Sumco Corp

TSE:3436

Earnings Call Transcript 2020-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 third quarter fiscal 2020. Before starting the presentation, allow me to confirm today's materials, which consist of 3 items. The brief statement on consolidated financial results for the third quarter of fiscal year 2020, the announcement regarding the revision to the dividend forecast and the presentation deck entitled Results For The Third Quarter of Fiscal 2020. 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; Representative Director and Vice Chairman, Michiharu Takii. Hiroshi Ito, General Manager of Accounting, is also on hand. We have set aside 60 minutes for the briefing. The briefing will end at 5:00 p.m.

Finally, a disclaimer. The estimates, expectations, forecasts and other future information discussed here 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 operating environment to be followed by an explanation of the earnings results by Vice Chairman, 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 Hashimoto. I will present an overview of our results and discuss the market environment. Starting with the overview of our third quarter results. Sales were JPY 71.6 billion; operating income, JPY 6.6 billion; ordinary income, JPY 5.9 billion; and net income attributable to owners of the parent was JPY 3.4 billion. The dollar yen rate for the quarter was JPY 105.9. We missed our guidance this time. Since the start of the COVID-19 outbreak, we significantly overshot our first quarter guidance, were broadly in line for second quarter but fell well short of our forecast in third quarter. We apologize for the deteriorating accuracy of our forecast.

There were 2 major factors in third quarter. The first was the extremely large-scale typhoon in September. Unfortunately, the Kyushu area where we have manufacturing plants was broadly impacted. The typhoon hit all of our main plants in Miyazaki, Saga, Imari and Nagasaki, although there was no direct damage to the facilities. However, for safety reasons, we asked our employees to shelter at home for 2 days, which impacted output. For better or worse, the second factor had positive and negative elements for Sumco. Sumco has a very high share of leading-edge wafers. We saw a strong surge in leading-edge demand, particularly from logic. Typical transitions to date had been more gradual, starting with small initial lots with volume increasing over time.

By the time volume reaches relatively high levels, we would normally have completed the debugging process, bringing down costs. However, this time, the customer chose to ramp up very quickly, and Sumco was effectively the sole supplier. Sumco devoted significant effort to meeting customer demand rather than allowing a shortfall at Sumco to force the customer to stop production. As a result, Sumco was producing large volumes at suboptimal yield, which substantially pushed up costs. These 2 factors led to the shortfall versus forecast.

Moving on to the forecast for fourth quarter. Typhoon season is not a factor in the December quarter. We have also largely completed the debugging process, so yields on leading-edge should return to more normal levels. Our fourth quarter guidance is for sales of JPY 72.5 billion, operating income of JPY 7.5 billion, ordinary income of JPY 7 billion and profit attributable to the owners of the parent of JPY 5 billion. Our ForEx assumption is JPY 105 to the dollar.

Next on to Slide 6 and shareholder returns. This is the forecast for the full year dividend. Guidance for the fiscal year-end dividend is JPY 9 for a full year total of JPY 27. In addition, we are guiding for a share buyback of JPY 2.5 billion as usual for a total shareholder return ratio of 41.1%, largely in line with the previous year.

Please turn to Slide 7 for a discussion of the market environment. As noted earlier, demand for 300-millimeter logic was very strong in third quarter. For memory, although there are slight signs of recovery, the signs are not particularly strong. Trends in memory are probably best described as not strong but normal. In 200-millimeter, order levels were very strong in second quarter, with customers building inventory on supply concerns about possible logistics issues, as previously noted. In third quarter, demand dropped off in reaction to the strong second quarter, particularly for consumer and automotive applications. On prices, we were able to maintain price levels for LTAs, but for diameters of 200-millimeter or smaller, spot prices were slightly weaker.

When we look at the outlook for fourth quarter, we expect the strong trends in 300-millimeter logic will persist. Demand for leading-edge wafers, in particular, is extremely strong. Customers are demanding even more volume, we are struggling to keep up. As yields improve, volumes will increase. We are making good progress on this front. For memory, demand is not bad and has not weakened. However, visibility is not great. For 200-millimeter and smaller diameters, we are seeing a rapid recovery. In particular, we are seeing strong orders for the year-end period of December into January.

Based on this, I think first quarter 2021 levels could well recover to match the levels we saw in second quarter 2020. Since the start of the COVID-19 outbreak, our forecasting has been less reliable, for which I apologize. Much will hinge on trends in 200-millimeter and smaller diameters. For now, we are seeing a rapid rebound in demand. We expect to maintain LTA price levels. In terms of the outlook for spot prices for 200-millimeter and smaller diameters, although we are not in an environment where it is possible to return to previous higher levels, we would expect to see a flat sequential trend. Looking further out, 300-millimeter demand should be supported by a number of trends such as teleworking and working from home and demand from 5G smartphones and data centers. All of these factors should support stable growth, primarily in logic. However, for 200-millimeter, while we are seeing a sharp near-term recovery, future prospects will depend on whether or not we see a global macro deterioration on the back of a resurgence in COVID-19.

What we can say is that, for now, we are seeing a rapid recovery in the near term. Whether we see a more sustained recovery in demand or in the near term the rebound falters will depend on consumer and industrial demand. For now, it is my view that we will return to more normal levels of demand in 2021.

Please turn to Slide 8. This is the wafer trends for 200-millimeter by quarter. The red line is 2020. As you can see, we saw a surge in demand in second quarter followed by a big drop-off in third quarter. For fourth quarter, we are expecting a relatively solid rebound, but more towards the end of the quarter, the December and January time frame. This is what we're seeing now. For now, we expect levels in 2021 will not be bad, but I will also say that at the moment, visibility is not great. The trend for 300-millimeter is strong. Third quarter was above 6.5 million wafers. I would expect fourth quarter to maintain this pace. Logic is driving the strength.

Moving on to Slide 10. This is our view of customer inventory in 300-millimeter. The bars represent the aggregation of logic and memory. As you can see, inventory has come down slightly. If you separate logic and memory, there is a difference in trends. Unfortunately, we cannot share the specific trends in logic and memory because of client confidentiality. But in general, inventory in logic is declining meaningfully. For memory, the situation varies from customer to customer with some continuing to see increases, while others have seen inventory declines. For memory players overall, inventory has not necessarily returned to healthy levels.

Please turn to Slide 11, which looks at the impact of U.S.-China trade friction on wafer demand, an area of keen interest for investors and analysts. There has been talk that SMIC might be the next trade ban target. We looked at China's share of the global wafer market. Looking just at wafer consumption by local Chinese players, we estimate they account for around 6% of the global total at this time. If we look at all wafer consumption in China, including non-Chinese fabs, such as Intel in Dalian, Samsung in Xian, TSMC and SK Hynix, it represents about 15% of global demand. If the U.S. were to expand the scope of its ban, they would most likely target local Chinese players for inclusion in the entity list. Even if all the Chinese players were banned, given they only represent 6% of the global market, the potential loss of demand could be easily covered by demand elsewhere. From that perspective, we do not believe it would be a major issue for the wafer market.

Next, we looked at the impact on the wafer market from end users. Please see Slide 12. Here, we look specifically at base station players. Going forward, Huawei is expected to suffer a significant loss of share. When we think about which companies will take up Huawei share, while Nokia and Ericsson are players, Samsung has recently made significant inroads in this market. Samsung has significant technological expertise and management capabilities. My sense is that Samsung may well be the major beneficiary from Huawei's share loss.

Please turn to Slide 13 for a look at base station market share by region. As you can see, Huawei has a very strong presence in EMEA, but only a tiny share of the North American market. Huawei's share for Asia Pacific is in between these 2 regions. I would expect that Samsung's share in Europe will increase significantly. Of course, Ericsson and Nokia should also gain share, but I expect Samsung will replace Huawei.

This slide specifically looks at the impact on the base station market but please turn to Slide 14 for the impact on smartphone handsets. As you can see on this slide, Huawei's share of smartphone handset shipments is around 20%. Each bar represents a single quarter, so you can multiply by 4 to get a sense for annual levels. At this level, even if Huawei were to go away, there would be many others that will step up. Players like Oppo and Vivo are keenly looking for opportunities and would be pleased to grow share. Given this, we don't expect a Huawei ban to have a big impact on overall demand.

That said, the Huawei portion would likely be split between a number of players. For us, therefore, it is likely that the business would be split across a number of our customers.

Moving to Slide 15. We have discussed this previously, but this looks at how much smartphone wafer consumption increases in the move to 5G. Very roughly speaking, wafer consumption increases 1.7 fold. However, although DRAM and NAND memory density is expected to increase, it appears that because of the application processors are expensive, some handset makers may be considering reducing DRAM and NAND content in order to lower the price point. Therefore, it is not entirely clear whether memory will increase as much as is shown here.

That said, so far, based on teardowns of 5G models, these figures are in line with the actual increases in handset memory.

On Slide 16, we show the outlook for 300-millimeter wafer demand related to smartphone handsets. With regard to trends in total handset numbers, there is a silver lining. The outlook is not all bad, although absolute numbers are down this year. Of course, simplistically speaking, a decline in total handset volumes is negative for wafer demand, all else unchanged. However, as discussed earlier, the move to 5G will drive an increase in wafer consumption on a per handset basis. If you look at the bar chart on the left, the orange portion of the bar is 5G. It currently accounts for about 20% of the total, but as shown on the chart, 5G's share of the total is expected to increase significantly over time, rising something like 10 points a year over the next few years.

The increase in 5G share will drive a higher wafer consumption. Therefore, even if total handset volumes are down, it doesn't necessarily mean that wafer volumes will also decline at the same rate. Instead, as we have seen to date, demand for 300-millimeter wafers is likely to increase over time. If we exclude 200-millimeter wafer demand for automotive and consumer use, overall, I believe demand for 300-millimeter wafers should show steady demand growth in 2021 and beyond, driven by logic.

On Page 17, we show the outlook for data center and server demand. We would expect to see similar demand trends here as well. We don't really have an accurate figure for data center wafer demand. What we show here is a relative forecast. Demand could grow by 300,000 to 400,000 wafers over the next 3 to 4 years. Our customers' customers are proactively investing, as you can see on the chart in the left.

Please turn to Slide 18 for a summary of the overall picture. The bottom line, in our view, is that there will be a shortage of 300-millimeter wafers in 2023. We would expect 2021 to be better than 2020. In Sumco's case, our learnings from the ramp-up of leading-edge will enable us to increase volume slightly, and I would expect the situation to be slightly better. Overall, leading-edge production consumes a larger proportion of capacity. This is true not only for Sumco but for the industry as a whole. Not only is leading-edge more time consuming, but yields tend to be lower. As a result, leading edge reduces overall capacity. Therefore, at the moment, if anything, the market is probably already seeing shortages on the logic side. There does not appear to be shortage in polished wafers. However, logic requires epi furnaces. Adding new epi furnace capacity takes 1 to 2 years. This makes it challenging to catch up when demand is ramping rapidly. We are working very hard to improve productivity and hope that we can catch up.

For 2022, I would expect to be virtually back to parity. So signing LTAs for 2022 and beyond will be challenging. From our perspective, as utilization rises, we want to bring price points back up, although our customers would probably disagree. So it will be challenging. It is true that there is probably a little more room for brownfield expansion within the industry now, but that is likely to be fully tapped out by 2022 or '23, so we will probably need to put our thinking caps on at that point. This completes my segment of the presentation.

I will now hand over to Vice Chairman, Takii to talk about the earnings results in more detail.

[Michiharu Takii, executive]

I am Vice Chairman, Takii. I will present the third quarter results in more detail. Please turn to Slide 20, where we show the cumulative 9-month results. As noted at the beginning by Chairman Hashimoto, third quarter sales and operating profit fell Q-on-Q. On a 9-month basis, sales fell close to JPY 10 billion year-on-year, operating income dropped approximately JPY 12 billion year-on-year and net profit declined JPY 8 billion year-on-year. CapEx was JPY 42.4 billion, down slightly year-on-year, but depreciation increased year-on-year. As such, the financial metrics at the bottom of the table such as OPM and the EBITDA margin declined on a year-on-year basis.

Please turn to Slide 21. This time, our results were well short of our forecast, so we have included a waterfall chart comparing our third quarter results to the forecast to better explain the differences. Sales fell JPY 1.4 billion short of our forecast. This breaks out as follows: there was a JPY 1 billion negative owing to the impact of the typhoon that hit the Kyushu region. In addition, we had factored in an improvement in productivity for leading-edge wafers, but yields were weaker than expected. The impact of the lower yields, which translated into lower sales was minus JPY 1.6 billion. We were able to offset the total top line negative of JPY 2.6 billion by efforts to generate revenues in areas for a net decline of JPY 1.4 billion.

Operating profit fell JPY 1.4 billion short of plan. First of all, costs were JPY 1.1 billion higher than expected. This includes a negative JPY 1.3 billion related to the lower-than-expected yields on leading-edge wafers, as alluded to earlier, which depressed the OPM. This effectively accounts for the bulk of the higher costs. Depreciation was lower than we had expected. Sales-related variance and others was higher-than-expected at JPY 0.8 billion. This reflects the impact of lower sales volume in leading-edge wafers and lower yields, which depressed OPM versus forecast owing to mix.

OP was around JPY 0.3 billion lower than expected. There was also a negative impact of around JPY 0.5 billion to JPY 0.6 billion related to the typhoon. These 2 factors explain the JPY 0.8 billion gap for sales-related variance. ForEx also shifted slightly toward a higher yen.

Please turn to Page 22. This is the Q-on-Q comparison for third quarter operating income. There was a significant sequential decline in both sales and operating income. The JPY 3.3 billion sales shortfall includes the JPY 1 billion negative impact from the typhoon and the JPY 1.6 billion negative impact from lower yields in leading-edge wafers. In addition to this, weakness in smaller diameters in third quarter contributed to the sales decline of JPY 3.3 billion.

The JPY 4.9 billion decline in operating profit breaks down as shown here. Costs increased JPY 2 billion, and depreciation increased JPY 0.9 billion. Within the JPY 2 billion increase in cost, the impact of lower yields in leading-edge wafers was JPY 1.3 billion, and the negative impact of electric power unit prices was JPY 0.5 billion to JPY 0.6 billion. Of the negative JPY 1.4 billion impact from sales-related variance, JPY 0.9 billion was related to the lower yields in leading-edge and the typhoon and the rest comes from weakness in smaller diameters. There was also a negative impact of JPY 0.6 billion from ForEx. All of these factors were negative in third quarter.

Please turn to Page 23. This is the waterfall chart for year-on-year changes to 9-month operating income. On a 9-month basis, sales fell JPY 9.8 billion and operating profit declined JPY 12.7 billion. ForEx also shifted towards a higher yen. With regard to costs, despite the push up in costs associated with the lower yields on leading-edge wafers, on a 9-month year-on-year basis, costs decreased by a net of JPY 1.5 billion. This is a function of the positive impact of rationalization of materials, electric power, labor and maintenance costs. Depreciation increased JPY 2.7 billion. Sales related variance had a hefty negative impact of JPY 9.3 billion.

Earlier, I touched upon various factors such as the impact of the typhoon, product mix and electric power costs, but another major factor was price declines, which had an impact of around JPY 10 billion. 60% of this comes from FST and the impact of spot prices. FST only does business on a spot basis for both 300-millimeter and 200-millimeter, so prices were off significantly. The remaining 40% is the result of slight declines in spot prices for 200-millimeter and 150-millimeter wafers in Japan. There were also positive factors such as an increase in 300-millimeter volumes, particular epitaxial wafers, which had a positive impact of close to JPY 2 billion. The year-on-year exchange rate impact was a relatively large negative JPY 2.2 billion, JPY 1.2 billion of which is related to dollar, yen move. The remaining JPY 1 billion is related to moves in the new Taiwan dollar.

Please turn to the next page. Turning to Slide 24 on the balance sheet. Starting at the top, cash flow was positive, so cash and cash equivalents increased JPY 4.7 billion. Raw materials and supplies fell JPY 6.4 billion. Of this, JPY 8 billion comes from declines in polysilicon inventory. Tangible and intangible assets increased by JPY 8.7 billion, the result of CapEx outweighing depreciation. Interest-bearing debt was unchanged. The capital surplus fell JPY 3.3 billion, reflecting the buyback and cancellation of treasury stock. The JPY 12 billion increase in retained earnings reflects the contributions of profits net of dividend payments.

Slide 25 looks at the cash flow statement. We generated operating cash flow of JPY 63.8 billion, reflecting JPY 60.7 billion from pretax income and depreciation and a decrease in inventories on the reduction in polysilicon inventories. After taking into account the outflow from investing cash flow, free cash flow was JPY 18.4 billion. After reflecting the outflows for dividend payment and the cancellation of treasury stock, this leaves us with the JPY 4.7 billion, which was added to cash and cash equivalents.

Please turn to Slide 27. These are our forecasts for the full year, reflecting the fourth quarter plan. On a full year basis, we are expecting sales to fall JPY 8.2 billion, OP to decline JPY 13.4 billion and net income attributable to the owners of the parent to drop JPY 7.9 billion.

Please turn to the next page. Slide 28 shows the waterfall chart, which breaks out the major components of change for fourth quarter operating income on a Q-on-Q basis. We expect fourth quarter sales to improve JPY 0.9 billion Q-on-Q, mainly on an increase in 300-millimeter wafers. Some of you may feel that the Q-on-Q increase is small given the magnitude of the decline in third quarter, but we have factored in the impact of year-end holidays as well as expected delays in customs clearance around the year-end. As a result, we are expecting a moderate increase in sales. For operating income, we expect a sequential improvement of JPY 0.9 billion. We expect a JPY 1.6 billion decrease in costs, effectively the reverse of what we saw in third quarter. We expect yields in leading-edge wafers to recover and unit prices for electric power to decline slightly. Depreciation is expected to increase as we are invested in facility for leading edge. We expect a JPY 1.1 billion positive from sales-related variance, which is greater than the total top line increase on higher 300-millimeter volumes and, in particular, the mix benefits of a higher proportion of epi wafers.

Please turn to Slide 29. This is the year-on-year waterfall chart for full year OP. On a full year basis, we expect sales to fall JPY 8.2 billion and operating income to decline JPY 13.4 billion. With regard to the JPY 8.2 billion decline in sales, our ForEx assumption is for a higher yen, which will have a negative impact of JPY 4 billion. I will go into more detail later, but we have factored in a negative impact of JPY 12 billion from price declines. On the positive side, we expect volumes to rise for a positive impact of JPY 8 billion. This reflects the fact that 300-millimeter wafers have improved on a year-on-year basis. The breakdown of the JPY 13.4 billion decline in operating income is as follows: as shown in the waterfall chart, we expect cost to improve JPY 1.8 billion, but depreciation will increase JPY 4.7 billion. We expect a negative JPY 6.8 billion from sales-related variance. Earlier, I mentioned that we have factored in a negative JPY 12 billion for price declines on a full year basis.

The split between Japan and Taiwan is roughly 50-50. The Taiwan portion of around JPY 6 billion is all spot driven. In Japan, we expect price declines for 200-millimeter and smaller diameters. In contrast, as also noted earlier, we expect volumes to increase. We have factored in a JPY 5 billion increase, particularly from higher volumes in 300-millimeter epi. We are expecting volumes in 200-millimeter and smaller to decline, but this is outweighed by the increase in 300-millimeter volumes. The expected ForEx impact is a negative JPY 3.7 billion, JPY 2 billion related to a higher yen and a further JPY 1.7 billion from a stronger new Taiwan dollar. This completes the analysis of our forecast.

Please turn to Slide 31. As you can see from the chart, while third quarter was lower, we are expecting to see a sequential improvement in fourth quarter. With regard to Slide 32 on EBITDA, similar to the previous slide, third quarter was weaker, but we are expecting to see a sequential improvement in fourth quarter.

This completes the presentation. Thank you. We will now open the floor to questions.

[Operator]

[Operator Instructions] We will start with Mr. Enomoto of Merrill Lynch.

[Takashi Enomoto, analyst]

My question is about 300-millimeter prices. What is your view of LTA prices versus spot prices for 2021? There is a gap between spot prices and LTA prices. I believe there are some LTAs that will be rolling over in 2021. Is there a risk that LTA prices for contracts set in 2021 will get dragged down by spot levels? On the other hand, there has been some talk of price hikes, so maybe spot prices could rise. What is your view for 2021, 300-millimeter wafer prices?

[Mayuki Hashimoto, executive]

Hashimoto will respond. My impression is that there will be differences in price levels as we have seen before. In terms of LTAs, in our case, many of our customers have extended, so that 2021 is basically covered under LTAs. I think the situation is similar for our peers as well. Given that most customers have already extended, LTA prices will be maintained. For spot, given that 90% of our business is covered by LTAs, really our own exposure to spot is the business in Taiwan. My impression is that spot prices are slightly softer than LTA prices. However, after 2021 or 2022, the market is likely to face shortages, so it isn't clear how far spot prices might fall.

Of course, the customers are aggressive on price now, but I would question whether they can maintain that stance through 2021 and 2022. More fundamentally, the scale of the spot market is not that large. That would be my view. I would also say that the situation is clearly different for epi and polished wafers. PW prices are likely to remain slightly soft for a little while longer, but there are already shortages in epi wafers. Given this, I would expect there would be room for some price increases.

As an example, in the case of a customer with an LTA that wants to procure volume in excess of its LTA volumes, it would probably be tough for that customer to buy the additional volume at the LTA price point. This is my sense of the current situation, particularly in the case of leading-edge wafers. Does that answer your question?

[Takashi Enomoto, analyst]

May I confirm one thing. Are you saying that in your case, there will be almost no LTA renewals next year?

[Mayuki Hashimoto, executive]

Yes, with regard to next year, around 90% is already on LTAs. So if we were to put into place new contracts, we would only have a little room on the epi side. For polished wafers, there wouldn't be a lot of LTAs.

[Operator]

The next question comes from Mr. Azuma of Jefferies Japan.

[Yoshihiro Azuma, analyst]

I have a question about the productivity issue you talked about earlier. If we compare you to your major competitor, from a historical perspective, Sumco is the combination of 3 separate companies. This has made it difficult for Sumco to standardize and unify many processes. On top of this, when you look at product mix, your mix skews heavily to epi, and you have a relatively higher exposure to foundry customers.

Given this, my impression of Sumco is that you are better at producing technically challenging products in smaller lots rather than mass-producing a single product. Obviously, this is on a relative basis compared to your major competitor, but if a customer wants a product that is extremely difficult to produce, then Sumco would be #1. But if the customer wants huge volumes of a standardized product that is not technically challenging, then I feel your competitor is better. From this perspective, how do you approach improving productivity?

[Mayuki Hashimoto, executive]

You are absolutely correct in your characterization of Sumco. We are indeed the result of 3 companies coming together. It has already been close to 20 years since the combination. So I recognize that you could argue that we have had more than enough time to address this. However, each historical entity has different processes, so it is not possible to take a plug-and-play approach. It is true that we have had to make incremental investments to harmonize processes within the company. At the same time, it does mean that we have access to diverse processes. This does give us an advantage in manufacturing leading edge products, although some skeptics have suggested that our skill at the leading edge is pain without gain in terms of profitability.

It is certainly true that we are not in the business of mass-producing huge volumes of the same product. It is also true that this approach does allow you to reduce cost significantly. This is what FST does. What we are trying to do now is to unify our processes, although it will be a gradual transition. Also, if you look at some of the other players like GlobalWafers or Shin-Etsu, they have huge plants located in countries like Malaysia, where there are clear differences in labor costs versus Japan, which is expensive. To address this, we are focused on leveraging AI for processes such as inspection, which are surprisingly labor-intensive. For leading edge, the inspection process is increasingly challenging. So we have set up a dedicated AI promotion department that is looking for ways to leverage AI technology more effectively.

We aim to take advantage of technology to further unify our processes. If you think about it, leading-edge products are not leading edge forever. At some point, a new generation emerges. The challenge for Sumco is to hold on to the high share that we hold in leading-edge products as the technology moves on. To do this, the most important thing is to leverage automation, in my view, as well as to try and standardize facilities.

[Yoshihiro Azuma, analyst]

Understood. So it isn't really about improving polishing to improve throughput?

[Mayuki Hashimoto, executive]

When you say polishing, are you referring to wafer processing?

[Yoshihiro Azuma, analyst]

Yes. In other words, when I compare you to your major competitor, I would view them more as a polished wafer house. So my impression is that they have a highly efficient polishing process.

[Mayuki Hashimoto, executive]

It is probably fair to say that we tend to be better at manufacturing small lots of many different products. If you look at foundries, they are also focused on a diverse range of products. The foundries have particular skills in speedily retooling their fabs to accommodate different products. For us, given the benefits of the technology diversity results from our history as a merged entity, there is no reason for us to shift gears down to adopt a mass production cookie-cutter approach. That would be a retrograde move, in my view, in that we would be spending huge amounts of money to shift to simpler products. Effectively, this runs counter to the differentiating factors that are our strength. Instead, we will continue to focus on the foundry business. If you visit a foundry's fab, their ability to make fast changeovers to their lines is very impressive. I believe that there is more that we can learn from the foundries on this point. This will continue to be an area of focus going forward.

Currently, we are at the worst part of the cycle for us, where profitability is challenging. Normally, the pace of transition would be a little slower, allowing us to generate better efficiencies that would support profitability. However, recently, we have seen that the transitions have become much faster, not just for some foundries, but across the board for customers that do leading edge. To make this happen, we are under pressure to provide significant volume in short time frame. For better or worse, we are effectively the only supplier. In this type of a situation, the risk for us is that we become the reason for a product stoppage. So we have no choice but to do the best that we can to avoid this. We believe that the ability to respond to our customers' needs translates into market share and profits in the future. That is what we did this time around, but it just so happened that the timing was suboptimal. If we did not produce leading-edge product but focused only on commodity products, it might be that we would have reported better profitability in a year like this year.

[Operator]

The next question is from Mr. Ikeda of Goldman Sachs.

[Atsushi Ikeda, analyst]

Shipments of 300-millimeter wafers have recently been at around the 6.7 million or close to the 6.8 million level. My impression is that demand has gotten stronger. How do you think customer inventory levels will trend going forward? Do you think that we might see customers slam on the brakes at some point? Also, it appears that shipment levels are getting close to capacity levels. As device makers increase output or wafer input levels, do you have a sense of urgency in terms of addressing future capacity issues as inquiry levels rise?

[Mayuki Hashimoto, executive]

On polished wafers, there are 4 major customers. If you look at their inventory levels, it is a mixed bag. Our customers that are preparing to ramp up new fabs do not have much inventory, but others appear to be carrying significant inventory. The latter are aggressive in terms of negotiating price, although we ourselves don't do much in polished wafers. However, for epi, I think we are already seeing a clear shortage. With leading-edge ramping rapidly and logic trending so strongly, I feel that there may be a shortage. It doesn't appear that customers are building up inventory.

Following the Huawei issue earlier this year when orders were temporarily paused, when the restart happened, inventory was worked down very rapidly. For leading edge, it appears that customers are now competing to lock in supply. In particular, it appears that iPhones are selling well, and the shift to 5G is proceeding smoothly. Wafer consumption increases as we migrate to 5G. Even if there were to be a correction in memory, I think it's unlikely there will be a correction in logic since chip numbers are aligned with handset volumes. So I think we will see a significant gap between the strength of demand for polished wafers versus epi. Does that answer your question?

[Atsushi Ikeda, analyst]

So tight epi demand should drive price hikes. Is that correct? If we look at your margin relative to your peers, it is very low. Given the limited number of players that can do leading-edge wafers, is improving margins not a key management objective?

[Mayuki Hashimoto, executive]

We recognize that our margin has been lagging that of our peers recently. When we analyze the situation, we have found that margins on polished wafers have been high. In the past, when we had higher margins, it was the case that margins on epi were higher than polished wafers. Given that the added value for epi wafers is higher, the margin should be higher. At the moment, the situation is reversed. We are focused on addressing this issue and feel that this is the right timing to address this issue.

[Atsushi Ikeda, analyst]

So you may be able to raise prices for 2021 and beyond?

[Mayuki Hashimoto, executive]

Yes. Obviously, it is not just up to us to decide unilaterally, but we will certainly be focused on doing our best to make it happen.

[Operator]

The next question is from Mr. Miyamoto of SMBC Nikko.

[Go Miyamoto, analyst]

I would like to ask about memory demand. Earlier when you were discussing the market environment, you indicated that while demand had not fallen, visibility is not great. You cited inventory levels as one reason for why visibility is not as good for memory as it is for logic. Other than that, if you look at the end markets, what do you see as factors for uncertainty? If we were to see a rebound in data center CapEx, would you expect that to support visibility into a full-fledged recovery in memory, please provide more color?

[Mayuki Hashimoto, executive]

One positive would be the expected increase in data center investments. Also, the slack from the absence of Huawei will be taken up by others. So unlikely that there would be a significant decline in end product volume. On the other hand, data centers are not the only heavy consumers of memory. Smartphones also use significant memory. The question is what happens here? For now, it appears that handset memory density has not declined. So I would expect that we would continue to see growth.

The other concern would be that a weak macro backdrop could hurt smartphone handset sales so that even if consumer take-up of 5G is good, they may offer handsets that use less memory and park data in the cloud. If that happens, then memory growth may not be that strong. If the question is, will memory growth be much stronger than logic, I wouldn't think so. My impression is that memory will not grow as much as logic. If your question is, will volumes increase year-on-year versus 2020, then yes, I would expect memory volumes to increase year-on-year. So in other words, in the near term, the situation is tough to read because of customer inventory levels. But in the medium term, you are comfortable that there will be growth in 2021.

[Go Miyamoto, analyst]

Do you think DRAM will grow more than NAND?

[Mayuki Hashimoto, executive]

Is your question about what will happen with DRAM and NAND?

[Go Miyamoto, analyst]

Yes, do you think that DRAM is likely to show stronger growth than NAND?

[Mayuki Hashimoto, executive]

That's what they say because DRAM is used in data centers. However, NAND can be quite volatile. Depending on which market research firm you believe, there are those that say DRAM won't grow much, but that NAND should grow. It depends on how you segment the categories. If you think data centers will grow, then the natural follow-on would be that DRAM is strong.

[Operator]

Next is Mr. Okazaki of Nomura Securities.

[Shigeki Okazaki, analyst]

I would like to confirm the factors that are depressing profitability for leading-edge wafers? In terms of factors going into the downward revision, you cited increased cost of JPY 1.3 billion in third quarter. You also highlighted JPY 0.3 billion in sales-related variance for a shortfall of JPY 1.6 billion. How much of each of these factors have you assumed you will get back in fourth quarter? Also, how have these factors trended so far in the quarter? And what are your expectations for the monthly -- on a monthly basis?

[Michiharu Takii, executive]

Takii will respond. If you look at Slide 28, you can see that we expect a positive sequential impact on cost of JPY 1.6 billion. This includes the impact of the JPY 1.3 billion that you alluded to in your question. The positive JPY 1.1 billion in sales-related variance includes JPY 0.3 billion you referred to in your question.

[Shigeki Okazaki, analyst]

In other words, you expect to fully recover these 2 items in fourth quarter?

[Michiharu Takii, executive]

Yes, we expect the issue to be fully resolved.

[Mayuki Hashimoto, executive]

If, I, Hashimoto, can jump in. We have been able to identify the issues through the debugging process. I would say that it wasn't we didn't know what the problems were, but simply that we were still in the early part of a learning curve in terms of manufacturing. Poor yields not only push up cost but eat into capacity that might have been used for other things. So it's a double whammy. That was why we were struggling. The sad part of being at the forefront of technology is that you are in uncharted territory, and there are many unknowns at the outset. There are many things that the customer themselves does not know. Through a reiterative process, we have been able to largely address the issues. So in fourth quarter, the focus will be on efficiency.

[Shigeki Okazaki, analyst]

As an add-on, when we think about the next transition to 3 nanometer, should we expect to see a similar pattern?

[Mayuki Hashimoto, executive]

If the ramp is very rapid, then possibly, yes. It sounds like we are making excuses. But up to now, we have seen the transitions to 16-nanometer, 10-nanometer and 7-nanometer over time. In previous generations, the transition had been slower, which gave us time to get up the learning curve in terms of wafer manufacturing. More recently, however, we have seen transitions that play out over a much shorter time frame. So it is very tough to keep up. In addition, we have increasingly been the only player that is certified in the early stage. So if we cannot provide wafers, the customers will have an issue. To avoid causing problems for our customers, we devote significant efforts to meeting demand, even though we are still at an early stage of the learning curve.

However, it depresses capacity and boosts costs, so it is a suboptimal situation for us, but it allows us to win the trust of our customers and to enjoy a high market share in leading edge. If we can hold on to the high market share we capture at the outset, there should be a significant improvement to our situation in the future. This is the expectation we have. So for now, we feel we should be hunkering down.

[Shigeki Okazaki, analyst]

I hope you will be successful in reaping additional rewards from your customers for your efforts in the future.

[Mayuki Hashimoto, executive]

Please continue to support us.

[Operator]

The next question is from Mr. Yamada of Mizuho Securities.

[Mikiya Yamada, analyst]

With regard to Slide 17 and the demand forecast for data centers and servers, where you show DRAM demand increasing with the shift to HBM, should we expect to see even more shortages in epi furnace capacity in second half of 2021 after PCI Express 5.0 is supported? If so, should we expect you to need to invest over a relatively prolonged period of time? If so, I would think there would be serious shortages in epi in 2022. Is this the right way to think about it?

[Mayuki Hashimoto, executive]

My answer is very simply, yes. In 2021, we will need to improve productivity significantly or I am concerned that we may not be able to meet our customers' requirements. This is a serious concern. We had originally expected epi to grow and had already made additional investments, but the growth appears to be exceeding our assumptions. I think everyone is in a similar situation near term. I think we all have no choice but to increase output.

[Mikiya Yamada, analyst]

Doesn't that mean that everyone will need to invest in more epi furnaces.

[Mayuki Hashimoto, executive]

Certainly, there will be a need to install more epi furnace capacity over time. But even if you make an investment decision today, the capacity will not come online for 18 months. As you know, we have top share in epi, so we have been making preparations.

[Mikiya Yamada, analyst]

So you effectively have things in hand?

[Mayuki Hashimoto, executive]

Yes.

[Operator]

Next is Mr. Watabe of Morgan Stanley MUFG Securities.

[Takato Watabe, analyst]

I have a simple question. With wafers on the verge of hitting a new high at 6.8 million wafers, why do you think your quarterly profits are stuck at around the JPY 7 billion level. The gap between you and the top player is about JPY 30 billion per quarter. How do you view this? And are there initiatives to help you narrow the gap?

[Mayuki Hashimoto, executive]

This is a question we always get asked. What I always say first is that there is a difference in terms of history when you compare us to Shin-Etsu. Shin-Etsu has benefited from 30 to 40 years under the brilliant leadership of Mr. Kanagawa. When he first took the helm, Shin-Etsu was a very different company from what it is today. You can ask anyone in any industry that goes up against Shin-Etsu, like a major chemicals player, and they won't have a neat answer. Clearly, Shin-Etsu has benefited from superior management. I have read all of Mr. Kanagawa's books. What I find most impressive is that he states the obvious, and he follows it through to its logical conclusion with discipline. It is even more impressive when you think about how consistent he has been over a very long period of time. So that is one difference, although you could argue that, that is simply an excuse.

Looking at what is different now, Shin-Etsu's product lineup is different from ours. They do thin-film SOI and FZ wafers, so they generate a high proportion of revenue from such high value-added products. This is different than us. We are not involved in these types of products. Additionally, these value-added products has seen a surge in growth recently. The second reason is Shin-Etsu takes a cookie-cutter approach and excels as mass production of a single product. Their approach to customers is also noteworthy in that they choose to rely heavily on certain customers where Shin-Etsu holds a dominant share. In contrast, there are customers where Shin-Etsu has a minimal presence. FST takes a similar approach. This approach makes it possible to lower costs significantly. Unlike us, Shin-Etsu does not have a lot of different processes. They choose to use a single process, which unfortunately, I believe, does have implications. To compete with Shin-Etsu, I believe we need to leverage our strength focusing on technologically challenging leading-edge product, producing many different products in small lots as efficiently as possible in order to generate profits.

[Takato Watabe, analyst]

From that perspective, you would need to change product mix or high prices to grow profit significantly, is that correct?

[Mayuki Hashimoto, executive]

Yes.

[Hiroshi Shibaya, executive]

Thank you. We will end the briefing here. Thank you for your participation.

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