Stock Splits + Semis

Data is everywhere. When orchestrated properly, it sings. At Morningstar, we analyze and enrich data, making it actionable and powerful for you. Morningstar, where data speaks.

The beauty of covering semiconductors or electronics hardware is nobody can forecast end market demand. It's always a black box.

Hello and welcome to the Barron's Streetwise Podcast. I'm Jack Howe, and the voice you just heard is Mehdi Hosseini. He's a semiconductor analyst at Susquehanna International Group, and he says that spending on computing power is about to peak, maybe dip, and that might have big implications for the stock market. In a moment, Mehdi will tell us more about his view, including which types of semiconductor companies he thinks spending will flow toward. That's later. First, we'll say a few words about stock splits. Listening in is absolutely no one. Our audio producer Jackson Cantrell is off this week. He's traveling. Can't say where. Rhymes with Guatemala. Can't say any more than that. Rather than bring you a rerun, I thought we'd bring you a quick, fresh episode. I've never done this by myself before. I don't know how this works. Who's going to be there to reel me back in if I get sidetracked? If I start talking about— it's probably going to be an old TV show— Six Million Dollar Man, right? If I start singing, who's going to tell me if I triggered royalties? How's that work? All by myself. Is that— was that expensive? Feels like that one should have been free. Let's see how it goes. I'm going to bring you to start with what I think is a bad investment strategy from really smart people. I'm not going to recommend it, but I do think it's interesting and it has to do with stock splits. Does everyone know how stock splits work? A stock split is just where a company arbitrarily says our stock's been doing really well, we've got a high price, we're going to split it into more shares at a lower price. It's pretty simple, but I have overheard conversations from people who don't seem to quite know. You ever heard someone who's an investing novice and they say, well, they want to buy this stock over here because it's trading at $750, so it must be doing really well. It's doing a lot better than that other stock over there that's only $30. And then someone else will come along and say, no, no, that's the wrong approach. What you should do is buy low and sell high. You should buy the $30 stock, maybe to become a $750 stock, and then you sell it.. And you can kind of see where both parties are coming from. If you were talking about some other type of good, if you were talking about a sport coat or a smartphone, then a higher price might mean it's probably a better thing. And also if a stock is $750, maybe it's a big recent gainer and it's been doing well, but that's not always the case. I saw a little while ago Regeneron Pharmaceuticals, it was in the $680-something range. It was down 28% over the past year. I saw Adobe, it was trading a while back in the $440s. That was down 30%. So, not everything that's trading at a high nominal price is on a rip-roaring run. But also, experienced investors know that share prices are kind of arbitrary things, right? Companies can just make them up to a point anyhow. When they go public for the first time, they or their investment bankers could just pick a price. And then they can adjust that price every so often through stock splits or if If things go poorly, they can do reverse splits. The nominal share price doesn't really tell you too much about a company. You have to know how many shares are outstanding, and that tells you the market value. And as for the buy low, sell high part, I can't stand that saying. It's kind of a truism. A truism is something that's like, it's correct, but it's so obvious that it's unhelpful. Like someone who says, it is what it is. Well, yes, it in fact is what it is.

Thank you.

I appreciate you telling me. But with buy low, sell high, I think that might actually be an untruism. I think it's both unhelpful and not especially true. First of all, it's really difficult to tell what's low and what's high. That's the whole thing that makes up a market. Every trade that you have is basically a disagreement between two parties about whether the given price for a stock is too high or too low. But also, stocks tend to rise over time. If you're an S&P 500 index investor, you have made out beautifully. 4 decades by buying high and then holding out for higher. If I wanted to rewrite buy low, sell high, it would probably go something like buy often and ideally hold until it's someone else's problem, because either you're dead or you're rich enough to give it away. That's not very pithy, so I don't think it's going to catch on. But the point is, neither of these two well-meaning people in the conversation that I laid out earlier really has it quite right. But I want you to forget all that because at the moment, the first person in the conversation might be pretty darn right, the one who wants to buy the $750 stock. And this is according to some recent observations by the Research Investment Committee at B of A Securities. Let me share some of those observations. Number 1, more stocks than usual have been splitting. There were 17 split announcements last year in the S&P 500. Companies like Nvidia, Broadcom, Walmart, and Chipotle, and that is the most since 2013. And if you're thinking that 17 doesn't sound like a lot, it's not a lot if we're using a longer lens. In other words, we've seen 44 splits over the past 5 years. It's a large portion of that, but it's nothing compared with the 364 splits that we saw over the 5 years that ended when that dot-com stock bubble popped all the way back in 2000. But there's a reason to suspect now that splits will become more popular from here because, and this is observation number 2, stock splits have been leading to big returns. Since 1980, stocks have returned an average of 25% over the year following split announcements. That's versus 12% for the market overall. And as good as those returns for splitters have been, Lately, they've been even better. Last year's class of splitters, they returned an average of 17% in the 6 months following the announcement. There's one more piece. Observation number 3: There are a lot of split candidates now. That's what happens when you go a long time with no splits and the market goes way up. You end up with stocks with high share prices. Among S&P 500 companies, there were recently 40 trading above $500 a share. I've got some thoughts on why stock splits might have dried up. Decades ago, trading odd lots— that's fewer than 100 shares— it could be a little cumbersome, a little more costly, and there was no one out there buying fractional shares. Maybe if you had a drip plan or a dividend reinvestment plan, but nobody called their broker and said, I'd like 10.4 shares of whatever. Today, when you put an order in online, a lot of brokerage firms ask, do you want to buy a number of shares or do you want to buy a dollar amount? Trading in odd lots and decimals is no big whoop, which means nominal prices are less important than ever. And even B of A, who produced this research on splits, points out that splits don't really affect fundamentals. Warren Buffett famously avoids splitting Berkshire Hathaway A-shares. Those were recently more than $700,000 per share. His belief is that the high price attracts only long-term investors, not flippers. My hunch, and this is only a hunch, I think that not splitting shares just became kind of big tech fashionable over the years, like wearing those Alpine vests to work. But now share prices are super high, splits are bringing big gains, and big gains are certainly fashionable, just like Alpine vests. Are those still fashionable? I feel like probably, which probably means definitely not. So even if the whole thing doesn't make sense, splitting your shares into more shares, that should not make your stock price go up. But if it is making stock prices go up all around you, you, maybe you're going to do it. There could be a coming split for LANCHE, and that raises the question of who's next. I want to stress again, I don't like this very much as an investment strategy. It's gimmicky. You should buy into good companies you want to hold for a long time, not try to guess who's going to split. Anyhow, a lot of sensible things aren't working at the moment. You know what's doing great? GigaChad. GigaChad is a meme coin and it represents, and I am quoting here from gigachad.com, Tokenized masculinity. GigaChad is up thousands of percentage points over the past year. It has a market value of hundreds of millions of dollars. I am leaving out some key fundamentals here. There's also a picture of a guy with a beard on the website, and there's phrases like, quote, "I don't care, I win." A phrase like that feels like it's worth about $14 million right there. I'm not an expert on tokenized masculinity valuation. One last thing to point out is that the list of split contenders, it reads a lot like the high flyers that investors are chasing anyway. The biggest commonality among past splitters is that they had rapid price run-ups before the announcements. So if you start with the $500+ stocks and you sort for ones that have done better over the past 1, 3, or 5 years than the average company that then decided to split, You find, according to B of A, names including Netflix, Fair Isaac, Eli Lilly, Meta Platforms, and Goldman Sachs. The point is that I don't recommend split chasing, but you do you. I just had the best idea for a meme coin. I Googled it and I don't think it's taken. GiggleChad. Now hear me out. It's like tokenized trying to be masculine, but you can't pull it off because you're laughing too hard every time you try to do it. Giggle chat. What you do is you get in on the ground floor, like fractions of a penny. You got to put it on the blockchain, Ethereum, warp speed, Hyperloop. I'm not a details man. Someone out there figure it out and make a billion dollars. If it doesn't work, don't tell them that Jack Howe from Barron's recommended it. And that is stock splits. Next up, chip spending. We'll hear from Mehdi Hosseini at Susquehanna. That's next after this quick break. Data is everywhere, but is it ready for consumption? Morningstar developed the language of global investment data so you have the right ingredients to help you shine. Morningstar, where data speaks. Model portfolios have evolved from a one-size-fits-all solution. In a recent roundtable conversation, Jeffrey Saif of BlackRock and Alessio DeLongis of of Invesco, break down what's driving the next model moment, discussing how a deeper level of customization is producing a better alignment with investor goals and creating a practice differentiator for financial advisors. Welcome back to the Barron's Streetwise Podcast. I'm Jack Howe. Jackson Cantrell is off this week. It's just me. One is the loneliest number that you'll ever do. No, focus, Jack. Semiconductor stocks have been driving the stock market. Some of them, not all of them. Everyone's talking about artificial intelligence. NVIDIA is leading the charge. It's been a lot of spending on computing power. But I had a conversation recently with Mehdi Hosseini. He's a semiconductor analyst over at Susquehanna. And he says that kind of spending looks like it's about to peak. That sounds a little scary, not just for anyone who has NVIDIA stock, but someone who has an S&P 500 fund. So I wanted to ask Mehdi more about what does that mean, where does it take us, and what should investors do? Let's hear some of that conversation now. Mehdi, so I know you cover ASML and we did something on them recently, and I know you cover a lot of stocks actually in the space. And I wanted to now just get a broad picture of your coverage and the group and what investors should be thinking about the relative value there, how attractive the group is. What do you think is going to happen with it from here? Can you give us the big picture?

Sure. My coverage is different than a traditional sell-side analyst. I try to look at the entire supply chain, and sometimes it's helpful in identifying opportunities, and sometimes it's very difficult because there's so many different subsectors. So the way I add value to investors or even company is by connecting the dots and looking at the entire supply chain. I cover some of the semi names, like memory. I cover semiconductor manufacturing, like foundries, TSMC. I cover a couple of IP companies like Arm, Rambus. And then I also cover OEMs and ODMs. We discussed Supermicro 6 months ago. I also covered Dell, Pure Storage. Those are the OEMs.

What's your rating on Supermicro these days?

It's still a sell, and I think my price target is half What's this?

I'm looking at the chart. It's been a disaster. You were bearish then, you know, you warned about it and it's been a disaster since we spoke with you. So you were correct in everything you told us there. Thank you. Do you find that there are times when you say, okay, I see something in this company I'm covering here and that tells me that change is coming for this other, not fully similar company over here. And under normal circumstances, you know, a regular analyst would not be covering both of these companies, but I do. Do, and I'm— and so therefore I'm getting an advanced warning of what's to come for this other company. Does that happen?

You nailed it. For instance, just to give you some context, and this is factual, not opinion, we were able to, uh, be ahead of the cloud spend peaking in 2018. The same thing in '22. And our last analysis suggests that the cloud spend Despite all the hype of the past 2 weeks, the cloud spend in terms of the year-over-year change in a quarterly spend should peak by middle of '25.

That sounds like bad news. Is that bad news?

It's not bad news. It's basically they will spend, they turn the faucet on, and then they turn it off to better capture economics. Don't forget all of these LLMs. Have to generate positive ROI.

Large language models for these chatbots and return on investment, right? So it's time to stop this spending spree and it's time to start getting a return on the money. But it— but isn't that bad news for some of these stocks? I thought some of these stocks were running up on the belief that everyone's going to spend freely on AI for many years to come.

The beauty of covering semiconductors or electronics hardware is Nobody can forecast end market demand. It's always a black box. So I have my way of doing this. I would like to argue that I have been more right than wrong, but the market still hasn't figured out how to forecast cloud demand. Back in '22, everyone was convinced that, oh, we're just going to go through the roof and the cloud spend is never going to decline. And then Boom, in '23, it kind of slowed down. We went through digestion. And by the way, when we talk about a peak in the growth, it doesn't necessarily mean it's followed by a decline, but the stock market is very sensitive to these inflection points. And I'm basically saying that, yeah, there is an inflection point middle of '25, and then we have to go through a digestion. And ultimately, enterprises have to find a way to inferencing, and they have to realize economic benefits of increased automation that AI brings us.

Mm. So what does an investor do if they subscribe to that belief of yours? I think that's going to happen. Will the same things that have worked in the past for investors continue to work, or do they need to take a different approach in the kinds of stocks that they're favoring?

More of the latter. I think if the past few years has been about investment in compute, I think over the next 12 to 18 months the focus will be on networking. The focus will be on interconnecting these AI server racks or interconnecting these AI facilities. So the shifting of dollars of spend is gonna change from compute to networking.

Got it. Okay, so you don't just want to pour investment dollars into compute now, you want the companies that are doing the things around the edges to enable all this computing power to, you know, be of more benefit to the customers. Are all these stocks expensive? I mean, I feel like not every semiconductor stock's had a big run-up, of course, but some of the companies that have been— anything AI-related has been a Wall Street favorite. Are the valuations any more attractive when you start to move past compute and look at some of these other areas that you're talking about, or are they still pretty expensive?

They are expensive, but if you think about the upside to the earnings, that would rationalize or justify higher multiple. There's more upper revision to earning estimates for the networking or these fiber-focused companies.

Okay. What are your favorites?

So in terms of instrumentation, especially for fiber, imagine if we're putting more satellite into orbit, that's another means of communication and Keysight, K EYS is in the sweet spot. Corning is another company that is in the fiber. They actually manufacture the fibers and the fiber is displacing copper. So as you're building out more high-speed data center to accommodate AI infrastructure, you interconnect them through fiber. That's where the Corning comes in.

Keysight's about a $31 billion company as I look at it. Corning, this is the same company that people of my age will remember from, from the dot-com stock bubble days back when everybody was excited about laying fiber.

You're correct. 25 years ago, they did have some exposure to telecom, but those assets are still there. And now that AI is driving investment in interconnection, well, those assets are generating revenue. Yeah. I also cover a company called Tower Semiconductor. This is a specialty foundry. They actually make the silicon photonics that is used for accommodating the interconnection of these very expensive AI compute chips.

Silicon photonics. I'm trying to think about what that might mean.

You're replacing a physical wire with A faster expedited means you're using photon rather than a copper to transmit data.

Oh, I see that. That sounds fast. Yes.

If you think about liquid cooling and everyone is complaining about electricity required for these AI server, well, if you can replace copper with silicon photonic, there will be a dramatic reduction in heat generated.

By liquid cooling, you mean that these computer chips get very hot as they get denser and smaller and faster, they generate tremendous heat and you have to dissipate that heat somehow. And one of the ways that people do it in these high-end installations is they use liquid cooling. And you're saying with this silicon photonics, it reduces the need for that?

Yes. If you replace copper with silicon photonics, you have a dramatic reduction in the heat dissipation. So you can just use air cooling.

That's one of the big challenges, isn't it? I mean, everybody always talks about, well, you have to get denser circuits and, you know, faster chips. And but at some point it's like, okay, you can make them faster, but you got to make them cooler. It doesn't do you any good to make them faster if they melt everything around them.

Exactly. And if you replace the copper with photon, you can accommodate faster interconnection, faster transmission while also minimizing the heat that would be dissipated.

Got it. Let me ask you, you know, you cover individual companies, so it is not part of your responsibility to oversee, you know, whether people's S&P 500 funds are going to do well for them. But people do own those funds, and those funds do have a lot of exposure to stocks that have been running up on AI excitement. This change that you expect coming, where you think that the spending on compute will peak sometime this year and that the direction of spending will change to some of these these, you know, companies on, on the periphery. Do you have any thoughts on what that might mean for someone's, you know, large-cap index fund? Are those same types of companies still going to do okay in this environment, do you think? I realize you don't cover them, by the way.

Yeah, we could talk about qualitatively. I think over the past 10 years, just using the cloud investment as a context, because companies haven't found a systematic way to forecast cloud spend. When this peaking happens, investors are always surprised and they always panic and you have a correction. And I think this time could prove to be— we could see the same kind of reaction from investors. And if I'm correct and that were to play out, yeah, some of these big indices that are, have a significant exposure to AI, we see a sell-off. And when, when there is a panic, everybody sells everything. I think it's very typical.

Yeah, they usually get it wrong in both directions, right? They say they get too excited on the way up, right? And then they say, wait, things aren't going to grow at this rate forever. And then they sell to the point where, where it's, where it's too cheap. And so that, that gets you concerned about what could be like a big sell-off. And let me ask you the same question, but let me now ask you about the Philly Semiconductor Index, because one of the really surprising things about this bull market is you've seen, all right, stocks are running up, AI is a big part of it, and that means chips. And here we have this index focused on chips. Wait a second, I thought this index would be doing better than it has been doing. I mean, the Philly Semiconductor Index, correct me if I'm wrong, that has not been lighting the world on fire because some of the stocks in there, I mean, you got like an Intel and companies like that that are dragging things down, or some other smaller chip companies that are not at the center of all this excitement. Am I right about that? And what do you think this would mean for the, for the Philly Semi Index?

Well, let me put it in perspective.

Over the past 10 years, I kind of forgot the full spelling, the SOX as they say.

Yeah, it has over the past 10 years traded at a 10% premium over S&P 500 forward-looking PE. SOX has over the past 10 years traded at an average of 10% premium. Now, what about the past couple years since AI took off? Well, that premium has gone up to like a 25%. So we have had a doubling of a premium applied to SOX index relative to S&P 500. So some of these names are very expensive or trading at a high multiple. And it reflects the AI evolution, which started in late '22 or sometime in '23. Now, there has been some pullback since last July, but that premium is still at twice the premium it has been in the past 10 years. And this is where, like, if I'm correct and the cloud spend were to peak middle of this year, I think that's gonna cause a lot of panic and that premium is gonna come down.

Let me ask you something now, long term, what do you think that people are going to be surprised about on the subject of AI? And I'm talking about over the next maybe 5 to 10 years. And this doesn't have to just be investors, this should just be ordinary people. And it doesn't have to be something that means whatever to anyone's portfolio. You know, everybody thinks they have a handle on what AI does. What is it going to do 5 or 10 years from now that people are gonna say, wow, I never saw that come. That's, you know, that's blowing my mind.

I think increased automation by accelerating the compute time, it's all positive. It will help different verticals. It helps with new way of corporations doing their work, or maybe they come with different services. But I think the most surprising factor will be, which based on my own experience going to the '90s and we had the PC and the Wintel Alliance and then servers in 2000 and the internet. I think this time the application that would lead to a scale of AI is gonna take longer to develop. You're not gonna have a one killer app that would capture the economic benefits of AI. It's gonna be many smaller killer apps and it would involve a small contribution to increase productivity.

It's interesting you say that because I feel like the internet took, you know, 20 years. I mean, everybody was very excited about it in the '90s, and I don't think it was immediately life-changing for most people. And it wasn't until, you know, the, the, the, the iPhone and the later generations of the iPhone and the cloud services and Netflix and all these other things that people's, you know, lives became much more online and, and changed in a lot of ways. Are we, are we talking about it's going to take at least that long, you think, for people to see the, the main impact of AI?

I don't know is a short answer, but if I were to go back to the 2000s, well, eBay was formed, Google was formed, e-commerce were formed. Those are the clear application, an online exchange of goods or a search. And then Apple and how Apple use iPhone to enable download of apps. And these apps have enabled services provided through your smartphone. To me, those are very clear and defined application of internet. Now with AI, we may not have a very distinct or visible application, but we may have smaller ones, but it's not gonna be as revolutionary as iPhone was to the download of apps. It may be smaller applications.

Thank you, Mehdi. And I want to thank— let's see, I want to thank the Six Million Dollar Man for sure. Jackson Cantrell is our producer. He was off this week. I want to thank Alexis Moore for handling the audio production. And thank all of you for listening. If you have a question you'd like answered on the podcast, just tape it on your phone, use the voice memo app. You can send it to jack.hough— that's H-O-U-G-H— at barons.com. You can subscribe to the podcast. I'm giving you a lot of instructions on how to listen to something you're already listening to, but you can subscribe, I gather, on Apple Podcasts, Spotify, wherever you listen. If you listen on Apple, you can write us a review. Thanks and see you next week. Data is everywhere. When orchestrated properly, it sings. At Morningstar, We analyze and enrich data, making it actionable and powerful for you. Morningstar. Where data speaks.