion. Costs increased JPY15.5 billion, which breaks down into a JPY9 billion negative for materials unit prices, a JPY7 billion negative from electric power costs and an increase in labor costs, particularly in first half when capacity utilization was high, of JPY2 billion. We were able to offset the higher cost with internal cost reduction efforts of JPY2 billion. Depreciation increased JPY11.9 billion. Sales-related variance was a negative of JPY22.8 billion, which breaks down into a JPY22 billion positive from higher prices related to the start of a new fiscal year, offset by a JPY45 billion negative as a result of lower volumes. The weaker yen had a positive impact of JPY13.6 billion year-on-year. Slide 23, please. Slide 23 shows the balance sheet. I will touch upon this later, but free cash flow was significantly negative in fiscal 2023. We tapped into cash and time deposits, reducing it by JPY100 billion. Raw materials and supplies rose, reflecting an increase in supplies as a part of initiatives to strengthen BCP. Tangible and intangible assets increased JPY250 billion. This is the impact of the Green Field CapEx of more than JPY300 billion on an acceptance basis discussed earlier less depreciation. Interest-bearing debt increased roughly JPY83 billion year-on-year. This breaks down to new borrowings at joint venture FST of JPY50 billion and at SUMCO in Japan of JPY30 billion. If we look at the capital account, retained earnings increased, reflecting the impact of profit growth. As shown at the bottom of the table, the equity ratio declined slightly, but equity per share increased. On the cash flow statement to the right, the sub total of pretax profits, extraordinary income and depreciation was JPY144 billion. However, there was an increase in inventory, and under Others, tax payments were relatively large. The resulting operating cash flow was slightly less than JPY100 billion. Capital expenditures were JPY310 billion, split between roughly JPY200 billion in Japan and JPY100 billion for FST in Taiwan. In response to the negative free cash flow, we undertook new borrowings and tapped into cash and time deposits, hence the decline in cash on hand. Slide 25, please. On Page 25, we show our forecast for first quarter. We expect a significant sequential drop in sales. Operating income is expected to be roughly half Q-on-Q. The figures are as shown on the table. Depreciation is projected to be JPY18 billion, down sequentially reflecting the start of a new fiscal year. As you can see from the various metrics, we are expecting a deterioration in profitability. Slide 26, please. Slide 26 shows the analysis of change in operating profit. First, the first quarter sequential change on the left. With regard to the Q-on-Q decline in operating income of JPY6.7 billion, we are expecting a slight increase in costs, reflecting ongoing increases in unit cost of materials. Depreciation, as noted earlier, is expected to decline Q-on-Q, reflecting the start of a new fiscal year. Sales-related variance reflects the significant expected drop in sales. This includes an expected drop in non-LTA spot prices of around JPY2 billion, primarily in smaller diameters. We also factor in a JPY6 billion decline to reflect lower volumes. On the back of lower sales, we project a deterioration in the profitability of sales. Our ForEx assumption is for a slightly stronger dollar yen at JPY145 to the dollar. Near-term, the yen is slightly weaker, but this is subject to change. Looking at the analysis of year-on-year change for first quarter, we expect significant declines in both sales and profits in first quarter. On costs, we expect to significantly reduce labor costs in first quarter for a year-on-year improvement. Electric power unit prices are also expected to improve, but this will be offset by a slight increase in materials cost. On a year-on-year basis, depreciation is expected to increase for a negative of JPY3.6 billion. Of the sales-related variance, as noted earlier, spot prices are falling, so JPY2 billion of the decline is related to price. The remaining roughly JPY20 billion negative reflects the overall large drop in sales volumes. We expect a positive impact from the weaker yen. Page 28, please. This shows EBITDA and sales. As you can see, we expect sales to fall sharply in first quarter. If we assume this will be the bottom, we expect to see a reversal of the trend in second quarter and third quarter of 2023. This is our assumption. This completes my section of the presentation.

Takayuki Komori: Thank you. We will now open the floor to questions.

A – Takayuki Komori: We will start with Mr. Enomoto of BofA Securities.

Takashi Enomoto: Before asking my question, I would like to take this opportunity to thank Mr. Takii for his service. Thank you so much. Moving on to my question, can you talk about how we should think about prices in 2024? Previously, you had indicated that you expect unit prices would rise in 2024. But based on the analysis of change in profits presented today, you have factored in a decline in spot prices and appear to be expecting LTA prices to be flat. Is this because of the pushouts in delivery timing for LTAs? So for previous deliveries that have been pushed into 2024, are you expecting prices will not rise? Please talk about how we should think about LTA pricing and why prices are not expected to rise.

Mayuki Hashimoto: At a high level, although there is variance in individual contracts, LTAs can be categorized into two groups. Those were the prices fixed over a contract period of, say, 5 years and those where there is a step-up feature where the prices rise every year, for example. For fixed price contracts, a delay in deliveries has no impact since the price is the same. For step-up contracts, however, if a delivery originally scheduled for 2024 were to be pushed into 2025, in an extreme case, all of the delivery will be made at a later date but at the 2024 price point. Obviously, that is an extreme example that is highly unlikely. However, the cumulative impact of gradual pushouts to date is having an impact on the blended price point. So prices, which should have risen this year, won’t rise much. For FST, the proportion of LTAs is not as high, so the decline in spot prices is having an impact. As a result, on a net basis, prices are not expected to rise.

Takashi Enomoto: With regard to previously pushed out deliveries, can we assume that you will be able to achieve the original price when the actual shipment is made? So for delayed shipments in 2024, when the shipment is actually completed, can we expect you will receive the higher 2024 price?

Mayuki Hashimoto: Yes, that’s correct.

Takashi Enomoto: Understood. I have high expectations for the completion of inventory adjustments. Thank you.

Mayuki Hashimoto: Me too.

Takayuki Komori: Thank you. Next is Mr. Watabe of Morgan Stanley.

Takato Watabe: Based on today’s discussion, if the rest of the industry tracks in line-with SUMCO’s first quarter sales forecast, it sounds as if industry shipment volume for 300 millimeter will likely drop below the level of 6 million wafers per month. Do you believe that this reflects the magnitude of the inventory adjustments? If you have an image of the magnitude of the inventory adjustments, either on a quarterly or monthly basis, could you share it with us?

Mayuki Hashimoto: SUMCO does business with almost all of the industry players, and I believe that the customers treat all of their suppliers equally and are not requesting shipment adjustments just from SUMCO. In such circumstances, customers are likely to be as equitable as possible. So I believe that the industry as a whole is in a similar position to us.

Takato Watabe: Is it difficult to talk about specific numbers at this time?

Mayuki Hashimoto: In terms of a specific image, there are some customers that expect a recovery from April, so it’s hard to say. One thing that I can say, however, is that first quarter will likely be the bottom in terms of wafer volumes. Obviously, this could change if there are new developments. Recently, there have been so many different geopolitical factors which makes it very difficult to predict. But barring some unforeseen circumstances, I believe first quarter should be the bottom, and this is the view emerging from our discussions with customers.

Takato Watabe: Understood. Thank you. Also, many thanks to Mr. Takii.

Michiharu Takii: Thank you for your kind comments. I didn’t take the opportunity to mention it in the presentation, but thank you all.

Takayuki Komori: Thank you. Next is Mr. Ikeda of Goldman Sachs.

Atsushi Ikeda: Can you provide a little more color on the LTA pricing, please? Currently, customers are carrying high levels of wafer inventory of around three months and are also facing challenges to profitability. In these circumstances, being able to maintain flat pricing may be the best you can hope for. Having said that, there are contracts in place under which you would hope to see prices rise. If the industry as a whole makes strong progress on inventory adjustments, lowering inventory to around 2x months, would it be easier to begin negotiating for higher prices from second quarter? Furthermore, can we expect the prices in the September quarter could rise by around 5%? Can you comment on the balance between LTA prices and inventory levels?

Mayuki Hashimoto: LTAs can be generally grouped into two types, contracts with fixed flat pricing and contracts where prices step up over time. Very simplistically, for the step-up contracts, using a 3-year contract of 3 million wafers as an example, typically, there would be one price point for this year, a higher price point for the next year and so on. Hypothetically, if the contract is for the purchase of 1 million wafers this year, although the push-outs generally are not as extreme, if we assume that delivery of 500,000 wafers is pushed into next year, it is difficult for us to ask for the customer to pay the 2025 price point for the pushed out 500,000 wafers, given that we know the customer is struggling. In such cases, we ask the customer to respect the price point set out in the contract. Obviously, we would not agree to cut prices, but agreeing to accept a push-out means that the anticipated increase in price also gets pushed out. That’s what’s happening now. However, there are many contracts where prices are fixed for the life of the LTA. There isn’t much impact from push-outs on these contracts. One more thing that complicates the picture is the spot business at FST in Taiwan. Although non-consolidated SUMCO is 100% LTAs, FST sells some of its wafers at spot prices, particularly when doing business with affiliates. Negotiations with affiliates, where FST is prioritizing the sale of wafers, can be challenging. In situations where affiliates are being asked to buy, price points can fluctuate. So, this is what’s happening now. However, if we look at the overall picture for the full year this year, on a blended basis, I don’t think that prices will fall on a year-on-year basis. Prices may rise slightly in the second half of the year, but I wouldn’t expect to see full year blended prices rise in the same way as they did in 2023. That is my view.

Atsushi Ikeda: Understood. So, prices could rise slightly this year, but may rise more significantly by around mid-2025. If we look at prices at the end of 2025 or in the first half of 2026, do you think they will revert back to the originally anticipated levels?

Mayuki Hashimoto: Yes, exactly.

Atsushi Ikeda: Understood. I have high hopes for SUMCO. Thank you.

Mayuki Hashimoto: The way to think about it is that all of the wafers have price points. So, there is a price for the first 1 million, another price for the next 1 million and so on. The rise in the blended price is impacted by the timing of deliveries.

Atsushi Ikeda: Understood. Thank you.

Takayuki Komori: Thank you. Next is Mr. Okazaki of Nomura.

Shigeki Okazaki: Thank you to CFO, Takii. I want to ask about sales volume and production volume. Three months ago, you indicated that you would be adjusting production to reduce your own inventory in fourth quarter following a cancellation, but suggested sales and production volume would return to normal in the March quarter. To what do you attribute the gap between what you have said last time and this time? Is it because customers are heavily focused on reducing inventory?

Mayuki Hashimoto: I don’t recall suggesting that inventory volumes would rebound from January. What I said was that we were expecting first quarter to be very tough.

Shigeki Okazaki: What I understood was that after the production cuts you planned for the December quarter, you expected to revert to normal levels in the March quarter.

Mayuki Hashimoto: In terms of the production cuts, given that total volume had fallen, we did not want to carry excess inventory, so we lowered our inventory to an appropriate level by cutting production volume. If you understood that to mean that we intended to increase production volume from January, then there may have been a misunderstanding. Fourth quarter into first quarter is the bottom for the market itself. In terms of what is happening in first quarter, although there are customers that reduced volume further, it is more that there were end users that are reducing wafer consumption further in first quarter. Overall, first quarter is probably flat to slightly lower on a sequential basis. Epi, in particular, is down. So, first quarter into second quarter is probably the bottom in my view. We have aligned our production to current conditions.

Shigeki Okazaki: With regard to second quarter on a sequential basis, as you mentioned earlier, the picture for the June quarter is still mixed, is that correct, and you expect a recovery in sales or production volumes to come in the September quarter?

Mayuki Hashimoto: That is true for epi. For SUMCO, the proportion of epi is higher, so I believe we should see a solid recovery. PW, however, is likely to take longer to recover. But even for epi, there are a variety of customers, some of which are doing well while others are not. The winners are doing relatively well, but there are some customers that are really struggling. The overall picture is mixed, so it is difficult to generalize. That said, I believe that generally epi will recover in the second half of the year.

Shigeki Okazaki: Understood. Thank you.

Takayuki Komori: Next is Mr. Nishiyama of Citigroup Securities.

Yuta Nishiyama: I would like to ask about the outlook for wafer demand. I can see on Page 15 that wafer input started to recover from the end of last year. If I look at the supply-demand chart on Page 19, SUMCO has revised up its forecast for 2024 by 7% relative to three months ago. In addition, the wafer industry, as a whole, is undertaking significant shipment adjustments. This suggests that the normalization of inventory may happen earlier than was expected three months ago, in my view. Can you comment on your view of these points?

Mayuki Hashimoto: I completely agree. In terms of the normalization of inventory, input levels should increase, although I would view first quarter as the bottom. However, conditions vary significantly by customer. The picture is mixed, which makes it difficult to generalize. What I can say is that, overall, first quarter is likely to be the bottom and that we should start to see recovery from second quarter. However, there will be customers where the recovery will be gradual while customers with strengths in DRAM or AI are likely to see a solid rebound. However, other applications, such as smartphones, are still weak. I will say it is very difficult to generalize.

Yuta Nishiyama: I see. When you talk about first quarter as the bottom, are you referring to sales? Profits are likely to be impacted by an increase in depreciation, I think.

Mayuki Hashimoto: I am talking about sales.

Yuta Nishiyama: Understood. Thank you.

Takayuki Komori: Next is Mr. Yoshida of CLSA.

Yu Yoshida: I want to ask about Slide 19 and the supply-demand balance. As mentioned in a previous question, you have revised up your demand forecast for 2024 from a negative growth to flat this time. Please talk about where you saw the largest change in your view in terms of applications. If possible, can you frame your comment in terms of epi. And for memory, could you distinguish between DRAM and NAND? Can you also share your assumption for customer wafer inventory levels as of the end of 2024, which underpins the flat forecast, i.e., how many months of inventory have you assumed?

Mayuki Hashimoto: We don’t have a digital read of customer inventory levels, so our analysis does not go that deep. The reason why we slightly revised our forecast to flat is because we expect an earlier recovery in epi. We also factored in an early recovery in DRAM, given the significant use of DRAM in generative AI.

Yu Yoshida: I see. Can I also ask what you are assuming in terms of the rate of change in production, either upward or downward, in order to achieve this forecast for flat shipments?

Mayuki Hashimoto: You mean the forecast for flat shipments, on an overall basis we expect it to be flat, although we are expecting a relatively sharp improvement in the second half of this year. Obviously, I could be wrong, but certainly, we would need to see a relatively solid recovery in smartphones to achieve this. That said, there will be an increase in smartphones enabled for AI. I also expect to see proactive growth in AI data centers. However, as you know, the customer’s production capacity for these applications is limited. They are working very hard to increase capacity. Fortunately for us, we are a major supplier to this customer, so I think it hinges on this customer. I don’t expect to see a dramatic recovery in conventional epi wafers, but do expect to see a rapid recovery in epi wafers for AI applications. I think it will hinge on this.

Yu Yoshida: Understood. Thank you.

Takayuki Komori: Next is Mr. Yamada of Mizuho Securities.

Mikiya Yamada: CFO, Takii, thank you for everything over the years. I would like to ask about how to think about LTA prices, apologies for being persistent. There was a discussion of specific prices for 2024 wafers, but the reason why overall prices are not at these levels now is because of the cumulative impact of previous push-outs. So, in thinking about price levels for 2025 and 2026, should we expect a similar trend?

Mayuki Hashimoto: The original contracts stipulate volume and price for each year. So, for instance, the contract would say x number of wafers for 2023 at a specific price. However, the contract does not state a price point for the remaining wafers in the event that not all of the wafers for a given year are delivered in that year. When customers ask to delay shipments, it is because they are struggling so it is difficult for us to say that the 2023 price is no longer valid for delayed shipments and that the price point is now higher. It is difficult to engage in this type of discussion unless both parties are doing well. Additionally, we have been involved in joint development to this point as well, so it makes it difficult to take a purely self-interested approach. I also recognize that the customer probably wants to ask for a lower price, but out of respect for their commitments, they do not. They feel bad about having to delay shipments. In this situation, both sides end up finding middle ground by sharing the pain. It is not specified in the contract, but there is an understanding that the portion of 2024 deliveries that is pushed out will be priced at the 2024 price point. Push-outs in the next year will follow the same pattern. The reverse is also true. If the customer ask to take more volume earlier, that portion would be at the higher price point. This is the unspoken understanding that we have with our customers.

Mikiya Yamada: In this situation, given that there is continued progress being made on process development, I am assuming the degree of difficulty for wafers in 2025 and 2026 will increase, does what you have just talked about mean that you must sell more advanced wafers at lower prices?

Mayuki Hashimoto: That’s different, we set prices for more technologically challenging wafers hire to begin with, so that is different.

Mikiya Yamada: In other words, the wafers are different, so the prices are different as well. A more advanced wafer is sold at a higher price, is that correct?

Mayuki Hashimoto: Yes.

Mikiya Yamada: Understood. Thank you. I have high expectations for an early normalization of inventory levels.

Mayuki Hashimoto: Me too. Please support us.

Takayuki Komori: Thank you for your questions. Thank you to everyone for joining the fiscal 2023 results briefing. We are grateful for your participation today. We will end the meeting here.

Mayuki Hashimoto: Thank you.

Michiharu Takii: Thank you.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.