This is former institutional investor Keiko Kimura. Since this is a former institutional investor channel. >> It is a quite difficult industry. As an institutional investor, I'm watching it most closely. >> Something like "go, go, go, full steam ahead." >> Hello everyone. I’m Izumida, a former institutional investor. >> Izumida-san. >> Yes? >> The framing looks completely different from usual. >> It does, doesn't it? >> A total change. >> Right? >> Yes. Is this the new studio? >> Oh, right, right. It’s the new studio . Yes. >> Well, as I mentioned in a video recently, we want to try a new project. >> Oh. >> Yes. >> So, let me introduce the person next to me. >> This is Keiko Kimura, a former institutional investor. >> Oh. So you’ve invited another former institutional investor. >> That’s right. >> I see. >> It is a former institutional investor channel, after all. >> Ah, indeed. Sorry about that. This is just my personal impression, but I thought institutional investors were only older men. >> I think the ratio of older men is quite high. >> Oh, is that so? >> Yes. Then, Izumida-san, could you please introduce Kimura-san? >> Yes. Actually, Kimura-san and I worked at the same previous job. >> Yes. >> And we were on the same tech team, researching the technology sector together. >> Oh, really? >> Specifically, I covered semiconductors. >> Yes. >> And Kimura-san covered semiconductor manufacturing equipment. >> Oh. >> That was our relationship. Even if we say "semiconductors," as we've covered on this channel, there are many different companies. >> That's true. For manufacturing equipment, there's companies like Tokyo Electron, right? >> Exactly. And I covered companies like Toshiba, Renesas, and Rohm. >> Ah, I see. So that was your relationship. >> Exactly. Since the value chain—the flow of production—is so closely linked, we would communicate and share information, go to interviews together, or visit factories. >> Wow. >> Yes. So, will we be seeing Kimura-san’s institutional investor perspective on this channel from now on ? >> That’s exactly right. You know, I talk about a lot of different things, but honestly, there are areas I'm good at and, well, areas I'm not. >> Yes. >> Since Kimura-san and I covered different sectors and industries, I think I'll handle the topics I'm strong in, and maybe have Kimura-san handle the ones she's strong in from now on. So, rather than me just rambling on, I’d like to have Kimura-san introduce herself. >> Yes. Oh, nice to meet you all. I'm Keiko Kimura, a former institutional investor. It's a pleasure to be here. >> It’s a bit funny that we have two people doing that introduction. >> That's amazing, isn't it? >> It’s been a while since I've spoken in a setting like this, so I’m actually quite nervous. Yes. But, you know, watching the two of you, I felt like I wanted to join in too. >> No, absolutely, please do. You're welcome to. >> Yes. Looking forward to it. >> Looking forward to it. Actually, I mentioned that Kimura-san and I worked together at our previous job. >> We've actually known each other since our twenties. >> Oh, is that so? >> That's right. >> You both look so young. No, Kimura-san is young too, you know. >> But you haven't changed at all, Izumi-san. >> No, you haven't changed either, Kimura-san. You're joking, right? >> What is this exchange of flattery? >> We've been like this since back then. Yes, that’s exactly it. >> Is that right? Well, I guess some things never change. Exactly. When I was in my twenties, I was at Fidelity, about my third year as a working professional. >> And back then, Kimura-san was an analyst at a foreign firm. >> Oh, I see. >> She was my client. >> Exactly, exactly. She was a client of mine. Like what they call sell-side and buy-side. >> Yes. Sell-side and buy-side. >> Ah. >> Yes, yes, yes, exactly. >> Kimura-san would provide the information, and Izumi-san would be the one listening. And then based on that, he would decide whether to buy or sell. Mm-hmm. And since I was in charge of small-cap stocks back then, and she was covering sectors related to small-cap stocks as well... >> So that's when you first met. Yes. >> And after that, Kimura-san, you went to Fidelity? >> Oh, that's right. >> We were there together, and well, there were several of us looking at tech, and we worked together on that team. >> I think Izumida-san also mentioned that there are a lot of older men among institutional investors. Was it an environment full of men? >> Ah, yes, it was. Especially on the sell-side, there were very few female analysts. >> Ah. >> Really, everywhere you went, and in the offices too, well, maybe not just "old men," but there were a lot of men. >> Well, I'm an old man now too. >> That's true. You certainly are. >> But when I went to Fidelity, maybe because it's a foreign firm, >> there were a lot of women. >> Yeah. Exactly. And what's more, many of the fund managers were women. >> Oh, really? Yes. >> Yeah. >> There were so many strong, high-achieving career women. >> Right. Come to think of it, in a previous video when you talked about being told something by a fund manager, you used female-coded speech. >> That one... >> I can imagine who it was. >> There are quite a few, aren't there? The strong female fund managers are mostly a specific group. >> Oh, is it a specific group? >> But they are all strong. >> So you went through those ordeals together. >> That's right. >> Exactly. We went through those trials together. Well, I was in charge of semiconductors and semiconductor manufacturing equipment during a very tough period—specifically, 2009, or rather, 2... >> The Lehman Shock? >> Yeah, 2008, 2009. >> Right. >> That was a difficult time, wasn't it? >> Yes. I mean, Elpida and others went bankrupt, too. Oh, that's right. Which side were you covering at Elpida? >> I was covering NAND flash at Toshiba. I thought you were covering DRAM. >> I see. Yeah. >> So, Elpida, which no longer exists. >> Yes, Elpida, which is gone now. Yes. >> So you witnessed both that difficult period and the point where the business was wound down. >> Along the way, well, sectors rotate, so , >> just as things hit bottom after the toughest times, thinking a super cycle might be coming, I'd >> move on to another assignment. >> I see. And then, I changed jobs to a global tech company, so I moved over to the industry side. >> Ah, I see. After this, they go to Google, you know. >> That's right. >> Yes. >> At Google, I was selling YouTube ads. >> That’s right. Ads, like YouTube ads. >> Exactly. >> And this is YouTube. >> So, you were selling the very ads that are on YouTube. >> Exactly. That's right. >> And now you're on the talent side. >> Yeah, I never imagined I'd be on the side that gets to appear on screen. You're a performer. >> I used to tell people, "You can run ads on videos like this." >> That's right. >> So, thanks to Mr. Kimura's sales efforts, even our channel is earning ad revenue now. >> Well, indirectly, I suppose so. >> Yeah. >> At Maneyr Academy, you can watch some video content on investment basics for free. You can watch them for free by registering via the Maneyr Academy URL in the video description. Maneyr Academy provides free access to stock report cards analyzing various companies. For those who want to learn more about stocks or analyze them on their own, you can download them for free by registering as a member. Please make use of it. As for today's topic, we've received a lot of feedback from viewers asking about the "arms race" in capital expenditure by hyperscalers and the AI frontier development race— things where capital spending is moving various stock prices, which we’ve discussed here before. Since there were so many requests for the latest updates and future outlooks, I made a request in advance, and this time, Mr. Kimura has put it all together for us. >> So, as you mentioned earlier, Mr. Kimura, since you’ve been researching semiconductor manufacturing equipment, we’d like to hear your perspective on how to view their fundamentals and stock prices. >> Yeah. >> Also, since you were at Google, I’d like to ask if you were directly involved in that, but having been on the buying side, you can offer a perspective from both sides, which is why I’d like you to explain things today. That’s what I’m hoping for. >> I see. Thank you. >> Let’s get right into the main topic. To start, the movements in semiconductors, AI-related stocks, and data center stocks are incredibly fast, aren't they? We often hear that capital investment is crucial, but I’m really curious about what institutional investors are looking at right now and how they make their decisions. Could you tell us about that? >> Yes. That’s exactly what we were doing when we were in charge of this area, wasn't it? >> We were. >> We really were. So, let me start by explaining from that perspective. >> Please do. >> Yes. There are many different stocks, but I’ve summarized the trends here to show you how the stock prices of representative semiconductor manufacturing equipment companies— Tokyo Electron, Advantest, Disco, and Screen—move and what they correlate with. The line graph shows the stock price trend. >> And the bar graph represents the year-on-year change in global semiconductor manufacturing equipment shipments. This refers to the equipment shipments, or rather, the global sales of semiconductor manufacturing equipment makers. >> Right. Right. >> Since this is the year-on-year growth rate, if you look at it, you can see that the stock price starts to fall a little before the bar graph starts to decline. >> Yeah. Yeah. Can you see how, a little before it shifts from negative to positive, the stock price also starts to rise, and this pattern keeps repeating? >> Looking at this, it's safe to interpret "shipments" as being nearly equal to sales, right? Indeed, looking at this, when shipments start to decline, the stock price drops sharply; on the other hand, it starts to rise a little bit before shipments actually increase. >> Yes, that's right. Actually, it's not that shipments themselves are decreasing. This is the growth rate of shipments, so >> even though shipments are actually still rising, if that growth slows down , the stock price will already start to drop. >> I see. Because this is the growth rate. >> That's right. >> So it's not the absolute amount. >> Exactly. Oh. >> And even when the shipment amount is still negative, once the margin of the decline starts to shrink, it begins to rise. Mm-hmm. >> So, it's such a difficult industry where stocks start to rise even though sales are falling, and they start to fall even though sales are rising. >> So the stock price moves to price things in ahead of time, right? >> Yes. That's correct. That’s why we have always looked at sales—well, there are orders before sales, but many companies don’t announce orders now —so rather than looking at whether sales are growing or what the amount is , we look further into the change in the growth rate, like whether the growth rate itself is falling or rising , which is a very professional way of looking at it. >> I always say that stocks are all about change, you know. >> Yes. >> They are nothing but change. >> I see. >> Exactly. And it's the rate of change of the change. >> That's right. And when it comes to the growth rate of the change, if you want the stock price to continue rising, it feels like the growth rate itself has to keep increasing. >> That's exactly it. >> So, in terms of absolute values, it has to grow exponentially. >> Otherwise, the stock price won't go up. >> That's right. It's... >> It’s quite difficult, isn't it? >> Well, that is one of our themes for today, actually. >> Yes. >> Yes. That is exactly it. So, listening to what you’ve just said, it comes down to how future capital investment in semiconductors—specifically investment to boost shipments—is going to grow, right? >> Oh, I think you're absolutely right. So , when looking at semiconductor manufacturing equipment stocks or semiconductor stocks in general, the largest portion of their capital investment currently comes from hyperscalers. Institutional investors focus most on not just how that investment looks right now, but on predicting how it will change in the future and where that rate of change will plateau. Mm. Mm. Is it even possible to figure something like that out? >> That’s where institutional investors show their worth. Oh, I see. >> Exactly. >> It’s about what approach you use to forecast. Well, forecasts are always wrong, but... >> Yes. >> You create a few scenarios: the best-case scenario, what happens if we do nothing, and the worst-case scenario . Mm-hmm. And then you say, "I believe this case is currently the most likely, " and that's your... >> Your "go-to" case. >> Right, your "go-to,""go-to" case. This is what I consider to be the main scenario. So, all these institutional investors are making their own forecasts, which is why there's such a wide gap in market consensus depending on who you ask? >> Exactly, exactly. >> That’s interesting. By the way, considering the current situation, could you tell us what kind of cases are being anticipated and what kind of future might be in store? >> Oh, yes. First, to look back at how the market currently views things, I’ve brought the latest capital expenditure data for each of the hyperscalers here. For this, we're looking at four hyperscalers: Alphabet, Meta, Amazon, and Microsoft. Because their capital investments have such a massive impact across so many areas, they actually provide guidance for us. They’ve released estimates for this year’s capital expenditure, and since it’s currently 2026, the current fiscal year for these companies is 2026, and they’ve provided these figures as their current-period guidance. >> Well, each of them is now around 200 billion dollars, and 200 billion dollars is about 30 trillion yen. >> Oh, that’s right. 30 trillion yen, yeah. >> Around that much, yeah. As for 2027, since that’s next year, few companies give guidance, but many are saying it will increase significantly. Companies like Alphabet and Microsoft have clearly stated that it will increase even more than in 2026. However, since they haven’t provided specific amounts, the market can probably price in up to 2026, but opinions likely diverge quite a bit for 2027. >> I see. >> Yeah. >> I see. If they write "significant increase" like Alphabet does, you think , "Oh, it’s going to grow even further from here," but without concrete numbers, you wonder how much exactly. >> That’s going to vary depending on who’s looking at it. >> I think it will change. >> So that’s what it comes down to. >> And as we mentioned earlier, the growth rate is what’s important. >> So depending on the actual amount, whether the growth rate decelerates or accelerates changes everything. >> That’s certainly true. So even if they say a "significant increase," I think everyone is really curious about how much that actually is. Exactly. We’re playing the game based on the rate of change, after all. >> Yes. >> It might be "significant" to the company, but to us, it might be like... " >> The growth rate slowed down, right?" >> Exactly. I see. So, in short, if you look at 2025 versus 2026, and then move to 2027, even if it increases compared to the growth from 2025 to 2026, if the percentage isn't as high as last year's , institutional investors will judge it as a deceleration in investment. >> Right. Even if the dollar amount increase is the same, the growth rate is slowing down. >> Yes. That’s exactly it. >> Ah, that’s true. >> But from the company's perspective, they’d say, "No, we're increasing it by the same amount." Yeah. That’s understandable, isn't it? Because they’ve increased it quite a bit. I mean, >> there’s a bit of a mismatch in the conversation there, like, "Didn't you say this at that time?" And they’re like, "No, but we are increasing it." Oh. >> Exactly. But from our perspective, the growth rate is decelerating. >> That’s right. And investors are very demanding, so >> they want the growth rate to accelerate ; they won't be satisfied unless the investment amount keeps growing faster and faster. >> That’s just how they are. Yeah. >> I see. So, the people we call investors in this world make various judgments based on those premises. Yeah. >> Right, right, right. Ah. >> That’s true. >> Even if a company carefully plans and builds up their growth curve, if that curve isn't steep enough, as we saw in the graph earlier, they can still get sold off for a while. When they announce their figures, the market prices them in immediately, so if they miss, the stock falls, and if they beat expectations, it rises, and the stock starts moving from there. >> They’ve already issued their guidance for this year, so if capital expenditure falls short of that figure, >> yeah. >> hyperscalers will get sold off significantly, right? And I think related stocks could also be sold off. >> It’s not really about current earnings anymore. Because it’s about the future. >> It’s focused on how the most important variable for future earnings will change. >> And for these hyperscalers, the most important factor >> is the growth rate of capital expenditure. So, I’ve included the actual capital expenditure of the four hyperscalers through 2025 and their 2026 outlook based on their guidance. As you can see, the growth rate started climbing around '23, began rising from '23 to '24, accelerated further from '24 to '25, and then more than doubles from '25 to '26. It's this one, isn't it? >> Yeah, so everyone has been doing this since about last year. >> With investment money increasingly pouring into related stocks and so much already priced in, the main focus now is what the growth rate will look like in '27. But based on what we said earlier, this current chart shows 358 for 2025. >> And then 669 for 2026, which is a massive jump—not quite double, but still significant. So when it comes to that growth rate, unless we see an incredible figure like 1,300 or 1,500 in 2027—essentially doubling it again —it's hard to see the stock prices continuing to rise. >> That's true. In the short term, yes. >> I wonder if they'll really put that much out there. >> Right. Depending on which scenario we follow, I’ve put together a capital expenditure forecast for the four hyperscalers based on one specific assumption. Regarding the scenario we're using: there are various ways to look at this, which makes it tricky, but since these hyperscalers are investing primarily for their data center business, and since they all report their order backlogs for the cloud services that drive those data centers, I’ve based this forecast on the assumption that the current pace of those orders continues. Yes. This graph shows the order backlog—the amount of cloud business that hasn't been recognized as revenue yet—and it represents the combined figures for Alphabet, Amazon, and Microsoft. The most recent figures go up to the June quarter, as they are announced quarterly, so that's the data included here. I think everyone is waiting for the numbers up to September, which should be announced soon, but for now, these are the figures through June. Looking at this, you can see that the order backlog is piling up. So, we have new orders coming in, and since some existing orders are realized as revenue , the remainder becomes this order backlog. >> With the order backlog and the cloud business revenue already realized, we can calculate how much the new orders were. >> I see. I see. >> And here is the reverse calculation: the new orders, or rather, the new cloud business orders. If you look here , this is last year's Q4, the fourth quarter. You can see it jumping up significantly from here, right? >> Yes. Around this time, companies like OpenAI and Anthropic, those LLM providers, started competing to secure data center capacity by placing orders, so the order volume started shooting up . Now, it has continued at a similar pace through the second quarter of 2026 , so I'm projecting what would happen if this high level remained flat. If these new orders continue through 2028 and 2029, we can estimate what the cloud business backlog and revenue will look like, and they are making capital investments to generate that revenue, right? >> That's right. >> So, if we can forecast the revenue, we can also estimate the necessary data center capacity and the resulting capital expenditure, which is what I’ve modeled. Yes. >> And the outlook for capital expenditure in that scenario actually looks like this. >> Huh? >> Huh? >> Huh? >> Huh? >> I see. Exactly. Well, apart from investor expectations, the business itself can't just pour in unlimited funds for capital investment without a clear outlook on generating actual revenue. >> Yes. >> You can't just keep investing indefinitely, can you? >> That's correct. I see. >> And when that levels off, it settles at the level of capital investment required for that amount. >> Oh, I see. >> Yes. So, in reality, unless new orders grow significantly, capital investment won't grow significantly either. >> Oh, well, that makes sense, doesn't it? >> I mean, if there's only demand for 100, >> investing in equipment for 1,000 or 2,000 would be a waste. It becomes pointless. >> Yeah. >> I see. Looking at this, from 2026 to 2027, well, it doesn't quite reach 1.5 times, but. >> That's true. Next, regarding the growth rate of this capital investment, looking at the current growth rate like this, 2026 is, yeah. It's accelerating. It's accelerating to 87%, but wait, what about 2027? >> Wait, this is the same as the stock price of the memory manufacturers. >> Yeah, that's right. When you look at it , it's like, "Is this year the peak?" So, it has already peaked. >> I see. Oh, I see. >> So, 110,000 yen. >> Right. Yes, yes, exactly. So, unless current orders accelerate... >> Right, but if it accelerates, we'll know again. >> Yes, exactly, that's it. >> I see, I see, I see. >> Huh. >> Now, this is, right? It's more about the growth rate of hyperscaler capital expenditure, not so much the hyperscaler's own stock price, but rather companies that supply products for that investment, semiconductor stocks, and data centers. Related stocks, yeah. >> These are the ones directly tied to stock prices. >> Exactly. >> That's what it comes down to. >> Yes. >> This capital investment is equal to the shipments from the semiconductor manufacturing equipment and materials makers mentioned earlier. >> Yeah. Exactly. Sales, right. Data centers are said to be about 40% building costs, and about 60% semiconductor or computing hardware-related capital investment. >> So, the market has started to become aware of 2027, yeah. >> Hmm. >> At this timing, it's hard to tell how much is really priced in, but if this is correct, well, it's priced in. >> Unless the outlook changes significantly, the stock price will likely move sideways. >> Yeah. >> It would move within a box range at current levels, and there's talk currently about the AI threat theory. >> Like, let's slow down the capital investment and development race. But all the capital investment up until now has been about competing. That number came out of everyone jumping in and saying "I want in, I want in, let's compete." >> I see. When talk turns to everyone calming down and taking a measured approach, even if the absolute amount is growing, the rate of change starts to drop. >> Ah. >> It becomes harder for stock prices to rise, and if they start talking about aligning their capital investment levels or development standards—not necessarily regulating it—the investment amount might even turn negative. >> I see. >> There's a possibility it could end up like that. In Dragon Quest terms, it's like switching from "Show No Mercy" to something more like "Everyone, Do Your Best." Yeah. Yeah. Oh, "Everyone, Do Your Best." Well, it hasn't quite reached "Focus on Healing" yet. >> It's not "Focus on Healing," but looking at these graphs, I got the impression it shifted to "Everyone, Do Your Best." >> That was the vibe. >> Competition usually goes too far, doesn't it? Like with semiconductors. >> Yeah. Yes. >> You know, with cutting-edge technology ... " >> Full steam ahead" situations happen all the time. Because everyone is competing , after all. >> Right. It's called an overshot; as a result of all these incredibly talented people thinking hard, it becomes excessive competition. >> Huh. Is that what they call the silicon cycle? >> Exactly. >> Ah, that's right. That's right. >> Ah, that's exactly it. >> I see. >> You said it well. Exactly, exactly. >> History tends to repeat itself, doesn't it? That's right. >> There's no learning from it. >> Wait, but isn't this time also...that? >> A supercycle? >> Ah, that's it. That's it. >> No, the moment you say "supercycle," it's over. >> I'm already tired of the word " supercycle." A bubble. A supercycle is something that, if you look at it on a 50-year or multi-decade scale, probably is a supercycle. >> Ah, so it changes depending on the time axis you look at. >> Time, yes. Even back then, when they were talking about a supercycle after the bottoming out during the Lehman Shock, it took about 20 years to get to where we are now, right? >> Oh, isn't "supercycle" a term that was born recently? >> No, it isn't. No, it's not. It's been around for a long time. Oh, is that so? About the tech supercycle. >> Yeah. Oh, and, >> Yeah, having started stock investing in the last few years, I’m like, "Oh, a new term popped up." Oh, I guess that happens too. I hope that happens. I was just feeling excited about it, but >> There are all sorts of things like " supercycles" and "golden cycles." >> Oh, I see. >> That this time is different. >> Exactly. The moment someone says "this time is different," I’m like, "Ah,"" >> Here it comes." >> You said that last time, too, didn't you? When you say "this time is different," somehow >> fund managers usually tell you it’s a bubble, implying a peak. >> Right, exactly. Yeah. Based on what we've discussed so far, regarding the growth rate of capital expenditure you showed earlier, >> Yeah. it was said that whether or not that continues to grow will determine the future of Japanese data center stocks. On the other hand, when considering the stock prices of the hyperscalers themselves, what becomes important? >> Well, what’s important when looking at hyperscaler stock prices is that in the past, for hyperscaler stocks, it was all about how fast sales were growing. If sales were growing, they were still in a growth phase, so that was good. But now, sales growth for their existing businesses is starting to settle down, so I think how much profit they can generate is becoming what's really important. >> Yeah. Yeah. >> Up until now, these were companies that were asset-light, able to grow sales without needing much capital investment , and that translated directly into profit. But now they are starting businesses that require significant capital expenditure, so everyone is looking very strictly at whether they can actually earn a solid return on these new businesses. >> Ah, I see. Meta and Google are basically ad businesses, so >> they’ve shifted from a world where they could generate steady sales through advertising without needing their own data centers, to a world where building data centers means their assets just keep piling up, right? >> That’s exactly right. And what impacts profit as assets increase is something called depreciation. >> I see. >> Yes. So, speaking of that, this is a simple summary of how much depreciation these hyperscalers, who are making massive capital investments, actually have to account for. >> Generally, the useful life for servers or network equipment—for Alphabet it’s 6 years now, and it’s not just Alphabet, but everyone sets it in a range of about 5 to 6 years—so the capital investment divided by 5 or 6 years is roughly what gets recorded as the annual depreciation expense. >> Oh, looking at this chart, it just crossed my mind: are the useful lives for servers or network equipment not fixed by law, but something the company can decide to some extent, like saying it's 4 years or 5 years? >> It is up to the company and their accounting. >> That's right. >> Oh, I see, so that's how it works. >> So, they are investing a massive amount of money into things like data centers right now, aren't they? And if they decide to amortize that over, say, 5 or 6 years, then for 5 years, it would simply be the total investment amount divided by 5. >> And that gets recorded as an expense every year. Doesn't just changing that period from 5 to 6 years, or 6 to 4 years, >> make a significant difference in the expense amount? >> Yeah. Well, if you do that, people might ask if you’re in trouble. >> It’s manipulation, right? >> Exactly. >> Ah, >> like, aren't they manipulating the numbers? >> You can't really change it often. Once you disclose it, yes. But looking at this, in 2023, Alphabet changed it from 4 to 6 years, or 5 to 6 years, so by extending it, they are splitting the expense recording more finely and... >> reducing >> the cost per year, so that's the kind of move it is, right? >> That’s correct. Yeah, that's right. >> Yeah. >> So, whether these useful life periods are truly accurate or not seems like something that could be up for debate. Well, as long as it's based on facts, like the hardware lifespan actually extending, that would be fine, but... >> Yeah. >> Five or six years is probably not an incorrect figure for the lifespan of the hardware; you could keep using it if you really wanted to. >> The problem is that it becomes cheapened. >> The card itself is fine, but there are all sorts of calculations performed by those semiconductors, right? With Claude having Opus, Haiku, and so on, as functions evolve, the hardware might not get used as much. >> It becomes cheapened, so some people say six years is too long and that three years is more realistic. >> NVIDIA’s architectures, for instance, keep coming out with Blackwell or whatever every two years or so, right? So, if you try to keep up with all of that, it seems like there should be a discussion about whether this depreciation period is really appropriate. >> Yes. There is definitely room for debate here, but for now, five to six years is the timeframe generally accepted by audit firms. >> That is the current situation for these four companies. Uh-huh. >> Based on this, is it possible to project depreciation, as well as sales and profit forecasts? >> Ah, yes. This here shows the depreciation forecast based on the aforementioned capital investment projections using current depreciation standards. As capital investment increases, depreciation will accumulate , so we expect depreciation expenses to continue rising like this through 2027, 2028, and 2029. Next, let's look at revenue. There are many ways to view this, and projections will vary depending on the perspective. However, the premise here is that existing businesses—like the ad business for Alphabet or Meta, and the e-commerce business for Amazon—continue to grow at their current rates. We then add the cloud revenue projected from the backlog we estimated earlier to get the total company revenue, which is what this forecast shows. So, cloud is still growing, and existing businesses are growing as well, which means overall revenue is increasing. >> Yeah. >> But as mentioned earlier, depreciation costs will increase at a faster pace than this revenue growth, so as a result, operating profit won't grow much. >> Oh. >> That is what we can imagine. >> I see. >> Yes. >> I see. After all, in this transition from an asset-light business to a capital-intensive one that requires heavy equipment investment, if revenue doesn't grow faster than the pace of depreciation from those investments... >> Right. ... >> profit isn't going to grow. >> So, the current model is based on the premise that revenue grows at that specific rate, right? So if, for example, the cloud portion of revenue were to shoot up exponentially, wouldn't that change how this looks? >> Actually, you can't say that for sure. >> Oh, really? >> Yes. Because this model predicts capital expenditure based on the cloud revenue forecast. >> I see. >> If anything were to change, it would require needing less capital expenditure to generate that cloud revenue. Yeah. >> For example, if semiconductor prices fall, or if you could generate cloud revenue without spending so much on building data centers—unless the ROIC or the return on investment improves— it might be difficult. >> Yeah. I see. >> There's a lot of discussion about ROIC and WACC lately, isn't there? >> Oh, that's true, we've certainly been covering those topics more often on Izumi as well. Originally, companies like Google and Meta had asset-light business models with high returns on very little investment, but now they've reached a point where they can't generate sales or returns without massive investment, and with interest rates rising globally, the ROIC for these hyperscalers is plummeting. >> Yeah. >> I thought it was interesting that even Nikkei wrote an article about how these lines are crossing. Well, oddly enough. If ROIC falls below WACC, it means they are effectively destroying corporate value. >> So they should stop? >> Exactly, it means they shouldn't be making capital expenditures, but then they can't be hyperscalers anymore, so the whole growth story just collapses. If the growth rate of operating profit were to follow that path, >> Yes. >> The stock price would be in for a pretty rough time from here. >> That's true. Ultimately, as you mentioned at the beginning, the rate of change >> Yeah. If investors value that metric, then there won't be much of a rate of change to see, right? In this situation . >> That's right. >> I see. >> However, since these hyperscalers have plenty of room to maneuver, there is a possibility they could boost profits by cutting costs further. >> I do think that's possible. But since we are assuming they don't do that, >> I feel they will likely try to grow their profits by cutting costs to improve the ROIC of their existing businesses. >> I see. >> Yes. Well, Mr. Kimura, you've explained the outlook for hyperscalers and the pace of their capital expenditure today . Could you summarize the key points for investors watching Japanese data center stocks, as well as the important factors for the outlook of the hyperscalers 'own stock prices? Well, yes, for data center-related stocks, the most critical part is the growth rate of the capital expenditures by hyperscalers, which serves as the source of their revenue. >> Yeah. >> It's not about whether capital investment increases or not, but rather what the pace of that increase will be. >> I think whether or not they will grow is what's most important. So, for hyperscalers, the question is whether their operating profit will grow going forward. And what impacts that most is what happens with the depreciation costs stemming from capital expenditure . And whether the ROIC of the data center and cloud businesses—which are currently generating returns from that capital expenditure—improves or not is what I think is extremely important. As an amateur, I had this idea that if the amount of capital expenditure just keeps increasing, the stock price would grow steadily, but... >> That’s not the case, is it? >> Well, it’s not that simple of a story . >> That’s right. >> Institutional investors really do rack their brains all the time trying to forecast this stuff, right? Yes. >> So, if the growth rate is constant, institutional investors end up pricing that in beforehand, right? >> Well, yeah, they price it in. And after pricing it in, they wait for the next change. >> The moment it's priced in, the stock price tends to stagnate. I’m always saying that. >> Ah, that’s true. >> If those expectations don't change, it’s tough for those making the forecasts, and it’s even tougher for the companies trying to meet them. >> Well, it’s tough for companies, but some just choose not to respond, you know. >> That’s true. >> Yeah. >> They accept that the stock price is just something the stock market decides on its own. >> Yes. >> Some managers manage that way, and that’s fine too. >> Yeah. >> So, the company just doesn't care about the stock price. >> Yeah. That’s the idea. >> Of course, if the stock price is valued in a way they can't accept, they might put out a message from the company. >> Well, sometimes that takes the form of buybacks, which I think is a way of communicating as IR, but in the stock market, investors just set the price themselves, don't they? >> Yeah. >> That’s what it comes down to, isn't it? >> So, as an individual investor, it seems like a good idea to keep the mindset of institutional investors in mind when investing in hyperscalers or data center-related stocks from now on. >> Yes. >> I see. That makes a lot of sense. Thank you very much. >> Thank you very much. >> Thank you very much. >> At Izumi-Oizumi, we explain useful investment information for both Japanese and US stocks, as well as how institutional investors view the news, in an easy-to-understand way. If you enjoyed this, please like the video and subscribe to our channel. Also, if there are any topics you'd like to ask Izumida or Kimura about, please leave them in the comments section. We'll see you in the next video. Goodbye. Bye-bye .