Sumco Corp

TSE:3436

Earnings Call Transcript 2021-Q3

[Hiroshi Shibaya, executive]

I am Hiroshi Shibaya of Sumco Corporation. Thank you for your participation today and for your continued support. This is the results briefing for Q3 2021. Before starting the presentation, allow me to confirm today's materials, which consists of 3 items: the brief statement on consolidated financial results for the 9 months ended September 30, 2021, the announcement regarding revision to dividend forecast and the presentation deck entitled Results for the Third Quarter of Fiscal 2021. The materials are available for download on our website at sumcosi.com. We will have presentations today from Representative Director, Chairman and CEO, Mayuki Hashimoto; and Vice President, CFO, Michiharu Takii. Hiroshi Ito, General Manager of Accounting is also on hand.

We have set aside 60 minutes for this briefing, which will end at 5:00 p.m. Next, a disclaimer. The estimates, expectations, forecasts and other 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. Chairman and CEO, Hashimoto, will discuss our forecast and our 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, executive]

I am Chairman Hashimoto. Please turn to Slide 5 of the presentation. This is the overview of our Q3 results. Relative to guidance of JPY 86 billion, Q3 sales were JPY 86.7 billion, an overshoot of JPY 0.7 billion. Operating income was JPY 14.8 billion, JPY 2.3 billion higher than our forecast of JPY 12.5 billion. Similarly, for ordinary income, the result was JPY 2.2 billion higher than our forecast. Net profit attributable to owners of the parent overshot guidance by JPY 1.6 billion as well.

Overall, we came in well ahead of our forecast. The significant overshoot was the result of a combination of a number of factors. On top of sales growth, we were able to ensure costs were relatively well controlled. As well, depreciation expenses were incurred at a slightly slower pace than we had originally anticipated. It is quite rare for Sumco to miss its forecast by this much, but the miss was to the upside.

Next is our guidance for Q4. We project sales of JPY 88.5 billion, up JPY 1.8 billion sequentially. We forecast operating income to decline JPY 0.8 billion Q-on-Q. This may be surprising, but in addition to an increase in depreciation expenses, our second largest 300-millimeter wafer plant will undergo periodic maintenance in Q4. As a result of the related maintenance expenses and the impact of reduced production, we expect a slight sequential decline in profits. Conditions in the underlying market remain unchanged with supply-demand continuing to tighten. We project ordinary income of JPY 12.5 billion and net profit attributable to owners of the parent of JPY 9 billion. Our ForEx assumption is JPY 113 to the dollar.

Next, on to Slide 6 for a discussion of shareholder returns. We are guiding for a year-end dividend per share of JPY 19, which will bring the total annual dividend per share to JPY 36. This translates into a dividend payout ratio of 30.4%, in line with our stated policy of a target payout ratio of 30%. We have no plans for share buybacks at this time. As you know, we recently undertook a public offering, reflecting our strong need for cash as we expand our production capacity. We do not expect to conduct share buybacks for the time being. However, we do intend to maintain our 30% dividend payout ratio going forward.

Please turn to Slide 7 for a discussion of the market environment for silicon wafers. Starting with the Q3 results, as you already know, demand for 300-millimeter wafers from logic customers continue to increase, but we have started to see a tightening in demand for memory customers as well. We no longer have any more floor space available to increase production capacity. Through the adoption of AI technology, we have been scrambling to raise productivity and have managed to eke out some improvements. However, fundamentally, supply is not keeping pace with demand at all.

Market conditions for 200-millimeter in smaller wafers are very similar, driven by strong demand from automotive, consumer electronics and industrial applications. Conditions are very tight. With regard to prices, obviously, given the tight supply-demand situation, spot prices have been rising. Customers have accepted all of the price hikes. For LTAs, customers are respecting LTA prices although it is perhaps more accurate to say that we are ensuring that LTA prices are respected. The outlook for 300-millimeter in the current Q4 2021 results reflects the very tight supply-demand conditions for both logic and memory and the continued increase in requests for more volume from customers. Unfortunately, we are not in a position to respond. For 200-millimeter and smaller diameters, the situation is the same.

Prices on existing LTAs obviously continue to be respected, but prices in the spot market are still rising. Please note that Sumco's spot business accounts for only a very small proportion of the total, so higher spot prices will have only a negligible impact on Q4 earnings. On greenfield investments in order to be able to gradually add capacity for leading-edge semiconductor silicon wafer such as 5-nanometer on a step-by-step basis, we recently announced we would be investing in a sizable new physical shell as well as utilities and production facilities. We will invest JPY 78.6 billion to build a physical shell and utilities facilities at the Imari site. We plan to invest another JPY 122.9 billion in wafer production equipment for Imari for a total investment of JPY 201.5 billion.

We will also invest JPY 16.5 billion in a new physical shell and utilities facilities and JPY 10.7 billion in wafer production equipment at Omura City, Nagasaki Prefecture for a total investment of JPY 27.2 billion. The aggregate greenfield investments amount to JPY 228.7 billion. In terms of the timing of completion, the blueprints are already done. Construction will start early next year. Our plan is to gradually phase in capacity from the second half of 2023, reaching full capacity production in the second quarter of 2025.

Please turn to Slide 8. This is the wafer trend for 200-millimeter by quarter. We show the trends for every year since 2014. As you can see, every year, the wafer industry has been able to nudge production capacity up slightly. No one in the industry has invested in new facilities to expand capacity, instead everyone has focused on improving productivity in order to eke out small increases in production. Currently, the industry is at around 6 million wafers per month. At this level, I believe everyone is operating at maximum capacity.

Next on Page 9 is the same chart for 300-millimeter wafers. In Q3, the industry finally reached 7.5 million wafers per month. Here as well, although there may be a little more room for debottlenecking and there may still be some players with room to do more through brownfield investments, I think the industry is virtually maxed out.

Going forward, particularly in 2022 and '23, I don't believe that there will be anymore room to increase volumes. In line with this, please turn to Slide 10. This chart shows estimated customer inventory for 300-millimeter wafers. We believe that customer inventory has dropped to just over 1 month. Our view is that wafer purchases are still relatively well balanced versus wafer inputs. However, if we look at the next slide, Slide 11, which breaks out customer 300-millimeter wafer inventory between logic and memory, the situation is significantly different. First, if we look at logic, the most recent data point for customer wafer inventory is the red dot on the right. The reason why I wanted to show this data point in red is because it has dropped to the lowest level on record. Logic customer inventories have fallen well below 1 month, a very challenging situation.

Wafer input levels remain significantly higher than wafer purchase levels. Effectively, logic players are consuming wafers almost as soon as they come in the door. In contrast, memory customer inventories, as shown in the lower bar chart, are still at around 1.3 months. The gap between wafer inputs and wafer purchases is not that large. Having said this, memory customers are seeing their inventory levels gradually decline. The reason why customers are so desperate is because virtually no new wafer production capacity is expected to come online in 2022 and '23. Almost all of the wafer players must undertake greenfield investments to increase capacity.

What this means is that while everyone, including the customers, is investing in 2022 and 2023, even if new fabs come online, wafer inventories are already severely depleted. Customers are doing everything they can to build up wafer inventory. But nonetheless, as we show here, inventory levels are still falling. Therefore, when customers think about 2022 and '23, it appears they have a sense of crisis.

Next, please turn to Slide 13. When we look at market conditions for semiconductor use silicon wafers split between epitaxial and polished wafers, all of the companies that have announced investments to expand capacity will be building capacity for leading-edge epi wafers. The reason why capacity expansion is concentrated on leading-edge epi can be inferred from the trend shown in this chart.

As you can see here, the blue bars represent polish wafer demand and the orange bars epitaxial wafer demand based on volume. If we look at trends over the last 20 years or so, you can see that polished wafers have experienced 2 downturns during this time frame, reflecting periods of oversupply where price pressure was fierce. In contrast with this, you can see epitaxial wafers have shown steady and consistent demand growth over the same period without a downturn, generating a CAGR of 11.3%. As you know, epi wafers are used for logic with foundries as the key players. Polished wafers are typically used for memory.

When the macro backdrop deteriorates, set makers can easily reduce memory density per device without significantly impacting functionality. Data can easily be stored in the cloud. However, the same is not true of logic. Each device requires logic chips. In our view, this is what leads to the very steady trends in epi wafer volumes. Turning to Slide 14. This shows the scale of CapEx investments by our customers. As you can see, the foundries, which consume epitaxial wafers are making ambitious investments. The CAGR for foundry CapEx growth is 8.1%. Overall CapEx growth, on the other hand, is a CAGR of 5.9%, while the CAGR for IDM CapEx which is primarily memory, is 4.6%. Foundries continue to actively make large-scale investments. For an analysis of the implications, please turn to Slide 15.

On Slide 15, we show trends in 300-millimeter capacity at foundries. The historical growth rate for foundries had tended to hover around the 10% level. However, if we look at the momentum of current capital expenditures, it suggests that growth rates will likely be around the 13% to 14% level going forward with a major caveat that this growth is subject to wafer supply capacity. This is the underlying expected trend. If we now look at Slide 16, this shows trends in the evolution of logic technology. If we look at 2009, this is when we saw the transition to 40 or 45 nanometers. The pace of the transition was relatively slow. It took multiple years for the new technology to account for 20% of the market.

Subsequent transitions generally took several years. Now, however, the transition is very fast. 5 nanometer is already more than 10% and is likely to get close to 20% this year. It continues to grow very rapidly. The transition suddenly became much faster from 2017. This coincides with the foundries raising their game. Foundries have also been early to invest and generally very proactive. It feels like the foundries are rapidly widening the gap with the IDMs. Sumco's strength is in leading edge. This is where we focus. When the transitions are faster, there is less time to improve margins by moving up the learning curve. This makes it challenging to implement cost reductions while keeping pace with the transition. 5 nanometer in particular, was very challenging for us in the beginning. Now we are doing very well, but catching up is tough.

Next is Slide 17. This slide should be familiar to you. Our customers' demand forecasts are very bullish, so we have toned it down somewhat, but the current forecast is for growth of 8.4%. Compared to this, greenfield investments should allow the industry to catch up eventually. However, it is important to recognize that there is a significant difference in absolute terms between a 5% increase in wafer production capacity when industry production was 3 million wafers per month versus 10 million wafers per month. This is true for not just Sumco, but its peers as well. The key differences are the sheer scale of investments required as well as the lead time before new capacity comes online.

Capacity growth at our suppliers has not tripled or quadrupled over the same time frame. With the exception of a small number of major suppliers, lead times for wafer production equipment are getting longer and longer. So even if we are proactive in trying to grow our own capacity, there are bottlenecks. We would prefer to increase capacity at a much faster pace, but there are multiple bottlenecks in the supply chain, which means it will take time. I believe our competitors are in the same boat. To me, this suggests that greenfield investments will not catch up to demand for a while. This is why we show the bars for 2024, '25 and '26 not matching the expected demand line. It is likely that our customers share this view.

In the past, customers did not want to sign LTAs when prices were high, but were keen to do so when prices were low. This is the exact opposite of what wafer makers want. We want to lock in customers when prices are high, but prefer not to enter into LTAs when prices are low. Now, however, the interest of both parties are in alignment. Customers want to lock in wafer volumes for as long as possible. This is because of the market outlook. The market is expected to see clear shortages in 2022 and '23. Even for 2024, '25 and '26, customers have indicated that they are eager to lock in supply under LTIs. There is a sense that the tightness in wafers is not a temporary phenomenon, which is why customers are actively seeking to secure wafer volumes.

This completes my section of the presentation. I will hand over to CFO, Takii, to talk about the recent public offering and details of the third quarter results.

[Michiharu Takii, executive]

I, Takii, will present the earnings, but I will start with an overview of the recent public offering. Please turn to Page 19. I believe you already know we did a public offering so I will briefly comment on only the results. The announcement of a public offering coincided with our announcement of capital expenditures for greenfield expansion. As alluded to at the beginning of the presentation by Chairman Hashimoto, we plan to invest a total of JPY 228.7 billion, of which a portion was to be funded by a public offering as noted here. We issued 60 million shares as initially planned, reflecting strong international demand, overseas investors accounted for 82% of the total offering, while domestic investors represented 18%. This is different from our original plan for a 75-25 split between overseas and domestic investors.

The offering was fully sold out. Shares were sold at a 3% discount. We drew down a total of JPY 120.6 billion. Assuming that the public offering had been done at the end of September, if we added the capital to the balance sheet, splitting the increase in net assets evenly between capital stock and the capital surplus, the shareholders' equity ratio would have risen to 62% and the DE ratio would have been as shown here. As a result of the offering, cash and cash equivalents increased, technically putting us in a net cash position. This completes the overview of the public offering. I will now talk about the Q3 results.

Please turn to Slide 21. This is the summary of the 9-month results. 9-month sales increased JPY 25.7 billion year-on-year. Operating profit, ordinary profit and net profits all rose more than JPY 6 billion, respectively. If we look at the quarterly progression, as I have said before, we believe Q2 was the turning point. In first half, prices were down slightly on a year-on-year basis, mainly reflecting a 10% decline in portion of business covered by LTAs. However, this impact will drop out in Q3 and Q4. We show a number of metrics at the bottom of the page, such as OPM, EBITDA margin, ROE and EPS. As you can see, all of these metrics improved year-on-year.

Next, on Page 22 is the analysis of sequential change in operating income for Q3. We had not expected Q3 profits to be this strong. The main driver was cost. We had initially expected an increase of JPY 1.2 billion. At the time we had forecast an increase of JPY 0.5 billion or JPY 0.6 billion in electric power unit prices, reflecting the impact of the summer season as well as an increase in labor costs. However, while electric power unit prices did indeed increase in line with seasonality, yields were very favorable and we also made progress in reducing materials costs. As a result, the sequential increase in cost was limited to JPY 0.1 billion.

We also did not incur as much in depreciation expenses as initially expected due to delays in completions, the increase was JPY 1.3 billion. The better-than-expected earnings were primarily driven by improvements on the cost side. Sales-related variance was also slightly better than expected for a JPY 3.8 billion sequential improvement. The JPY 3.8 billion reflects a positive JPY 1 billion price impact on the back of rising spot prices as discussed previously, but the remainder is from increases in volume for both 300-millimeter and 200-millimeter wafers. The sequential change in foreign exchange impact was effectively 0.

Let's turn to Page 23, which shows the year-on-year change in 9-month operating profits. On a 9-month basis, sales rose JPY 25.7 billion, while operating profit was up JPY 6.8 billion. The yen depreciated versus the U.S. dollar, but was more than offset by the year-on-year increases in costs. The JPY 3 billion year-on-year increase reflected an increase in labor and repair and maintenance costs. R&D expenditures also rose year-on-year. However, this was offset by significant improvements in yields year-on-year and progress in reducing materials costs. Depreciation expense rose JPY 4.1 billion year-on-year. Sales-related variance and others improved JPY 15 billion year-on-year. Prices in the first half were down year-on-year.

As noted previously, this is the result of a decline of 10% in the proportion of LTAs in place for 300-millimeter wafers for an impact of roughly JPY 3 billion to JPY 4 billion. Spot prices were down slightly on a year-on-year basis as well for a total price impact of around JPY 5 billion for 200-millimeter and 300-millimeter combined. That said, as we have said several times already, the price declines were in first half. Increased volumes drove a year-on-year improvement in 200-millimeter and 300-millimeter of around JPY 20 billion, as highlighted earlier in the presentation by Chairman Hashimoto. The ForEx impact was a negative JPY 1.1 billion, largely related to moves in the new Taiwan dollar and the impact on our affiliate.

While yen trends were favorable for a positive impact of JPY 0.4 billion year-on-year, the new Taiwan dollar strengthened for a negative impact of JPY 1.5 billion. Moving on to the next page, which is the balance sheet. Cash and time deposits on the balance sheet increased JPY 11.5 billion. I will go into more detail on this in explaining our cash flow on the next page. Raw materials and supplies declined JPY 6.1 billion, reflecting the reductions in polysilicon inventory. On an annual basis, the decline in polysilicon inventory has an impact of more than JPY 10 billion. The Q3 figures reflect 3 quarters' worth of impact.

The outstanding balance of interest-bearing debt fell JPY 6.7 billion. If we look at the capital account, the JPY 2.5 billion decline in the capital surplus reflects the impact of buying back and returning treasure shares in first half. Retained earnings increased by JPY 19.4 billion, reflecting the impact of profits attributable to the owners of the parent less dividends paid, a combination of last year's fiscal year-end and this year's interim dividend. The shareholders' equity ratio improved to 54.5%, up 1.4 percentage points.

Moving on to the next page and a discussion of the cash flow statement. The subtotal of sources of funds under operating cash flow was JPY 71.4 billion. There was also a positive impact from a decrease in inventories. In investment cash flow, the net others line item includes cash, which had been transferred into a time deposit. From an accounting perspective, apparently, this action is to be reflected under investment cash flow. This is matched by the adjustment of JPY 4.1 billion under use of cash near the bottom of the table. After deducting for dividend payments and share buybacks and the repayment of debt from free cash flow, the ins and out of the time deposit mentioned earlier under the investment cash flow effectively offset each other from a cash perspective.

Summing up, all of these factors leaves us with a net increase in cash and time deposits of JPY 11.5 billion after reflecting currency translation adjustments. Next, on Page 27, I will comment on the full year forecast. The Q4 forecasts are as stated here. On a full year basis, sales are expected to increase JPY 41.6 billion year-on-year. Full year operating income is projected to exceed JPY 50 billion. Our net profit guidance is JPY 35.9 billion. The forecast for various metrics such as OPM and ROE are as shown at the bottom of the table. Please note the JPY 1.5 billion nonoperating loss we expect for Q4. Of this amount, JPY 0.9 billion is various expenses related to the recent public offering. This is the reason for the increase in nonoperating losses in Q4. Please turn to Page 28.

This is the sequential analysis of changes for Q4 operating profit. The single biggest factor behind the sequential decline in operating profit is the periodic maintenance at the mainstay Imari plant scheduled in Q4, which is pushing up repair and maintenance costs. We also expect an increase in depreciation. In sales-related variance, we are assuming volumes will be essentially flat Q-on-Q. So the increase of JPY 0.8 billion is mainly driven by the expectations of higher spot prices. The ForEx impact reflects the JPY 3 change to our dollar yen assumption.

Finally, Page 29 shows the analysis of changes in operating income for the full year forecast. If you look at the waterfall chart, again, we expect cost to increase for a JPY 6.2 billion negative. Earlier, I discussed the increase in labor and repair and maintenance costs. This will be slightly offset by improvements in yields and progress in reducing materials costs. Depreciation is expected to increase JPY 5.6 billion year-on-year. Under sales-related variance, I touched upon a JPY 5 billion impact from price declines in the 9-month results. But given the slight sequential improvement in spot prices from Q3 into Q4, the negative from prices on a full year basis should be slightly lower than JPY 5 billion. However, the impact from the increase in volumes is expected to be a positive JPY 28 billion to JPY 29 billion, supported by growth in both 200-millimeter and 300-millimeter wafers.

When we look at the impact of foreign exchange on a full year basis, the strong new Taiwan dollar has been a drag on profits until the end of Q3. This figure includes a JPY 2 billion plus/negative from the new Taiwan dollar. However, this is offset by a positive impact of more than JPY 2 billion for dollar yen for a net positive impact of JPY 0.8 billion. Our outstanding balance in U.S. dollars is around $10 billion. So the positive on yen weakening is around JPY 3 billion, which has been offset by the impact of the strong new Taiwan dollar. Turning to the reference section. Please open to Page 31. If we look at quarterly sales and operating profit, you can see that quarterly sales have surpassed the peak level seen in 2018 when profits were relatively high. Moving on to the next page, Page 32. You can see the EBITDA margin has also been gradually improving.

Compared to the levels of Q1 and Q2, the EBITDA margin has improved significantly. Please review the remaining pages at your leisure. This completes my remarks.

[Hiroshi Shibaya, executive]

Thank you. We will now open the floor to questions. We ask you to kindly limit yourself to 1 question each. We will start with Mr. Enomoto of Bank of America Securities.

[Takashi Enomoto, analyst]

I am Enomoto of BofA Securities. I would like to know more about the new LTAs put in place with customers related to the new capacity increases you have announced. For instance, how much of your investment do you expect to recover? How much higher are unit prices? What is the duration of the LTAs? On new LTA prices, is the approach to have prices gradually rise every year or will prices increase in 1 fell swoop? Also do the LTAs applied to only wafers to be supplied from the new facilities or have you put in place new LTAs, which cover both existing and new commitments? I would be grateful for any color you can share on the new LTA.

[Mayuki Hashimoto, executive]

First of all, with regard to the LTAs related to the new capacity expansion, we have applied the law of one price, i.e., identical goods will be sold at the same price. The new LTAs combine existing and new commitments under a single agreement. The price point will gradually rise over time to a peak level. Our intention was to set pricing at a level that would allow us to recoup our investment in 5 years, which is what we have done. All of the new expected capacity has been spoken for. A number of customers have agreed to the new LTAs. Does that answer your question?

[Takashi Enomoto, analyst]

Given that you are investing more than JPY 200 billion, will you be able to fully recover your investment over time or just the infrastructure?

[Mayuki Hashimoto, executive]

We can fully recover our investment.

[Takashi Enomoto, analyst]

You had previously suggested prices would need to rise 50% to 60%. Is this the right image? Were the price hikes close to these levels?

[Operator]

Under the new LTAs, we don't distinguish between existing and new supply commitments. Pricing is the same for identical products. If the new LTAs were just for the new supply, then prices would need to double. But given that we have combined existing and new commitments and are applying prices on a pro rata basis, the percentage increase is different for each customer. The blended price would be impacted by the scale of the existing LTA, so the magnitude of price hike varies by customer. What I can say is that the magnitude of the price hike is significant.

[Hiroshi Shibaya, executive]

Thank you. Next is Mr. Ikeda.

[Atsushi Ikeda, analyst]

I am Ikeda of Goldman Sachs. I would like to ask a follow-up to the previous question. You indicated that you have structured the LTAs to recover your investment in 5 years. If we look at prices for leading-edge epitaxial wafers, I believe you would need an average price of around $240 or $250. Would we be correct in assuming price hikes to this level? I believe you are still in negotiation for memory, but reflecting the very tight conditions for polished wafers, we have heard that customer inventory levels are now at around 2 weeks. Given this, I am expecting that memory use wafer prices will also rise significantly into the next fiscal year. Taking all of this into account, if we look at price levels for 2022 through 2024 or '25, my view is that overall average selling prices will be higher when compared to prices in the previous up cycle in 2017 and '18. What do you think? I would be interested in hearing your view.

[Mayuki Hashimoto, executive]

With regard to epi prices, I would agree that we will see strong price increases. However, for polished wafers, I believe prices have already risen significantly in the last up cycle. So I would not expect to see further significant price rises for polished wafers. Wafer makers are much more focused on increasing epitaxial wafer production capacity, particularly the majors. So given the focus is on expanding epi wafer capacity, I would expect to see larger price rises on epi. Also, a price point of $240 seems too high, unless you were talking about wafers for CMOS image sensors.

For polished wafers, even if prices were to rise, the magnitude of the increase is unlikely to be as large as the rise in epitaxial prices in my view. That said, the supply-demand balance is out of sync and has been for a very long time. So it is a little difficult to read how things will play out. What is true is that the customers are anxious and they are prepared to agree to a fixed price for as much volume as possible.

[Atsushi Ikeda, analyst]

I believe that epitaxial wafer prices have effectively been fixed out to 2026, but that you are still conducting some negotiations on prices for memory use wafers?

[Mayuki Hashimoto, executive]

No, actually, we have virtually sold everything to 2026, although there are some cases where the formal signing has yet to take place. We have reached agreements on virtually all of our capacity.

[Atsushi Ikeda, analyst]

This is for both epi and polished wafers?

[Mayuki Hashimoto, executive]

Yes, both.

[Atsushi Ikeda, analyst]

Understood. If I could just confirm 1 point. Would I be correct in assuming that the new contracts supersede existing contracts that cover the period from 2022 and that, therefore, prices will rise from 2022?

[Mayuki Hashimoto, executive]

Yes, that's fine.

[Hiroshi Shibaya, executive]

Next is Mr. Watabe.

[Takato Watabe, analyst]

I am Watabe of Morgan Stanley. On Slide 7, you show details of the greenfield capacity expansion plans for wafer production capacity at both Imari and Omura. There has been much discussion in the market about the expected capacity of the new facilities with suggestions that capacity could be 200,000, 300,000 or 400,000. Can you provide clarification on capacity on a combined basis, including Omura?

[Mayuki Hashimoto, executive]

With regard to the new capacity, we already have signed agreements in place with our customers, so we aren't in a position to share detailed information. What I can say is that as a general comment, to increase polished wafer production capacity by 100,000 requires an investment of around JPY 38 billion. If you are producing epitaxial wafers, particularly leading-edge wafers, then you would need to invest an additional JPY 10 billion plus for an epitaxial furnace enabled for 100,000. The requisite inspection equipment is also very expensive, typically costing more than JPY 10 billion. So if you are fabricating leading-edge wafers, the total investment would be JPY 60 billion or JPY 65 billion. In some case, the new capacity is 100% leading-edge wafers. So the investment amount is relatively large. If you are only doing polished wafers, then the total investment would be less than JPY 40 billion. That's what I can say as a general comment.

[Takato Watabe, analyst]

Understood. I am convinced that expected capacity is about $200,000.

[Hiroshi Shibaya, executive]

Next is Mr. Nishiyama.

[Yuta Nishiyama, analyst]

I am Nishiyama of Citigroup Securities. My question is about the greenfield investments. You have said you are aiming to recover your investments of slightly less than YPY 230 billion within 5 years. If we take this statement at face value, it is very positive. If you have to cite what you consider to be downside risk, what would those be? As an example, for some of the LTAs from 2017, there were instances where you had to allow customers to push out deliveries. Also, subject to the timing of the silicon cycle, if 2024 turned out to coincide with the downward leg of the cycle, given that the wafer supply volumes are expected to increase at around that time, is there not a potential risk that the market could temporarily face excess capacity? If there are differences between the new LTAs that would help better in-force contracts compared to previous LTAs, I believe the equity market would have a higher degree of comfort with the capacity expansion, what do you think?

[Mayuki Hashimoto, executive]

In terms of how the new LTAs are positioned, as you know, many customers signed LTAs in 2018. The customers that have signed new LTAs this time are those customers that had LTAs dating to 2018, but respective the LTA conditions even when the market was weaker in 2020. If the question is whether these customers have only signed new LTAs exclusively with Sumco, I believe that they have also put LTAs in place with peers to a certain extent. However, these customers have indicated that if wafer consumption declined, they would cut off their other suppliers, but would respect the volume requirements of their LTAs with Sumco.

The difference this time around is that Sumco is investing to increase capacity and we need to be able to recover our investment. We are not asking customers to pay higher prices on wafers being produced at existing facilities. Given our need to recoup our investments, we were more vigilant in confirming these points with our customers. That said, in the event of sudden and unexpected developments, like an escalation of trade friction between the U.S. and China, a rapid strengthening of the yen or a surge in crude oil prices, we are thinking about how best to cope with these kinds of developments and aim to respond as appropriate. The key in such situations is to be prepared to respond quickly, which is a major focus for Sumco. There is no such thing as a safety net. There are always risks in business. However, with the recent capacity expansion plans, we were much more rigorous in confirming all of the relevant factors than we were last time. In particular, with polished wafers, there is a lot of price volatility. This is why PW is perceived to be more risky.

We are focused on leading-edge wafers. There is only 1 other company that is enabled for leading edge. Expanding capacity just for leading edge is challenging from a technological standpoint and barriers to entry are also high. From that perspective, I believe our investments are safer than more generic polished wafer capacity investments. In addition to this, there are customers with whom we have had LTAs over long periods of time and have particularly strong relationships. From this perspective, we have been very focused on expanding capacity as safely as possible. There is no such thing as a riskless business, but what I can say is that I feel that we have done our utmost to address potential risks.

[Yuta Nishiyama, analyst]

If you compare the current situation to the previous cycle, this time you undertook a public offering and improved your financial soundness. Did this also make it easier to negotiate with customers?

[Mayuki Hashimoto, executive]

It is certainly true that we became cash-rich as a result of the PO. We will move quickly to build the physical shelf this time, but we are not proposing to completely populate the space from the outset. We will be building a large shell, which is slightly larger than needed for the proposed capacity expansion, so there will be room for some go to expand in line with market growth in a timely manner as necessary. This is in line with our policy. The one area of surprise was our suppliers. As discussed earlier, the scale represented by a 5% increase when the market is 3 million wafers a month versus 10 million a month is completely different. On an absolute basis, the increase is triple the number of waivers. However, our suppliers have not tripled or quadrupled in scale over the same time period. At most, they have grown by 10% to 20%.

In this situation, if we were to try to ramp up capacity very rapidly, lead times on equipment would get longer and longer. In this industry, supplier relationships tend to be exclusive even from the inquiry stage. It is probably the same for Shin-Etsu, but generally, our suppliers do not sell to our competitors, but only to Sumco. What we gained through this is the ability to share significant know-how with our partners. This is the case for much of the equipment that is required for wafer fabrication. This is why no one in the wafer industry can ramp up significant volume overnight.

In the past, when Sumco doubled its capacity, we were still talking about a global scale of 800,000. Even if we use the most conservative baseline assumption of PPP GDP growth at an annual rate of 5%, the absolute scale of a 5% increase is completely different between then and now. This is why wafer makers are struggling to match demand growth with capacity growth. This is the single biggest difference between last time and now. Because of this, wafer makers are not in a position to tip the market into oversupply even if they could.

Previously, I personally have been concerned about the risk of excess capacity in 2024. But when we actually sat down to plan capacity expansion and the PO, it became clear that it is just not possible to ramp capacity quickly enough to trigger an oversupply. Our customers are full of complaints about the current capacity expansion. They want to know why we can't do more. I am being pestered from morning to night. I expect that our peers are in the same boat. I am skeptical that all of the wafer makers will actually be able to achieve the capacity increases they have announced for 2022, '23, '24 and '25.

[Hiroshi Shibaya, executive]

Next is Mr. Okazaki.

[Shigeki Okazaki, analyst]

I am Okazaki of Nomura Securities. With regard to prices, I would like to confirm a few things. If we look at the quarterly progression, I believe prices primarily for spot have driven sequential improvements of JPY 1 billion per quarter. Should we expect to see this trend continue into next year?

[Mayuki Hashimoto, executive]

Next year, I expect to see a significant jump in prices.

[Shigeki Okazaki, analyst]

Are you referring to just spot prices?

[Mayuki Hashimoto, executive]

No, next year, Sumco will have virtually nothing in spot in 300-millimeter.

[Shigeki Okazaki, analyst]

You are saying that you have effectively locked in all of your capacity?

[Mayuki Hashimoto, executive]

Yes, most of what we had in spot has been locked in.

[Shigeki Okazaki, analyst]

Sumco had a high proportion of 300-millimeter wafers under LTAs to begin with. Does the same apply to 200-millimeter as well?

[Mayuki Hashimoto, executive]

200-millimeter is not that high.

[Shigeki Okazaki, analyst]

I see.

[Mayuki Hashimoto, executive]

Our approach had been to leave some capacity available for spot transactions, but when we look at near-term customer demand for 200-millimeter, we have not been able to keep up. We do have LTAs in place. But unlike 300-millimeter, the duration of 200-millimeter LTAs is shorter. This is because entering into an LTA carries risk for both sides. As well, the incentive for us to recover cash is not as high on 200-millimeter as it is on 300-millimeter because depreciation is already largely done. So the focus for 200-millimeter is to act as a cash-generating business to support the 300-millimeter business. So we don't have very long-dated LTAs for 200 millimeter, but recently, we have signed some contracts for 3 years.

[Hiroshi Shibaya, executive]

Thank you. We are getting close to the allotted time, so we will take questions from 2 more people. Next is Mr. Yamada.

[Mikiya Yamada, analyst]

I am Yamada of Mizuho Securities. I have 1 question. You indicated that you expect silicon wafer prices to start rising from next fiscal year, reflecting the market environment. If that is the case, if we think about the JPY 120 billion in proceeds from the public offering plus your capital, given that capital is likely to grow on the back of continued contributions from net profits even before the new facilities begin operating, you will have JPY 180 billion or JPY 200 billion after paying dividends. In other words, I believe you could actually fund this round of investments with capital on hand. Am I right? If so, why did you do a public offering? Why didn't you use debt financing this time?

[Mayuki Hashimoto, executive]

The answer to your question is tied to when the physical shell will be fully populated. From our perspective, we wanted to have funding in place so we could build out fabrication capacity whenever we needed to. As discussed earlier, all of the pieces fell into place. our preparations, including the conviction necessary to proceed as well as the customers. The only exception was our suppliers. There is not much that can be done to speed up delivery times for equipment. However, our customers are not at all satisfied with the scale of capacity expansion we announced this time. They continue to say that they will need more wafers in future. When we tried to bridge the gap, we felt that we needed the financial wherewithal in the form of cash to have the ability to strike when the iron is hot. Borrowing from the banks each time we needed funding would work if we were talking about a one-off increase in capacity, but this is not necessarily the case this time.

[Mikiya Yamada, analyst]

In other words, you were looking further out to the future in deciding to do the PO. If you need to expand capacity in the future, you would also be able to tap into capital that has built up over time to expand capacity quietly without having to make announcement. Is that right?

[Mayuki Hashimoto, executive]

Yes, that's it.

[Hiroshi Shibaya, executive]

Finally, our last question is from Mr. Azuma.

[Yoshihiro Azuma, analyst]

I am Azuma of Jefferies. Extrapolating from the discussion of the current capacity expansion, if we look at Slide 17, it appears that the greenfield capacity expected to come online in 2024 is about 800,000 -- no, excuse me, 1 million wafers per month. This likely includes MEMC Korea, however. So if we exclude this, the incremental portion is probably around 800,000. I believe actually achieving an increase of 800,000 by the end of 2024 would be close to impossible. I think the most that could be achieved would be an increase of 600,000. Is this overly conservative?

[Mayuki Hashimoto, executive]

You raised a good point. This is not the level you get if you simply aggregate the forecast. However, our view was we probably did not have full visibility. The forecast we show here does include some guestimates. So you may be right that the market may not get to this level. It is true for us as well as our peers, but my intuitive reaction is that the market has grown very significantly. Even if a customer came to us and said supply us with more wafers because we expect 5% growth, it can't be done because we can't catch up with demand. Going forward, we and our suppliers will need to try many things in order to keep up with the much larger scale of incremental growth. The wafer manufacturing industry as a whole has not been able to keep up. I think there could well be a risk that wafers bottleneck that slows semiconductor growth.

[Yoshihiro Azuma, analyst]

I see. Understood. It may well be that there is no choice but to be conservative in factoring in the increase in capacity by other players. My personal view would be that the German company can only really get to half of their publicly stated target at best. I don't think we can take their statement at face value.

[Mayuki Hashimoto, executive]

We also looked into what our competitors, not necessarily the German player, were doing. Everyone has 100 or more steps to get through if you really break it down. There are many SMEs and family companies that are involved with the individual stages where we, for instance, would be providing education and support. These kinds of small businesses are not positioned to suddenly deliver on the kind of volume required at Imari. It means that we need to think about educating our next suppliers. However, this is not an easy business. It is very challenging. Our suppliers have truly become a bottleneck, which is a challenge. The semiconductor market will continue to grow strongly, but wafer makers are struggling to keep up with the current pace. This is why shortages are likely in 2024 and '25. We are not being overly optimistic. It is simply the reality. I believe our customers share this view.

[Yoshihiro Azuma, analyst]

One last thing that I would like to ask. On this chart, below the green section is brownfield and existing capacity. Judging from this chart, the implication is that capacity will be flat. However, if there is a transition to 3 nanometers, given the ingot raising speed and polishing process is likely to get longer, won't this portion of capacity actually decline?

[Mayuki Hashimoto, executive]

Yes, but we are currently addressing this through the adoption of AI to boost production. In our case, production volumes actually increased last year even though we did not expand capacity. If everything were to migrate to 5-nanometer, the same thing would happen. Production volumes would decline temporarily. As this is our livelihood, we are actively focused on getting back to or exceeding the original levels of productivity. In particular, we have concentrated on adopting AI recently. It has made it possible to monitor the equipment and facilities very accurately on a real-time basis. It has proven to be very beneficial for improving productivity. We are using AI to drive a recovery in productivity after advances in technology raise the degree of difficulty. We have our hands full at the moment. So we haven't been able to adopt AI to further boost existing products.

[Hiroshi Shibaya, executive]

Thank you. We are grateful for your participation today. We will end the meeting here. We look forward to your continued support. Thank you.

[Statements in English on this transcript were spoken by an interpreter present on the live call.]