Don't sleep on (analog) semiconductors

0:09Alex King:Welcome back to Investing Experts, everybody. It is always great to talk to Alex King from Sestrian Capital Research, who runs the investing group on Seeking Alpha called Growth Investor Pro. We had Alex on not too long ago, and the headline out of that conversation was, don't sleep on software. Alex, what are you thinking about as we head into the last week of September, the last couple of weeks of September? How are you thinking about, we saw a strong rally out of tech yesterday. How are you thinking about the tech space? What are you looking at? What are you thinking about there? Well, first of all, hi, Reena.

0:47Thank you, as always, for inviting me back. I'm always happy to talk for hours and hours and hours on these things. So I appreciate the opportunity. Thank you.

0:54Alex King:And we're grateful to have you. Yeah, you say that to everybody. So last time I said, yeah, don't sleep on software. Wait a minute. Software's done really well since then. So it's nice to be right sometimes. I would say at the moment, don't sleep on semiconductor. And the reason I would say that is because, you know, the sector has fallen so far from fashion. You know, there was a huge run-up in chips between the post-liberation day lows in April 25 and around June, early June in 2026. And then, as everybody knows, the Situational Awareness Fund was smashed against the rocks, I personally think, as collateral damage in a big fight between Jane Street and Citadel.

1:35It's Citadel 1. And with the decline in the Ashen Brennan positions, an awful lot of damage was done to chip stocks and to chip-adjacent stocks of the various so-called data centre bottleneck traits. And since then, if you had calls in any of those names, if you had been buying in the latter stages of the rally and you sort of hope for it to recover soon, your positions are destroyed. And sentiment has, retail sentiment at any rate, has been destroyed in the sector and nobody wants to look at it, which of course is the opportunity. So I think semiconductors are pretty interesting right now. We can talk in depth or not in too much depth about those.

2:18But so I would say number message, don't sleep on semiconductor and don't sleep on analog semiconductor in particular.

2:24Alex King:Yeah, get into it. Let's get into semiconductors. What do you like about them? What are you thinking about? How would you encourage investors to be contextualizing the news out of that space? Yeah. So if we step back and we look at the fundamental economics driving semiconductor spend, it's pretty simple. You have these hyperscalers that over the last 10, 15 years have built up this absolutely colossal cash pile, all of them, Meta, Google, Microsoft, all the names we're familiar with. And these companies just didn't know what to do with the cash. They didn't really pay much of a dividend. There were some buybacks here or there, but in essence, they were just building cash piles larger than most any country I can think of.

3:08And that was all great for their stocks. And along comes then this huge transfer of value out of the hyperscalers balance sheets and into component suppliers. And that's driven, as we all know, by AI CapEx. So along comes the next generation of computing, which LLM models were all designed and built in a world where nobody really thought about constraints of memory or processor cycles or power. Because when you're building these things in a lab, not at scale, you know, these things aren't an issue. To recount the history that we're all familiar with, ChatGPT launches chatbots and their more complex derivatives become a commodity item that consumers begin to use, that enterprises begin to adopt.

3:58The OpenClaw application gets launched into the wild. Then all of a sudden, things like co-work also become prevalent and token usage, token demand explodes. And with that, demand for memory, processor cycles, and power explodes. and with that you see a wholesale transfer of money out of those swollen balance sheets of the hyperscalers and into the balance sheets of the chip companies and the component suppliers and people like Bloom Energy who step in to fill the void where the power grids just can't supply capacity. Okay, so we all know this and then first of all along comes now a big narrative that says well, maybe all this stuff is overblown.

4:42Maybe we don't need this stuff in the end. And maybe this law proved to be a bust, just as was the fiber overbuild and the data center overbuild in the late 1990s. And all of a sudden, sentiment flips negative. And then in addition, we get this whole, maybe AI will kill us all next week or next month, and then sentiment starts to sour again. Now, remarkably and not coincidentally, this all coincides with the stocks having sold off quite heavily. And if you look across that data center bottleneck group of trades, you'll see even the semiconductor sector ETF, SXX or SMH is down, I think from memory, something like 25 % from peak to trough from June down to the July 31 lows.

5:29Some of the names, the more familiar ones, your Microns, Intels, and AMD in particular, have recovered. So AMD put in a new high yesterday. But a lot of them are still bumping along at the lows. And I think the sort of bravery that was encouraged into retail investors by all the usual talking heads on social media is gone. Because lots of those people have bought calls or bought late in the rally and, again, just lost a whole ton of money. And with it, vowed to never go near the sector ever again. Now, let's think back to earlier this year. This happened in February and March in software. software is bumping along the bottom, massive decline from the September 2025 lows, and we were all being told that AI meant the end of software.

6:17And of course, you don't need enterprise software ever again because it all vibe code our way to an enterprise accounting system. Funnily enough, that turns out to be not true. And as we talked about on the last time you kindly had me on the show, software was in the ascendant and it continues to be so. And I think we're there in chips again. Some of these names have run. I personally wouldn't chase AMD up here. It can go higher, anything can happen, but the easy money has been made. But if we look at the less flashy, less well-known, poorly understood analogue and mixed signal sector, I think there are some pretty interesting opportunities.

6:51So do I think Nvidia can go higher? Yep. Do I think Intel can go higher? Yep. Do I think ARM can go higher? Sure. Micron, sure. For sure. All those things, I think they can go up. And for disclosure, I'm long the semiconductor sector via sector ETFs, which covers all those But there's a very interesting niche in analogue and big signal that's driven by robotics. So robotics is the physical AI trade, if you think about it that way, as some people do. And we are seeing robotics start to be commonplace, not just across industrial settings, but we are seeing the beginnings of the humanoid robot market.

7:33and it's easy to get carried away with this. You know, we're all going to have robots brushing our teeth next week. It will happen more slowly than that. But humanoid robotics is a real thing. It is starting. And if you look at the semiconductor component demand for even a fairly run-of-the-mill humanoid robot that's produced by your Boston Dynamics or Unitree or any of the other companies doing this, the chip component and the passive component requirements is absolutely enormous. The joints that these things use are complex. So the number of motors, motor sensors, power regulators, power converters, all these bits and pieces, the unit count of component is huge.

8:14And these things aren't produced by NVIDIA. You know, they're not produced by Intel. They're not produced by Micron. You do need processor. You do need memory. You do need storage. And it's all those things. But you also need analog components. And for anyone who isn't familiar, analog components just mean anything in semiconductors that interacts with the real world, the real analog world, where values are in a continuum, a waveform, they're not in ones and zeros. So processes and memory, purely digital, only deal with ones and zeros. A temperature sensor has to deal with an analog concept, temperature, turn that at some point into a digital signal for consumption elsewhere.

8:47So the analogue mix signal sector is populated with chips, with vendors in that sector. And the leading names include people like Texas Instruments, On Semiconductor, and then you have Passive Suppliers, VCA, Intertechnology, names like this. And this sector isn't well understood, it's not well known, but I think these stocks are starting to bottom out. Very good financial performance. We have a note out today on Seeking Alpha Premium covering Texas Instruments, TXN, I'm long that stock. And if you take a look at that note, you should be able to see it on premium pretty soon. You can see great revenue growth, really good cash flow margins, nicely delivered balance sheets, and yet the stock is sort of bumping along close to the 200 day and looks like it's holding up.

9:34So I think there's a good opportunity there. And I think we're going to see grown up institutions and funds accumulating these names at a time when retail has been told to be fearful of semiconductor. So that's my take on that sector and subsector.

9:47Alex King:I appreciate that. Why do you think that part of the sector is so misunderstood? Well, it's hard. Okay. It's boring, first of all. I mean, does anyone really know what a firehistor does or care? Is the CEO of Texas Instruments on stage in a leather jacket with lasers talking about the factory of the future. No, I mean, they are old line chip businesses. And if we rewind 10, 15 years, most people didn't invest in semiconductors because it was too technical, too weird, too hard, too volatile. People can sort of conceptualize software, they use software every day, but it's hard to conceptualize chips.

10:31Everyone's piled into NVIDIA There are two reasons. One, because anyone can read the numbers, look at the chart and say, fantastic business, fantastic stock. But you also have a very charismatic CEO who can both talk to engineers credibly and talk to the general investing public about why their bits and pieces are essential to the future. It's an easy stock to understand. But Texas Instruments, OnSemi, go look at the website and I guarantee you'll be falling asleep within 2.4 minutes. So people don't like to put the work in and or that they find it off putting and they think it's too technical.

11:13And because I can't understand the product, then I'm going to be very careful about going near the stock. In addition, VShay is a great example. Take a VSH. And again, I'm long stock. This is an old line and storied business. It produces really low value and pretty dull components, but it produces them at high quality and has temporary monopolies, if you like, in certain important segments. And so it's never going to set anyone's imagination alight. But if you can cool off, just look at the numbers, look at the stock charts and build a sort of basic understanding of why these components are important to certain trends in the industry.

11:56So robotics is probably the next most important physical trend for chips. Then I think these things make more sense. Analog also, even amongst semi-doubter people, inspires a bit of, you know, thanks, I'll leave that alone. And that stems from, if you have a dangerously small amount of knowledge about semiconductor design, as I do, you know that chip design is hard. All chip design is difficult, but analog chip design is really hard. And that's because the real world is weird. Ones and zeros are much easier to model than the real world. And chip design and manufacture today is difficult anyway because of the tiny feature sizes and the power requirements and the cooling requirements and all of that.

12:38That's hard anyway. But if you then factor in the fact that, again, the real world is an interface and we have to deal with unpredictable moving objects, waveform data, have to operate in unpredictable temperature and weather environments, it's really tough to design, manufacture and understand. And still today in analog semiconductor design, you still have an element of magic in the design. You still need some human intuition to solve some problems in analog design. And so it's a somewhat esoteric area of the industry from a design perspective. The manufacturing processes generally use older fabs, previous generations of digital fabs that have been adapted.

13:29and that again that's a bit off-putting to the average investor they go well if the hot new thing is you know all about whether intel's new uh tiny feature size fab can be successful it's that's the thing in chips why do i care about this 10 or 15 year old fab over here that's churning out boring products so it tends to get overlooked and you know i don't suggest for a moment that the ultimate potential of a vshay or a texas instruments is anything like where nvidia could get to or Broadcom could get to. But as an investment opportunity right now, I think they're overlooked, forgotten. The stocks have been hammered with the Q2, early Q3 sell-off, and I think are a nice risk-reward place right now.

14:08And why I say that is because with that proximity to the 200-day with all that, you can set a fairly proximate stop that's 10%, 12%, 13 % below where the stocks are today. And as long as you don't size your position stupidly, if you're stopped out, so what? You lost 10%, not the end of the world. But the upside, there's two, three, four X, maybe more upside in these things in exchange for that 10, 15 % downside. So again, I think risk reward perspective is pretty good for a long opportunity right now.

14:36Alex King:So when it comes to risk reward, how would you share, how would you articulate your feelings about Texas Instruments in particular? What's your case for being bullish and what risks are afoot in your mind at the very least? So bullishness, I would say as follows. If you, again, if you get a chance, and I'll post a link to this when this podcast is published, take a look at the note we have out on premium today. But the bullish case goes like this. You've seen material revenue acceleration, you know, from negative revenue growth quarter on prior year quarter a couple of years ago to, you know, substantially double digit quarter on prior year quarter growth right now.

15:15And growth has been accelerating and the Q3 guide is for further slight acceleration. They're clearly bullish. TTM basis for memory, TTM revenue growth is about 20 % right now. And again, a guide for Q3, we'll see a slight acceleration of that. That's bullish clearly. Balance sheet looks really good. So leverage was, I want to say 5.5, 5.8 times. Trading 12 month operating cash flow minus CapEx a couple of years ago. It's now down to, again, from memory about 1.3 times. So balance sheet looks really nice. And there's plenty of cash and liquid short-term investments on the balance sheet. So revenue good, balance sheet good.

15:51Margins, I guess on the face of things, are bullish. Cashflow margins are really high, a big uptick in that in recent years. Margins in chip companies you have to be a little bit careful of because even though if you're in short trousers, you've been told that chips are no longer cyclical, but I have news, chips are still cyclical. And when chip companies see a combination of very high revenue growth and and also high cash flow margins, you have to sort of sit back and go, are we approaching peak cycle from a fundamental perspective? And that may be true. So if let's say the AI doom trade is true and this stuff all comes to a screeching halt, if in fact robotics turns out to be a bust and that comes to a halt, then it's going to turn out that, well, margins are at peak.

16:41And in fact, we've got manufacturing surplus, not shortage. And so, you know, perhaps this is a margin rollover point. And you always have to be careful of that when you're investing in chip companies with such high cash flow margins. But that's just, you know, have that in the back of your mind. It's not a reason, I think, to not invest today. So bullish revenue growth, sector opportunity for volume growth, unit volume growth into robotics, balance sheet strength, and probably margin is a positive for now. bearishness, you know, as follows, really, which is possible peak margins. But beyond that, you could say, well, where are we in the overall market cycle?

17:25You know, are we in the early stages of a bull market on a longer term timeframe? Probably not. You know, are we near the end? Is it going to fall off a cliff tomorrow? Probably not. But are we, you know, later part of the bull cycle? My guess is we are. So you'll see, you know, your recession needs to say the US economy is going to collapse next year and inflation is going to moon or deflation is going to crush everything and the stock market is going to fall off because it must do because it's gone up so much since the 2022 lows. And they may be right, but probably life isn't as dramatic as they say.

17:58But are we approaching a longer term top? Yes, my guess is we could be. And so I think you just have to be a bit careful piling into relatively high beta stocks if that's where we are in the overworld bull market. Again, I don't think that's a reason to not invest for me. It's a reason to go, you know, just have some context. You know, don't think that the whole world is in flames and ruins and everyone's super negative on stocks and therefore it's probably a very good time to go in. You know, there's been a temporary setback in chip stocks. Some people have been miserable at chips. But overall, as we can see, the market's still very bullish.

18:33It's been bullish for a long time. And so you just have to be a bit wary that you don't go, you know, all in right as a market top approaches. So I would say cycle risk, you know, in semiconductors, market cycle risk. Those are the sort of biggest things in my mind. Of course, you have company specific issues. So could something go wrong with each one of these businesses? Of course. Could no buyers show up for V-shade? Of course. But you can solve those with a stock chart and a stop. So I think the bigger things are possible peak chip margins, you know, potentially approaching the latter stages of an overall bull market.

19:08Alex King:Are there any particular metrics or any particular data points that you've kind of changed around given where we are in the cycle, where we are in the markets, or are you steadfast in your process? It's not dependent on the navigations of the market. No, no, we're completely market neutral in our work. So a view on the market is driven entirely by price and volume. So obviously there are a thousand other inputs to that, but our take is that there are probably some people who can take all those inputs, macro numbers, micro company numbers, throw all that into a jumble machine and from that work out where the S &P is going to be in a year's time.

19:53There probably are some people who can do that, but I know that we're not amongst those people. And so our view is, look, you can make reasonable estimations of market bottoms and tops. And you can spot a market bottom because the S &P is generally trundling along at the lows and volume is picking up. It tells you that sentiment's bad, but large accounts are likely to be buying. And you can make a reasonable estimation of market tops because something similar happens, which is you can't get past a certain level and volume is picking up, which tells you that probably large accounts are selling.

20:27And obviously nobody's perfect. Nobody can call these things for the day. But broadly, I would say our analysis of market cycles is not bad. And if I look through the time we published on Seeking Alpha, we've generally got the tops and bottoms pretty much right. So I feel okay that we can spot tops and bottoms. And I also think, well, there's always a bull market and a bear market somewhere. We cover multiple sectors in our work. So we cover tech extensively, as you know, we cover defense, we cover biotech, energy, space, and a whole bunch of other things. And so our view is, look, you're only wrong if you're constantly looking for a bull market in the same things, be that US equity indices or tech or energy.

21:10But if you can learn to rotate through sectors, then you can do well, good market or bad market. So our overlay goes like this. Where are we in the market? We use the S &P as a good proxy for that. Is the whole market selling off? Okay, well, in which case, very simple, just have some short S &P exposure. And you don't need complex instruments to do that. You can either just take a short SPI position, or you can take a long SH position, a one-times inverse ETF. You can dress that up with all manner of more complex things, but just long SPI in a full market, long SH in a bear market, that will work just fine.

21:45So you've got overall market positioning. And then within sectors, we both manually and algorithmically look at which sectors are on the rise and which are on the decline. And we do that with the sector ETFs that sit within the S &P. So, you know, your XLK is for tech, XLE is for energy, XLV is for health care and so on. And if you look, if you just chart these things out for free, as anybody can use the Seeking Alpha site to do it, use 10 or 15 sector ETFs, you can see they all run to different profiles. And by and large, often but not always, if you look to see which sectors, which sector ETFs are sort of bumbling along at the lows and everyone's hating on them and nobody wants to own them, there's a good chance that that particular sector will be up a fair bit in the next nine, 12 months.

22:31And similarly, if you look at the sector that everybody's talking up and it's going to go up forever, there's a good chance that's approaching a top. So we do that manually in our work. And we have a really super successful algorithmic signal service called Signal Flow Sector Rotation that does this quantitatively. And by either approach, we found a good way to navigate through market cycles. Obviously, we're not perfect and obviously we make mistakes all the time. But at the highest level, it works pretty well for us. And that process applies, you know, bull market, bear market. Sideways markets are probably the hardest for our method.

23:06And the best thing you can do in sideways markets, if you have the stomach for it, is, you know, sell options for premium. I personally don't like to do that. Sort of options scare me a bit. But, you know, trending bull or trending bear market holds no fear for us.

23:20Alex King:You know, you were talking about hyperscalers. And I read this article yesterday on Seeking Alpha, and the headline was from the Apollo Global Management Chief Economist, Torsten Slocken. He said that the credit case for the largest cloud and AI infrastructure builders rests on one widely shared forecast, that their combined operating cash flow will triple from about$600 billion to$2 trillion, which is exactly what my tax accountant is looking ahead for my portfolio. So I was wondering what you thought about that. I think that the rational human response to all these things says, well, we're clearly in a bubble.

23:59Clearly, everyone's gotten overexcited. There's clearly been an overbuild. And of course, none of this stuff is going to come to fruition anytime soon. And if you don't at least consider that case, you're an idiot and you're not saving money. If that turns out to be true, okay, well, it's pretty easy to navigate because if you just have a healthy dose of scepticism. And if you own Google, for instance, as I do, and you just watch the numbers and you watch the stock chart, and you don't have a sort of quasi-religious belief that will go up forever, then guess what? If the numbers cool off and or the stock cools off, well, sell Google and get out.

24:36I mean, it's not hard. I think the more unusual case, you know that great Musk quote about the most entertaining outcome being the most likely. It's just not necessarily the most entertaining for those affected. The most entertaining outcome here, I think, is that actually the CapEx forecasts tend to turn out to be true and the cash flow forecasts turn out to be true. In other words, I think the most entertaining outcome will be actually all these hyperscalers do achieve a return on CapEx. You know, they do turn all that invested CapEx into accelerated revenue growth and cash generation and then deleverage.