[REPLAY] Jeremy Raper: Cycles, AI, and Activism

Brandon Beylo (0:00)

Jeremy, it has been over a year, I think, since we've done our last podcast. I know it was at the, so I moved. So it's at the last podcast we did, I was at my old house.

And so we are in a new house, and I have a four month old daughter. So-

Jeremy Raper (0:18)

You did mention that in the intro, congratulations.

Brandon Beylo (0:20)

Yeah, I've got-

Jeremy Raper (0:20)

That's awesome, that's the most important news.

Brandon Beylo (0:22)

I've got two girls.

Jeremy Raper (0:24)

Oh wow, there you go.

Brandon Beylo (0:25)

Yeah, one's two and a half, and then the other one, like I said, has kind of turned four or five months.

And it's insane. It's two is so much harder than one. Oh my gosh.

Jeremy Raper (0:35)

Two girls are in it. Two girls, I got one of each. So, yeah, I got a boy and a girl.

Brandon Beylo (0:39)

So- I mean, girls are way more chill. Like our friends, that voice. They're so chill.

Jeremy Raper (0:43)

They're aggressive. Yeah, they're terror full time, but it's fun, man. It's like, that's what it's about. So I'm glad you have two. Are you guys done?

Brandon Beylo (0:51)

Or you- You know, it depends on the night.

If the baby sleeps well, then maybe there's a chance. If the baby doesn't sleep well, then we're done.

Jeremy Raper (1:01)

It's not- Yeah, give it a while. Four months is pretty still hot zone, right? You got to give it a bit of a cooling off period.

Brandon Beylo (1:07)

So- Yeah. So what are you going to do?

We're going to talk a lot about Australia. You kind of made, I mean, kind of a public announcement like, hey, there's so many ideas in Australia that I'm going to be writing about them again on your blog.

And then I also want to discuss this AI semiconductors trade through the lens of cyclicality. And you were kind of one of the earlier people to basically go on Twitter and say, hey, I think these things are still cyclical, and cyclicals are going to do what they always do. And I think you're starting to see that play out over at least the last couple of weeks, maybe one to two weeks. So let's start. Maybe we'll start there with AI.

Jeremy Raper (1:54)

Just because I was trying to make a pretty nuanced point that maybe got lost in translation. The point was not that this is not post-cyclical or pre-cyclical or still cyclical or not. So there's two debates to be had. There's, is this still cyclical or not? The fundamental debate, because it's everyone who's examined, not semiconductors specifically, but I was looking at memory specifically, because memory companies have been a very long-term participant in the markets. They've been around for generation, essentially, and they've been as cyclical as long as they've been listed companies, right? So if anyone who's even looked briefly at, say, Micron or SanDisk or some of these companies will see that there's huge volatility in earnings, the long-term average return on equity of Micron, for example, I think one of Micron's is telling is about 10 percent. So more or less the cost of capital.

The cost of equity, that is. Nothing says a perfectly balanced commodity market like your long-term return over 10, 15 years being equal to your cost of equity. So when people get on here and say these are now post-cyclical, no longer cyclical businesses because something fundamental has changed structurally.

So I don't cover memory as closely as I used to, but there are elements of that thesis that are believable. So there used to be many, many memory manufacturers, okay? A lot of them went bankrupt. A lot of them were Japanese. They went bankrupt, got taken over by Micron Board, a Japanese company called Elpida, for example, during the financial crisis. There are a couple of Taiwanese manufacturers who went under and got taken out. And so the industry is consolidated from 10 players, whatever it was 20 years ago, to maybe four scale players today and three in certain sub-segments. So you can make an argument that, okay, it's a much more oligopolistic market, therefore, earnings structurally should be higher.

So you have one argument which you can debate, and there are people who are looking at the evolution of the market and the products they make and can make that call. The point I was making was actually more about how you trade these stocks, okay? So whether it's going to be no longer as cyclical as it was, or less deeply cyclical than it was, or even not cyclical at all, and maybe earnings have graduated to a new high plateau that they otherwise never earned before. The problem is, they're going to go from earning 100% ROE, 100-150% return on equity, which is the current run rate for Micron or whatever, because they're just making insane, crazy, peak, peak, peak, peak, never before generated earnings, to something that is going to be much less than that, right? So the new normal might not be 10% ROE, it might be 30% ROE, still amazing. It might be 40% ROE. The point is you're going from 150 to 40 It's a massive air pocket for the market. So how do you trade it? Well, of course, in the market, oh shit, Sandisk, EBITDA, I'm sorry, I can swear, right? Can I swear?