Earnings Call: Q4 2026

May 13, 2026

Summary

FY2025 sales grew and operating profit returned to positive territory, but SiC impairments drove a large net loss. FY2026 guidance targets higher sales and profit, supported by reforms and AI-server/SiC growth, amid cost and market risks.

Good morning, everyone. This is Azuma. Allow us to deliver our presentation while seated. Let me start our financial results. First of all, our FY 2025 full year results. Net sales ended 7.3% plus year-on-year at JPY 481.1 billion. We ended with a large operating loss last year. This year, we achieved an operating profit of JPY 10.8 billion. Since we posted big impairment losses, net profit ended in negative JPY 158.4 billion. This fiscal's plan for FY 2026 is to achieve a 6% year-on-year growth of net sales, reaching JPY 510 billion. Operating profit to grow largely by 176.1% to a level of JPY 30 billion. Net profit is planned to turn positive to a level of JPY 29 billion. In FY 2025, we saw the gold prices going up. Yet, since we increased our sales and reduced our depreciation and fixed expenses, we were able to turn our operating profit into positive figures. As we posted impairment losses mainly on fixed assets in the SiC business, the net loss widened significantly. In FY 2026, we are starting to see a positive impact from these structural reforms. On the other hand, the impact is being limited due to tensions in the Middle East and the soaring gold prices. In line with the impairment loss in FY 2025, we have been able to reduce our depreciation largely, but still due to these situations, our OP is planned at a JPY 30 billion level. Talking about inventory, which is always being told as high, we have been able to destock mainly the raw material inventories. We are targeting to bring the turnover to below six months by the end of FY 2026. About dividends. We are planning to keep on paying JPY 25 per share for the term end. Together with the interim dividend, it will be an annual total of JPY 50 in FY 2026 as well. For business and management integration, or let's call it the three-part integration, the due diligence has kicked off. Mitsubishi Electric's power device business, Toshiba D&S Semi business, and we at ROHM are currently discussing for a possible power device business integration. This shows our previous FY 2025 full year sales. The consumer market segment, mainly amusement, grew the most by 13% year-on-year. Automotive up by 5.8%, industrial up 7%. Each market segment except for communication has been growing. Although we are not seeing major changes, computer and storage market segment is an area we will have to grow more. Going to the segment breakdown, LSI or the ICs went up 7% and discrete up 9.7% year-on-year. We grew except for the modules. Our intention is to continuously grow both the IC and discrete semi device businesses. Looking at the regional customer nationality split, Japan grew by 7.6%, which was big. Europe and other Asia showed a bigger growth rate, but in terms of the absolute value, Japan has been growing outstandingly. We still haven't achieved our target of having a 50% plus overseas business mix, and that's something we are wanting to do. This shows how we were able to recover our operating loss of JPY 40 billion to an operating profit of JPY 10.8 billion. First, our sales increased by JPY 32.7 billion. Organic increase was JPY 35.6 billion, but we were negatively affected by the FOREX by JPY 2.9 billion, thus becomes JPY 32.7 billion. Automotive market segment grew by over 5%, equal to JPY 13.1 billion. Consumer up by JPY 12.1 billion. These two showed a large increase. Talking about the negative trend, in line with the increase in sales, the variable expenses increased. The other is the gold prices. Last fiscal, we were estimating a level of JPY 17,000. Currently, it is going beyond JPY 20,000. That's another factor burdening the variables. For the SiC devices, due to the expanding adoption for automotive inverters, as we mentioned during the third quarter briefing, we posted quality assurance related costs in the third and fourth quarters. These add up to a negative impact of JPY 10.9 billion. Moving to the fixed cost. We have organically reduced this by JPY 31.3 billion. Since we recorded an impairment loss and we also changed the depreciation method from the declining balance method to the straight line method, these have helped us reduce the depreciation largely. Further, as the R&D spend for our SiC has peaked out, we were able to shrink the fixed cost by JPY 29 billion. All in all, this leads to an operating profit of JPY 10.8 billion for fiscal 2025. Let me explain why we ended with a big net loss. First, the previous fiscal ended with a net loss of JPY 50 billion. We saw a positive impact of JPY 50.9 billion from the OP. For the non-operating part, the previous fiscal, we had the dividend income, but this time no, and thus ended with a negative JPY 2 billion impact. On top of that, we recorded an impairment loss. Here we see a negative impact of JPY 163.1 billion. There is a positive JPY 5.8 billion impact from income taxes. But in total, the full-year net profit resulted in a loss of JPY 158.4 billion. This is more about the impairment losses on fixed assets. The left side describes the macroeconomic changes touching on the slowing growth of battery EVs. The battery EV market forecast peaked out in 2023, and we have been revising down our forecast since then. Recently, the U.S. market is shifting back to ICE or petrol cars, and Japanese OEMs are delaying their battery EV developments. In the midterm plan we made in fiscal 2025, we factored in these market situations and made our SiC business sales target to be more solid. Having this as a baseline, when making a forecast on how the market will grow beyond 2028, we anticipate our investment return to be slower, thus decided to post an impairment loss. Talking about business environment surrounding ROHM, it is true that excess upfront investment had to be made, and at the same time, we are seeing the rise of Chinese SiC devices. In the past, our sales in China were big, but due to aggressive price down pressures, our share is declining. Our SiC substrate business was having around an annual sales of JPY 10 billion, but now that the 6-inch substrates are not selling so much. Although we are continuously working on the 8-inch, we are seeing a shrink in the 6-inch business, and that is another reason why we decided to record impairment losses. This shows our fiscal 2026 full-year sales forecast. We expect the automotive market segment to grow by 5.6%, 5% with industrial, and 6.8% with consumer and others. The automotive production volume is pretty stable. Although the SiC substrate external sales business is shrinking, we expect the sales of SiC devices to grow largely amongst the European and Japanese customers. That is why we guide this market to grow positively. The industrial market segment was slow for some time due to a prolonged inventory adjustment, but that has largely been solved, and we are seeing a recovery in the FA segment. Thus, we expect this market to grow too. Going to consumer and others, there will be a slight decrease in the Chinese white goods market due to the end of subsidies. On the other hand, sales in the amusement segment remain strong. The market is expanding and adoption in the server and storages is growing. Therefore, we expect a positive growth here too. Fiscal 2026 plan hasn't changed largely from fiscal 2025. Left end shows the market breakdown. Computer and storage is expected to grow largely by around 20%. Automotive and industrial are to grow around 5% plus each. Now into the segment breakdown. The discrete semi devices are expected to go up by 8.4%, and the ICs and discrete are the two segments accounting for the major portion. Going to the customer nationality breakdown. I mentioned earlier that our target is to make our overseas sales to over 50%. For this fiscal, sales in Japan is to be 53.2% of total. Domestic sales are still higher. Other Asia, mainly Taiwan, is growing, as well as the Americas. Yet, Japan's sales mix is still accounting for more than half. This waterfall chart shows how we plan to achieve JPY 30 billion of operating profit in fiscal 2026. The starting point is JPY 10.8 billion on the left end. We expect our sales to increase by JPY 28.9 billion, including a positive impact from the FOREX. Organic growth will be JPY 24 billion. Amongst different market segments, we expect growth in automotive and computer and storage. In consumer segment, including amusement, is to grow, but automotive and computer and storage are to be growing even bigger by over JPY 10 billion each. Now, impact from material cost and inventory. We will see a negative impact of JPY 11 billion from material costs, while inventory adjustment is improved by JPY 1 billion, and in total, a JPY 10 billion negative impact from the variables. Impact from the fixed cost might appear to be not big enough, where in fact we recorded an impairment loss. We only expect a JPY 300 million positive impact. In line with the impairment losses posted in fiscal 2025, depreciation costs will go down significantly, but due to tensions in the Middle East, we estimate the materials and transportation costs to go up by over JPY 5 billion. Thanks to the structural reforms we are doing, we expect a positive impact of around JPY 15 billion. Yet, we still see various expenses going up. Although we are improving our savings, these will unfortunately be offset by the soaring expenses. Thus, our efforts in reducing the fixed costs end up to be just around JPY 300 million. Based on these factors, we believe our JPY 10.8 billion of operating profit is to go up to JPY 30 billion. This is about our CAPEX. During our first midterm plan, we made excessive investments. Our CAPEX spend reached to cumulative JPY 608.2 billion during the course of five years. The next midterm plan period is three years, and we plan to invest approximately JPY 150 billion in total. The initial year will be slightly high at a level of JPY 60 billion, but if we level the JPY 150 billion in three years, it will be roughly a JPY 50 billion of CAPEX spend every year. Our inventory levels have been high for some time. The WIP or Work-In-Progress inventory level is still slightly high, but our intention is to bring down the finished goods level to 1.2 to one-month level by the term end. For the WIP, since we have been able to achieve a cost down, we are being able to reduce the absolute value, but not the turnover months. We continue to de-stock and shorten the turnover months for the WIP. The raw material inventories were at a three-month level during its peak, but we have been de-stocking to a level below two months. We believe we can bring it down to an adequate level. Talking about shareholder return. Since our free cash flow has turned positive in fiscal 2025 and same to be expected in fiscal 2026, we intend to keep our annual dividend at JPY 50. Payout ratio is to be 67%, the fiscal end same with the total return ratio. We have been able to continuously achieve our target of over 30% of payout ratio. From this part, let me talk about our midterm management plan and structural reforms. The left side remains same. In the three-year period, we target to achieve over JPY 500 billion of net sales, over 20% of OP margin, and over 9% of ROE. The non-financial targets remain same, too. We keep on reducing GHG emissions and use more renewable energy. This shows how we plan to generate JPY 100 billion of operating profit in three years. As the letters in bold describe, we are reorganizing our manufacturing sites, discontinuing unprofitable products, doing price hikes, switching from gold to copper wires. Our customers are reviewing these now. Thus, these initiatives will start to contribute from fiscal 2027. The letters in red are levers that have been accomplished. We optimized power procurement costs and made a major improvement. Improvements in productivity and logistic costs are progressing, too. Talking about gold bar prices, so far, we have been able to improve by around JPY 4 billion. But compared to last year, prices keep on increasing, and we have seen it go up by JPY 13 billion. We are speaking to our customers, and rather than asking to have the cost passed to price, we are negotiating with them to consider switching from gold to copper so we can see a major improvement. We have been sending these samples using copper to our customers, and therefore, we expect to improve the situation by switching to copper wires. Now about sales growth. Previously, we forecasted our sales growth from AI servers to be very small, which disappointed our stakeholders. This time we believe this will grow steadily. More will be explained later. This is about reorganization of the manufacturing sites. Before we reported about the closure of our IC plants in Kyoto and Okayama. As you can see on the right end, this time we have accomplished the agreement to transfer the equity of Dalian Assembly Plant. This reorganization is not limited to Dalian. We are working on other sites too. As I always mention, the most critical thing is to maintain stable production. I hope you understand that we give you updates once things have been decided. Let us now shift to SiC and AI topics. I will hand it over to Mr. Ino and Mr. Takeshi. This is Ino. I will be covering our SiC business strategy topics. Here is the SiC sales trend. The black bars show the SiC business total, which includes the substrate sales. Fiscal 2025 sales turned out to be most likely equal to what we forecasted and disclosed a year ago. There has been some changes in the mix. The substrate business declined both in volume and sales last year. On the other hand, our device and module business grew, and in total, we achieved our forecasted sales. The red bars show the sales trend for automotive inverters. Our fiscal 2025 figures grew as we had planned. In fiscal 2026, we target to hit over JPY 50 billion of sales, and the breakdown is indicated on the right. We expect to see over 30% of SiC sales growth this year. The substrate business is to decline this year too, whereas the device and module business to be growing more, over 55%. In total, including the substrate, we expect to see an over 30% growth. The red bars represent automotive inverters, which is for sure driving the growth. In the SiC domain, we are starting to see growth for AI servers, and this fiscal, we plan to achieve 2.5 times sales compared to the previous fiscal. The absolute value is still small. Thus, we will start to see its contribution towards 2030. Currently, the market is centered on silicon, but we believe the use of SiCs for AI servers is to be growing. As we issued a press release the other day, we have been subsidized by the NEDO for the 8-inch development, and the project has been completed two years ahead of schedule, and we start its mass production this fiscal. Although not written here, we will see the contribution of the mass production of the 5th-generation products on a full-fledged basis from this year. This slide elaborates on the traction inverters. The left end chart shows the battery electric vehicle production volume market forecast, as we all know. There was a peak from the end of 2023 up to around January to March 2024. After that, as the red dotted line shows, the forecast has been declining recently. The current main market is China, but moving forward, we know that the EU and Americas will be ramping up. The right hand shows not the markets where the products are sold, but a breakdown of where the OEMs are based in. The arrows in the center show our customer volume forecast based on the confirmed businesses that we have been awarded. Fiscal 2025 was around 1 million units of inverters, and we forecast this volume to go up to a 3 million unit level in fiscal 2028. There will be some changes in the mix. But we indicated half year ago in our midterm plan that we expect the volume to triple. Lately, we still see this forecast to be maintained as it is. For fiscal 2029 and onwards, as it says in the left end chart, the growth rate is to be dropping as we are considering the trend to be on the lower range. For fiscal 2026, the right side charts show China, EU and Americas, and Japan/Korea. The current market is skewed towards China, let's say around 65%. Up to fiscal 2025, majority of sales came from China. From fiscal 2026 and onwards, we expect the sales in non-China areas to start showing up. That will help us to sustain the business without depending on China. As of 2028, even excluding China, we will have our business being well-established and still will be able to add the China business portion on top of it. In that sense, 2026 will be a big turning point for us. What is more important for us is to improve our profitability. Our target is to achieve breakeven and generate profit in fiscal 2028. Compared to what we showed in fiscal 2025, we have come up with a much more detailed profitability improvement roadmap. The first is to improve the SiC substrate business. The 6-inch in-house production will be downsized so that we can concentrate on the development and production of the 8-inch. We will switch more from make to buy for substrates. For the in-sourced 8-inch production, we will concentrate on further improving cost competitiveness. For the devices, there are three main initiatives. One is to improve yield, and this is important, as you can see on the right. Last fiscal, during the third quarter, we saw a temporary decline due to quality-related issues, but then it recovered in the fourth quarter. In line with our midterm plan, we intend to continuously improve the yield. For the 5th-generation MOSFET, development has been completed and a full-fledged mass production and shipment will begin this fiscal. By largely reviewing the way to build the device structure, we have been achieving better yield performance from the beginning compared to the fourth gen products. We will increase our mix of gen 5 products as they show higher performance on smaller chips. Of course, profitability too is better with gen 5. As shown on the bottom right chart, we will be increasing the 8-inch, and amongst the 8-inch work more on the gen 5 products to reduce cost. Another thing is to increase volume. By putting focus on the profitable gen 5 products and modules, we will improve our portfolio and profitability. In this way, we plan to hit the breakeven and go farther beyond. From this part, I, Tateishi, would like to talk about our AI server business and its progress. This is about the data center CAPEX and demand forecast. As you may all know, the data center-related CAPEX is growing at a very fast speed at a 40% CAGR level. It became double from last year to this year. Moving forward, it is forecasted to grow at a range of between 1.5x and 2x. Looking at the demand of power devices used for AI servers, which also includes the DrMOS, which is a MOSFET, I will elaborate on it later. From 2025 up till 2030, we foresee a CAGR of 48%. That means the demand will become 1.5 times bigger every year. Considering the power used for server racks, we calculated the volume of power devices needed, and that is how we at ROHM came up with this forecast. This shows an existing server, a typical one from sometime back in 2023. The power consumption is 13 kilowatts. When we look at the next gen AI servers, the power consumption becomes 1,000 kilowatts, which is 72 times bigger. Based on this, we estimate the count of power devices and analog devices to be used. The amount of power devices used will be 36 times more and 56 times more with analog devices. Hereby, we expect this market to grow significantly. When thinking about how to approach this market and about ROHM's path to success or competitive advantage, we have the silicon MOSFET. As Mr. Ino mentioned earlier, we have the SiC coming into AI servers. There is the GaN device and MOSFET. As part of the controlling technology, there is the analog technology. The nano technology has been promoted for some time. By combining the power and analog technologies, we at ROHM can offer all of the semiconductors that are needed for a power system. As Asia's only one-stop manufacturer, we collaborate with leading players in the AI server market. As we have already announced, we collaborate with Nvidia on the 800-volt power supply system. As part of the actual implementation with Delta, we collaborate in developing and making the power supply system together. This is about the DrMOS I mentioned earlier. It is also known as smart power devices. When we think how this grows back in 2023, the GPU was about 700 watts. This is expected to be 5.2 times bigger in 2027 to a level of 3.6 kilowatts. The final power device to supply this is the DrMOS. Currently, around 20 units are used, but this is to be increasing to around 50 units per GPU. That means 2.5x. To sell the DrMOS, you need the MPC multi-phase controller. Unless you combine this, you wouldn't be able to sell DrMOS. These always come in a combo. For the current GPUs, the solution does already exist and are being sold. Currently, we are developing the second generation DrMOS. The device standalone performance appears to be -40%. Actually, the lower the resistance per area is, the better. We have succeeded in improving the resistance by 40%. The evaluation results of performance are depicted on the right hand. Compared to the first generation products industry standard, we can improve the loss by 0.5%. This might sound like a small change, but actually means a huge thing. Today, a server rack consumes around one megawatt of power, and if we improve the loss by 0.1%, it means we save one kilowatt. Improving the loss by 0.5% means we save five kilowatts per server rack. This helps reduce power cost and on top of that helps reduce the cost for the systems to cool down the racks. Even if the semi standalone might not have an outstanding performance, there is a very strong requirement to improve the loss. Since we have most likely completed the development of this technology, the DrMOS part of the business has been included in our sales forecast this time. This shows the sales target for server businesses. FY 2025 was JPY 17 billion, and we plan to achieve around JPY 25 billion in FY 2026. We intend to keep on growing to hit JPY 100 billion in fiscal 2030. The mix by this time will be to have approximately 30% of DrMOS. Other than that, now that we see an increase in SiC sales, there is the SST which converts the AC power supply to DC. The transformers will be reduced whereas the more SiC devices will be used and that is the trend we are seeing today. SiC usage is growing due to its strong properties. When the power voltage eventually goes down to a level of 800 volts, that is when the GaN devices are needed and when it goes slightly lower, the silicon devices, then finally DrMOS and the power devices. In addition, the MPC multi-phase controllers to control these will be required. Combining all these will create a synergy in the sales amount and thus we have revised our target from the previous JPY 30 billion to a level of over JPY 100 billion. We would like to move to the business and management integration topics. There was a share acquisition proposal from DENSO. We at ROHM established a special committee composed of independent outside directors and others and discussions were carried out since then. Whilst a number of conversations were made with the DENSO management and business operation people, DENSO side informed us that they pull out their proposal of acquiring 100% share of ROHM. Therefore we at ROHM concluded and stopped the reviews at the special committee. DENSO mentioned that they decided to withdraw the proposal after a comprehensive review, but at the same time mentioned that they are wanting to further advance the collaboration with ROHM in the analog devices. They also stated that for them to roll out their analog technologies for the automotive to other market segments like consumer and industrial, it is essential and effective for them to partner with ROHM. That is why we would like to further progress our co-creation activities with them. We will continue discussions on the potential collaborations based on the existing strategic partnership. Our focus with them will be the analog ICs, but we will not limit that to automotive but include consumer and industrial market segments and make an effective use of the IPs that both companies have. We will continuously sell to the automotive industry, but at the same time expand to other fields. Although we had personal exchanges in the design review phases, we will further deepen this kind of personnel exchanges by sending our human assets to each other. Although it will not be 100% acquisition of shares, the business with DENSO will help both to widen the scope of business areas. Therefore, we will further strengthen and deepen our collaboration with them. Regarding this topic, well, there are many things I cannot easily mention. The due diligence just started off, and the three parties are working positively towards an integration. The key message here is that this possible integration is not at the opposing end of the DENSO's proposal. The reason is that the three-party collaboration is focusing on the power devices. Toshiba and ROHM are wanting to integrate overall, because if we can complement each other and strengthen the power device, we can contribute to DENSO and other automotive parts manufacturers by offering quality power devices. For analog devices, we collaborate with DENSO to expand our business beyond Japan and also to other domains like industrial. Thus, these two topics are something that can coexist. Some people might be thinking this three-party integration to be an anti-takeover measure, but that is not the point, and that is what I would like to stress here. Before the due diligence kicked off, I remember saying that we can get back to you with the progress and outcome by sometime in summer. Our overall target remains same, wanting to form a power device business that will be globally competitive. But since we have to talk more about our human assets, plans, and other topics, it might take some more time to be able to give you some updates. Yet, I will try to share with you more as early as possible. Last but not least, about our capital policy, let me hand it over to Mr. Kenevan. I, Kenevan, would like to cover this topic. As President Azuma said, while the discussions for business integration with Toshiba and Mitsubishi Electric are ongoing, at the end of the day, if we reach an agreement, we might have to slightly revise our capital policy. Regarding cash generation, we will continue to recover our business performance, reduce working capital through cash conversion cycle improvement, optimize non-operating assets, and sell investment securities. These remain same as before. We will only use our capital for things needed for our businesses. Looking at these strategic investments, if we realize a strategic investment opportunity and see a necessity to do financing, we might expand the investment capacity. Having said that, in terms of shareholder return, we keep our target of maintaining over 30% of dividend payout ratio. Talking about liquid funds, we were thinking of doing a buyback as part of the shareholder return. If we decide to use our cash for strategic investments, then the share buyback plan might be postponed slightly. We have a large asset of the LP investment in Toshiba. Upon realization of returns, we will be revisiting our capital policy, including a possible shareholder return. When we get the returns, our cash levels will go up. We will consider a share buyback. Regarding CAPEX, it remains the same, meaning we control it to an appropriate level. And for financial discipline, for us to promote these strategic investment projects, including a possible M&A, if we need additional financing, the D/E ratio will temporarily be over 0.5. Yet, we will keep it below 0.7. Cash on hand will be kept to a level of three-month worth of sales, which remains same. And the chart on right illustrates what I have mentioned. This was all from our slide deck. Our strong intention is to achieve a record high net sales of JPY 510 billion. We might aim to hit even higher than that. Unfortunately, the operating profit that we guided is probably not as high as you expected. We are strongly intending to get rid of all the negative factors and make ROHM back to an even profitable company. Hereby, we continuously look forward to your ongoing support. Thank you.