May 29, 2018 Medium-term Management Plan Briefing Q&A

Because we do not comment on specific figures concerning growth rates or amounts for each application, please refer to the graph on page 14 of the presentation materials for the approximate trends. Until now, greenfield investments in DRAM have been strong, but for CY2019 we believe that investment in 3D NAND will grow.

The demand for memory is particularly strong, and memory customers have been investing strongly in DRAM since H2 CY2017, and the increase in production capacity is proceeding. From now on, we expect that investment by memory customers will switch from DRAM to NAND.

Of the investment in China for CY2020, we see that investment by foreign manufacturers and Chinese manufacturers will be about 50/50. Most of the investment by Chinese local manufacturers is for memory.

As you are aware, we expect the main scenario to be $62B. There is confidence that the WFE market will continue to grow even larger, and we have set an operating margin target based on the development investment and capital expenditure to prepare for that. In the $55B case, the model is calculated including the assumption that the market will fall to some extent due to the risk of a financial crisis, geopolitical risk, or depending on progress in the miniaturization of semiconductors.

While a $50B case is also possible if there were to be a financial crisis, geopolitical risk and/or delays in miniaturization from the roadmap, Tokyo Electron (TEL) is making its investments according to the scenario of cases that exceed $62B. The largest decrease in sales for TEL was the 44% YoY drop at the time of the 2008 financial crisis. With this kind of decline in mind, we are building a system that can respond to the risk of market fluctuation by introducing the outsourcing of production to our partner companies and employee compensation based on the overall company performance.

We are afraid that we cannot comment on whether or not the target ROE values factor in any buybacks. As we have said in the past, we will flexibly consider share buybacks. We would like to make a decision from all points of view including market trends and the scale of share buybacks, and carry out share buybacks at the time that it brings the most value for shareholders. We believe that cash that is generated should be used in a form that can produce the maximum value, and should be used in an agile manner based on the possibility of securing capital for additional business chances when there is growth potential.

As shown in the graph on page 62 of the presentation materials, SAM growth is the same as the WFE market growth, and we see that the SAM ratio will remain constant. Although there will be no decline in the SAM ratio, there will probably not be a large increase in it either. In the lithography processes for which EUV is introduced, the number of exposures will fall to about 1/3. On the other hand, the productivity of EUV is lower compared to that of the current ArF immersion, so the number of EUV exposure tools and coater/developers required will increase. We believe that these two factors will offset each other. We see coater/developer unit prices rising slightly because they will be equipped with units for defect reduction and CDU*3 improvement for EUV. If there are changes to the productivity and exposure dose of EUV, the story will be different, but at this point in time our outlook is as shown on the graph. Business opportunities will increase starting in CY2022 for features such PSCAR™*4 to improve productivity for generations from N5 onward.

I would like to ask about the SAM ratio for etch systems in the WFE market shown on page 71 of the presentation materials. What is the background for seeing only slow growth from CY2018? In addition, it is said that Micron will continue to use multiple exposures without using EUV. Would you see an upside for the SAM ratio if comes to this?

The SAM ratio growth remains slow because the number of exposures in litho etch patterning processes after quadruple patterning decreases to one by introducing EUV, decreasing the number of patterning process steps. At this point, EUV is forecast to be introduced first in logic, and because it is not clear when it can be introduced in DRAM, it is not included in our forecast. There will be a sufficient possibility of changes to this forecast based on technological developments in the future.
From CY2017 to CY2021, the outlook is for an increase in 4 points in the share of the exposure tool market in the overall WFE market due to the introduction of EUV. Etch and deposition systems will both grow by 1 point in the same time period.

In the current ArF immersion, it is necessary to repeat exposures many times. Also, self-aligned multiple patterning has overlay issues when you repeat it several times, which decreases the yields. Therefore, from a technical viewpoint, the introduction of EUV is probably inevitable in order to advance miniaturization further. However, we do not expect a large drop in business opportunities for our etch and deposition systems.

It will be difficult to use EUV in all patterning processes for cost reasons as well as technological reasons. Because we see that the conventional double/quadruple patterning will continue to remain in light of the progress of miniaturization, we do not believe that the deposition market will fall greatly.

In CY2017, our share of the deposition system market grew as expected for ALD and single-wafer CVD systems. As you point out, our share in the oxidation/diffusion market fell, but this was not because we actually lost competitiveness. The background of the drop was that our share in the market changed due to the investment mix of customers, because it is not the case that we have the same share for all customers. We are not overly worried about the fact that our share fell. We plan to continue to increase our share by adding even more value.

The POR that we talked about last year at the Medium-term Management Plan briefing was a development POR, and that became a high-volume manufacturing POR.

As you point out, non-volatile memory devices such as MRAM and PCRAM contribute to addressing power consumption issues, and these will probably be utilized in the future. However, the NAND composition rate to WFE market is the largest among all non-volatile memory, and otherwise MRAM and others will be introduced somewhat later. In fact, the introduction plan itself has been pushed back longer compared to a year ago.

The ONON (oxidation film/nitride film) layers, in other words the charge trap structure, has become the mainstream approach.