Full Transcript: First Principles with Andy Constan - Episode 1

Andy Constan on Investing Through Market Bubbles

Justin: Welcome to the first episode of First Principles with Andy Constan. Andy, thank you so much for doing this with us. We’re excited for you to be here with us today.

Andy: Yeah, I’m excited too. It’s gonna be interesting.

Justin: There are a lot of shows out there that give people opinions on what’s going on in the markets, and we’re gonna do some of that on this show. But I think the overarching goal, what we hope to accomplish here, is to have a discussion that goes deeper on a lot of different subjects. And what we really wanna do is focus on lessons and frameworks behind what is happening in the market so we can help our listeners and our audience develop a better understanding of what actually it is that drives the things in the markets and drives the economy.

And so it’s gonna be less about what to think and more about how to think, and we couldn’t think of a better person to launch this show with than Andy. So it’s gonna be fun, it’s gonna be exciting, it’s gonna be informative. And we appreciate people watching and supporting our guests like Andy.

We’re gonna use your first, or a recent Substack post that you did, where you were talking about bubbles. And it’s a very, I think, timely post. I think it’s on a lot of investors’ minds in the current market today. Are there bubbles forming around us or are there not? But I think, you know, today’s discussion is gonna be about talking about what bubbles are from your perspective, how they form, how investors can recognize them, and how we see things playing out in the current market and what, how investors might be thinking about things.

So that’s the topic of today’s discussion. That’s the first, sort of, topic of the show. And yeah, so let’s just get into it, Andy. How do you... When you think about defining a bubble, I mean, in retrospect it’s always everybody’s like, “Oh, yeah, it was obvious we were in a bubble and we all should have seen it,” or “We did see it.” But how do you define a bubble and how do you think we can see it maybe in real time?

Andy: Right. So I don’t think you can. I don’t think it’s easily defined in foresight. It’s incredibly easy to define in retrospect. And I think that when people hear that term, particularly in this day and age where everyone wants to know are you a buyer or a seller, are you long, are you short?

Well, and those are important, very important things, obviously. But when you talk about a bubble, it’s hard enough to appreciate what’s different in a bubble-like regime versus... That’s hard. It’s impossible to determine when a bubble’s gonna pop. And so I want — this conversation has to be about what is different in a bubble regime, not, “Hey, the bubble’s about to pop.”

Like, that’s a call. As a trader, as an investor, I’m gonna change my portfolio if I think I see something that indicates the bubble is popped or is about to pop. But I will tell you, that’s like the holy grail of investing. Being able to pick the top not only of just a general market, but an actual bubble?

That’s the holy grail. I mean, nobody can... You can’t find the holy grail. It doesn’t exist.

Justin: How do you think you would separate a market that is expensive and maybe over-owned by investors from one that is actually seeing bubble-like characteristics, but that’s being defined or influenced by a real technological breakthrough?

‘Cause that’s kind of where we are today, right? I mean, the market’s expensive. A lot of people are owning these AI-related stocks. They’re all baked into the Mag Seven to some extent. But then you have this AI breakthrough that’s happening. So how should an investor sort of grapple between those two things, do you think?

Andy: Well, I mean, the first thing you said is something that you hear a lot about in markets. Overbought, expensive. I think the first thing I do as an investor is just respect the fact that the current market prices are the current market prices, and that they’re neither expensive nor cheap. They’re neither overbought or oversold.

I mean, just even those words — overbought or oversold — is not the way I think about things in general. And the reason is, is because at this very moment, every single investor on Earth has — by and large, I mean, I guess maybe somebody’s about to trade — but by and large, every single investor on Earth, no matter what their horizon, has exactly what they wanna own.

If they’re short, they have exactly what they wanna short. They have the leverage they want. They have the cash they want. Everybody’s at equilibrium. And so there is no such thing as overbought or over-owned or oversold. It’s just sometimes various cohorts are doing things in markets that appear unsustainable.

And so the idea is that when people use the term overbought, they mean some of them — whoever they are — have bought more than they should have and are likely going to have to liquidate. That’s a classic definition of a rout. So firstly, I just humbly say that when I look out at investing, I think no one really knows. The best thing to do is to recognize that you don’t know which way the markets or the economy or technology or anything else is gonna go.

And so you just own a passive, well-diversified investment pool, a set of investments, and just go about your life. So that’s the number one principle when I think about all this stuff. But I think there are differences — significant — assuming every market price is accurate, there are still differences in the sort of regimes that exist. And a bubble regime is what I think we have entered into, so let me get to that.

So what does that mean? Well, bubble regimes have certain root causes and certain sets of characteristics that you can loosely define as bubble-like. And again, it’s not a precise thing. Like, no one can predict that we’re actually in a bubble until after the fact. But for me, I’ve seen a number of them, and I’ve studied even more. I’ll only cover the things that I’ve actually seen in my real career, but I’ve studied a lot of other things. The environment today looks very similar to other bubble environments.

And I look at four of them that we’ve had in my career. The first one was really kicked off in 1981, 1982, and persisted through the crash of 1987. And it wasn’t a bubble the whole time, but it was a regime that ultimately delivered a bubble.

Similarly, the internet bubble, which started — I like to say it kicked off when Netscape Navigator was invented in 1995. And many of us became aware of the power of connectivity. That wasn’t really a bubble for a number of years, and then became a bubble.

2005 to 2008 didn’t really start as a bubble, became a bubble. Then we go into things that are not risky equity or credit assets, but are bonds. Government bonds after the GFC, where short-term interest rates were set to zero globally — went in one direction for an extended period of time. They rallied for an extended period of time. And then ultimately when COVID hit, they became a bubble. Rapidly accelerated to basically zero interest rates. And that bubble resolved the following year. And then today, a lot of the things that I look at are bubble-like.

So let me get back to how I think about those. Those are the five cases I’d like to sort of think about. Let’s get started.

These types of environments typically start with something new. And something new in the internet boom and, if we’re in a bubble today, the AI boom — was technology, was some new thing. And you can look back to — again, before my time — you can look back to a variety of industrial revolution technological advancements. You can look to China where they made a huge productivity move bringing people from the farms to the factories. You can look at major productivity changes, as they tend to lead to some sort of bubble-like equity outcome.

In 1982 through ‘87, the new thing — it wasn’t really new technology. We had just ended a major inflationary episode. The United States deregulated the financial industry, in particular the savings and loan industry. There was a small technology advancement, which was the invention of Lotus 1-2-3, which allowed people to easily scenario analyze companies. And there was the innovation of Mike Milken in terms of creating a market for high-yield debt. And that kicked off the thing that was really new to the markets, and that was the LBO. And so when I think of the 1987 crash, I think it was impacted by lots and lots of things, and all bubbles have lots of things going on in them. But in 1987, that bubble was driven a lot by a trend toward the LBO.

We know what kicked off the something new in ‘95. In 2005 through 2008, where you had the housing boom, we had a period of time where globalization had essentially ended inflation. And with the end of inflation, financial conditions could be left very accommodative with no risk of inflation. And that created a levering up in banks and in the housing market. So the new thing was the end of inflation and globalization. And that was a driver for what ultimately turned into a bubble. And by the way, this is what I think. I could be wrong. This is just how I’m thinking through these things.

Now, as I said, ZIRP and QE drove a bubble in bonds, which ultimately peaked when the economy was shut down during COVID.

And then today we have the ChatGPT moment, which — I don’t remember what you thought about it — but I thought on January 10th of 2023, when Microsoft made its investment in OpenAI. For many of us, we’d been playing with the first version, the first public version of ChatGPT. A new version had just come out. And for any of us who have done any sort of statistical analysis through their careers, there’s been a slow burn of regressions leading to neural networks, leading to machine learning, all happening as compute power increased. That’s been a 40-year slow burn in terms of what ultimately inflected with that pretty much one-off event. The AI trade has been one direction since then, basically.

And so I like to think of those as the precursor to the bubble behavior, which is either a significant regulatory change, a significant easing, or a significant technological development. Or lastly, a significant exogenous event. The bond bubble would not have occurred without COVID.

And then you have escalation events — that happens along the path of that framework, and that’s when you go into a bubble. And for me, those things were just an explosion of deals in 1987. In ‘98, the long-term capital easing ramped and escalated the tech bubble. In 2005, financial engineering, in particular tranched CDOs, tranched mortgage product — doubled, tripled, multiplied the leverage, squared the leverage, whatever you might wanna call it — and escalated the housing bubble. Obviously the pandemic itself was the final thing that caused the bond bubble to go parabolic.

And then I think we saw some unnecessary easing of financial conditions. Today we had a super hot inflation print. It’s been, I don’t know, 62 months since inflation is above target. And in 2023, and even in late 2022 before this whole AI trade got started, the central banks — in particular the Fed — eased to deal with financial stability around the banking crisis, the small banking crisis we saw in the spring of 2023, and they gave up on their inflation mandate, and that escalated this thing.

So those are the things. There’s the root conditions, which don’t have to be a bubble, but root conditions can become a bubble. Then there’s the escalation events, and then there’s the peaking, and I think we’re in that phase right now. We’re in the peaking phase. Now, how long that can last? Quite some time. You know, we saw Long-Term Capital got bailed out in October of ‘98. It took almost a year. A year later, the NASDAQ still had enough oomph to rally 60% in six months leading up to the final peak. So it can happen. It can take some time. So let me take a breath. Anything you want before I—

Jack: Well, actually, a few different things. One is I wanted to ask you — you were investing through it — did you have any takeaways from the Japan bubble, like of the ‘80s?

Andy: So let me be clear. I worked for Salomon Brothers. Salomon Brothers had a massive presence in all these markets that I just mentioned. And as a personal investor, you’re highly limited in terms of what you can actually do.

But one thing I’d remember doing is buying the Nikkei put warrants that existed in 1989 as the bubble peaked.

Justin: You got those cleared with compliance, right?

Andy: You had to hold them for like — oh... 60 days or something.

Justin: Okay.

Andy: It wasn’t like a trading thing.

Justin: Yeah.

Andy: Throughout my whole career on the sell side, you just can’t play. For good reason. For one, it’s distracting. I think that’s the principal reason why they do it. There’s the compliance reason, which is insider trading, all that, market manipulation, getting in front of clients, all that sort of nonsense, but I just don’t think they want you trading all day in the end. Or at least I thought that initially. In fact, I still do. So yeah, you can’t really do much investing, but I did get to see every other type of investor flowing through the market. But yeah, I mean, the Japanese bubble has many aspects of those same things. In particular, it was a housing bubble, it was an equity bubble, it was an easy money-driven bubble, it was all manner of things.

Jack: Yeah, what was interesting to me — hearing you go through those — it was ‘87. ‘Cause I wasn’t investing in ‘87, but when you typically hear people refer to historical bubbles, ‘87 is one people talk about as a crash, but they don’t really talk about as a bubble. So it was interesting hearing you talk about the events that led to that.

Andy: You have to, you know, most people don’t know that — from, if you bought stocks on January 1st, 1987, and you sold them on January 1st, 1988, you broke even. That the crash of 1987 just brought back, just gave away your 1987 returns. The peak of the equity market ahead of the crash was up, I think 31, 31 and a half percent ahead of the crash, and all it did is give back. So I think you can’t look at the 1987 crash without looking at the first nine months of 1987, which looked bubbly. But I do agree, not many people talk about that as a bubble.

I talk about it because I find there’s an interesting market mechanism dynamic there regarding portfolio insurance that maybe for a later time we’ll talk about. But portfolio insurance and zero-DTE options right now rhyme, and those had a very big impact on the ramping pre-crash, and then had a massive impact on the crash itself. So it’s just an interesting, to me, interesting dynamic.

Jack: Yeah, that was gonna be my question, ‘cause I think that was probably the most mechanical of the bubbles you would describe, maybe?

Andy: Yeah. There’s an aspect of every bubble that has sort of normal mechanical activity.

Jack: How do you think about — having gone through these — like, I’ve gone through a few myself and I always question, like, am I any better at this, or do I still get wrapped up in the whole thing? Like right now I’m talking about AI as a transformative technology, so it’s different than other things. Like, do you feel like you get better as you go through these, or do you feel like there’s a human nature to this that it’s just very hard to learn the lessons of the past ones?

Andy: Well, the problem is most people’s careers are relatively short, and so we don’t get to have many lessons over time.

But what I would say is without a doubt, human nature across these bubbles doesn’t change. Like, guys who were my age back then may have seen bubbles before and may or may not have participated in the bubbles that we’ve seen when we were in the younger part of our career. But the people today that haven’t had any experience — they’re not well-equipped to understand what it’s like.

And the bubble is perfect. The reason why a bubble regime is so difficult is because it plays precisely on human nature. You see your neighbor up 100% on some semiconductor stock, it matters to you. You are affected by that. You are not affected by them up 15% on their S&P when you’ve been in cash.

But if they’re getting enormously and suddenly rich — which is unique to bubbles. I mean, not unique. Of course there are people that sometimes get lucky, but when not only this neighbor but this neighbor and this neighbor are getting enormously rich suddenly, that only happens in a bubble and it’s incredibly compelling.

I don’t see how, honestly, I don’t see how we humans can combat that without incredible discipline. I don’t think we can. I think it’s one of those human nature things — bubbles beget behavior. That’s the point.

Jack: I remember the same thing like in ‘08, because I knew some people who were buying three and four houses on these stated income loans with nothing down and they were just making a killing, or at least on paper the houses were going crazy and it’s like, why am I not doing this? Why am I not buying these houses?

Andy: Yeah, I missed — I have to admit, I did not see that part, like the housing bubble per se. Yeah, I saw it. It just didn’t affect my neighborhood or my community, my friends and other investors. That said, credit investing and the way to make money using the securities that came out of that was highly speculative.

Jack: How do you think — can you talk a little bit about the late ‘90s again? ‘Cause you wrote in the piece about the analog between that, and you talked about the ‘94 bond market crisis and Greenspan’s pivot. Can you just talk a little bit more about the late ‘90s and the analog you see to today?

Andy: Right. Again, you start with conditions, and the conditions are — we had right before Netscape Navigator came out, which I’m pinning as being at the time a big deal.

Like, it was a sudden wake-up to the rest of the world.

One year prior to that, the central bank had gone in, had basically bankrupted the mortgage market, bankrupted the Orange County pension fund, and created a massacre in the bond market and then pivoted. And when they pivoted, they eased financial conditions in a meaningful way. So the money creation, the credit availability, the price of money started ahead of this ‘95 technological event very easy.

So that’s a good start. And so then you have this period of time between ‘95 and really ‘97 where people thought, “Wow, this is gonna be a big deal, and we’re gonna need all manner of capital investment.” And that capital investment started being funded and started flowing through to market prices and deals were getting done and so on.

And that was a period of time in which we all knew what was happening, but at the same time, there was no sort of like euphoria. It was just, “This is gonna be a big deal.”

So when I think about that, I think about the prelims to what we’re involved in now. And the prelims to that were — the stock market bottomed in, what, early October, maybe October 6th or 7th of 2022 after the Fed had, you know, gone through this hiking cycle and announced QT and started doing QT.

Bond market had sold off from its bubble. And all of a sudden we have the beginning of an easing cycle. And so that set us up to a point — you know — Meta was in the ‘80s and every stock that is now 10X that or 5X that were all double-digit stocks, and now they’re all deep into the triple digits.

All the Mag Seven, Nvidia. And then we had January of 2023, and that was the Netscape Navigator moment. Soon after that, we had the SVB crisis, and the next couple of years was a very nice runway for technology, but we hadn’t had the escalation, the meaningful escalation.

And that brings us back to ‘97 and ‘98. In ‘97 and ‘98, what happened? We had an Asian crisis because all this easy money flowed outside of the United States and went into Indonesian companies and Thai companies, and there was a speculative frenzy in Asia that unwound suddenly, causing US stocks to crash and then recover.

And then we had the long-term capital crisis in ‘98. And so when I think of those proxies, I said, “Hmm, October ‘97 we had an Asian crisis. October ‘98 we had a long-term capital crisis. That’s interesting.” 2025, we had Liberation Day. 2026, we had the war in Iran, and we had major sell-offs. Similar timing. So now you have those things. They’ve been resolved. The financial conditions have been kept easy because of the desire for a successful resolution of those things.

In ‘99 and in today, you’re having an expectations explosion, which I point to in one of my posts. In March, right in the middle of the Iran war, the chairman of NVIDIA came on and made a very, very aggressive announcement about CapEx. Now we all knew CapEx was coming, but it had a step change in expectations for semiconductors. Semiconductor stocks ended up falling, but that step change ultimately — when the war was resolved — immediately got recognized and started a parabolic move in semiconductor stocks that we’re still in the midst of.

And so you have to look for that step change in expectations, a world in which the main asset that’s in a bubble is flowing through for expectations for earnings that are simply never gonna be bad. And you had that in ‘99 as well. So we’re somewhere in that phase where you have massive expectations of immediate benefit from the technology that has resulted in a parabolic move in assets.

Jack: Yeah, and I do wonder — when we look back at this — I wonder if we’ll look at the announcement of Claude Mythos, ‘cause that was very tied to the move in semis recently. I wonder if we’ll look at that as a big event in terms of what ignited a parabolic phase here.

Andy: Well, I think that one lines up better with timing, but I think there’s a lot — I don’t know everything — but what I see, a lot of the bullishness on semiconductors is the step change in earnings expectations. We already expected 60%, 70% increase year over year in earnings. In March, you had a change from 60% to 70% earnings growth for the next couple of years to 100%. I point to that, but I think yours lines up better on timing, and that’s the way things work. You just have this constant — what’s the right word? — you prime the pump with the main thing, and then each time you get an accelerant that makes you go the next phase up.

Jack: Can you talk a little bit more about Long-Term Capital? ‘Cause you talked about that in the piece as what happened in the wake of that as something that might have ignited the last phase of that bubble. But can you just talk about what happened with Long-Term Capital? I mean, you’ve talked about bubble-like regimes. Is that the type of thing you’d expect to be happening in a bubble-like regime? Even though they weren’t really necessarily directly associated with the bubble.

Andy: Right. That’s an interesting point.

One of the things that happen during a bubble — you always look for what I call contagions that could either cause the bubble to extend or are consistent with the sort of post-bubble world. Long-Term Capital — it’s interesting.

One of the contagions you can have in a bubble is those who are fighting the bubble being bankrupted. But generally, those don’t have meaningful contagions because whatever they have to dump, they dump, and whatever they need to unwind — there’s so much liquidity around, there’s so much available capital around — losses can be absorbed by the system.

So during the period of time when a bubble is inflating, you rarely have contagion. So I don’t think the Long-Term Capital thing was caused by the stock market rally. There’s some tweaky little stuff about their vol position that probably had some impact, but it’s not really there.

Long-Term Capital was over-levered in primarily fixed income instruments and got a margin call. The problem is that the central bank massively overreacted. This is what they did — they arranged for the entire fund — which, by the way, the numbers are laughable how small they are right now. They forced, I believe the number was, they forced 13 banks or 11 banks called the consortium to come up with $1.3 billion.

That’s B for billion, not T for trillion.

Jack: That’s crazy.

Andy: Nothing. $1.3 billion to buy the positions that Long-Term Capital had and assume their positions. It was nothing. But they still cut interest rates significantly to make sure this didn’t become a financial crisis. It wasn’t gonna become a financial crisis. It was taken care of. That was that.

But they still did these... Not only did they cut, but they did surprise cuts. And this is from the same guy who, 40% ago, had used the term irrational exuberance to describe the stock market. He was cutting into a stock market that was up 40% from when he made those comments. I think that was the number.

And so that was like adding rocket fuel to the bubble.

Jack: One of the interesting things you had in this — we were just interviewing Cliff Asness yesterday. And he was talking about this period. And one of the things he said is, “Andy beat me to it.” He was like, “I was gonna talk to you guys about this idea that one of the huge differences here is this idea that tech was so much smaller in the ‘90s as a portion of the S&P or as a portion of the market than it is now, and this is a very different thing because tech is so much bigger now.”

And when he was gonna say that on the podcast, you had come out with a blog post ahead of it and talked about this. So I’m just wondering — why does that matter? ‘Cause that is a really, really important point. I mean, tech was very, very small going into the beginning of that bubble, and now tech is sort of at the beginning of this bubble a very big part of the market. Like, why does that matter?

Andy: I think... I saw somebody talk about how tech could become 100% of the economy, and it’s just like, no. No, it cannot. It cannot become 100% of the economy — ‘cause I’m gonna still want some Domino’s or Burger King or whatever. I’m gonna wanna go play golf. I’m gonna wanna get my hair cut. Whatever.

Jack: Until tech cuts your hair, I guess, which maybe someday is gonna happen. I’m not sure.

Andy: I do get a haircut occasionally, and a beard trim. Listen, I just can’t imagine... So anyway, the point being, there’s a bunch of stuff that the economy does that it’s gonna do, and it’s gonna require stuff that’s not semiconductors.

So the share of the economy matters. And so when an industry has no share, 4% share, and it can grow — it can 4X — well, that offers a real significant earnings and stock price upside. But if you’re already at 15X, you can’t go to 60X. You can’t get a 4X, because there’s just not a lot... Some of the pie slices of the economy are claimed. They’re never gonna be tech.

So I think that’s the point, which is if you’re starting at a small base, you can get a 4X return, but you can’t get a 4X return on a limited pie if you’re already a significant factor in not only investments, but in literal earnings share of GDP.

Now, does that mean that we can’t see another 5, 6, 10% share growth in earnings coming from the economy in tech over the next few years? We can. I don’t know where it’s gonna come from and who’s gonna pay for it. And so when I think about what the upside for this is...

I look at it and I’ve described this as a pie where great new technology — and this is true throughout history — great new technology increases the size of the GDP pie. It does it because people get a new tool and can generate more output with that tool. Full stop. That increases the size of the pie.

And a majority of that pie increase, the increase in that pie, can go to the tool maker. I got no problem with that. It can mostly go to the tool maker, and that creates an increased share of the pie. But unless that happens, any bit of share that technology takes comes from somebody else.

And so that speaks to jobs, it speaks to incomes. If technology is gonna take more of the share, everything else is gonna get less of the share. That means everybody’s income that is involved in that lesser share is gonna fall.

So who’s gonna be the customer to buy the new GDP from the tech? That’s the thing that I think is... It’s all about the S-curve in the end of the day. That’s what a lot of people describe it as. You can grow a lot when you’re small. You can grow steeply until you get to a point, and then your growth levels off. It’s all the same basic conversation.

Jack: I do wonder — we already talked about how you can’t time these things — but I do wonder, the idea that tech started bigger here, like, does that mean there’s maybe more potential for the bubble to get bigger and maybe also the more potential it’s gonna be more problematic for the market eventually? Because tech started out as such a big part, you know, before the thing even started?

Andy: Again, I don’t think it can double from a big base. And I do think it can double from a small base. So I think that is fundamental. Does it matter that it’s big on the way down? Sure, I guess a little bit in that it can fall harder, but I don’t think that’s... Well, I guess it’s possible.

I’m a big believer in statistical analysis that turns into neural network, that turns into machine learning, that as compute gets bigger and bigger and bigger, we can do more and more interesting things. That becomes AI. It’s been a 40-year journey for me to watch things happen in the world. Amazon has been able to anticipate my buying needs for 20 years now. They know exactly what I wanna buy, exactly when I wanna do it. They didn’t call it AI back then, but it was AI, right? So these things have been going on. We’re not going away from it.

Now, is the new version of AI that’s got people excited going to be extra disruptive? Maybe, or maybe it won’t work. I’m a big believer long term in this tool, and I do believe technology will continue — isn’t gonna shrink. The question is, is it going to grow as rapidly as expected? That’s really, I think, the only thing that we’re talking about here. It’s like talking about fiber. They overbuilt fiber. Now we use every strand. So I don’t think that’s the story here. I think it’s just a matter of expectations and pricing.

Jack: Do you think — you’ve mentioned some policy errors in previous bubbles — what do you think the biggest lessons for policymakers are? If you look back through the bubbles you’ve lived through, what are the biggest lessons for policymakers in terms of the mistakes they’ve made in bubbles?

Andy: Gosh, that’s hard. I mean, there are so many mistakes. I’m pretty cynical about the ability for the government to do anything but steal from the future and give to the current.

I guess — we had today someone, a senior figure in the Korean political system, suggesting that they are gonna tax Korean AI to deliver a citizen dividend. And so that’s an interesting thing. It’s a redistribution. It says, “These are the guys that are making all the money. We’re gonna take it from them and give it to those guys.” And so governments love to do that stuff. They do it all the time.

You look at housing — great example. Why was housing in a bubble? Lots and lots of reasons, but perhaps the greatest reason is that this country’s politicians have always favored homeownership and created subsidies for cheaper mortgages. And so when you do that, you have an impact.

So yeah, I think you’re in a world in which there is major disruption — so I do think this AI is gonna be disruptive — and in many bubbles we’ve seen disruption. What does disruption mean? It means people lose their jobs. The LBO bubble, if you call it that, was the hollowing out of the manufacturing sector in the United States. That’s what we did. The LBOs were — according to Dan Rostenkowski, who was the House Ways and Means chair, ultimately a criminal, I think — socially destructive. And I think we all look back and say, “Yeah, we lost our manufacturing.” We’ve got a vice president whose entire career has been about leveraging the anger in that cohort.

And you can believe it was a good idea or not, but Congress blamed LBOs and decided to target them. And so it’s very likely that policymakers can, by their actions, stimulate bubbles, and when they try to mess with who gets the benefit of the bubble and try to redistribute that benefit, they can kill the bubble.

Do I see anything on the horizon? Yeah, I do. There’s no chance that politicians in this country will allow the complete hollowing out of the income potential of a significant portion — possibly a gross majority — of the population from being able to earn a living, to allow shareholders of AI companies to become oligarchs.

They may allow it for some period of time, depending on who’s in charge, but at some point the country’s gonna rise up and there’s gonna be a politician who says, “Give me your money.” Now, is that something that’s gonna happen in the near term? No, but it always happens. So you can absolutely count on it that at some point they’re gonna take the money from capital and redistribute it.

Justin: I wanna go back to the point you made about the earnings growth revisions and how there was that step-up change — they went from projecting 50 to 60% growth, to maybe double that for some of these semiconductor and other tech-related companies.

So within this bubble regime, is it that the earnings growth estimates are too optimistic, or is it the reaction on the stocks is too optimistic, or maybe some combination of both? Because maybe if the earnings growth comes through, then maybe some of these price moves can be substantiated. I’m kind of asking.

Andy: Right. So, there’s this idea of valuation, which has never been a good metric for understanding markets at all. Like, you know, we’ve seen for a number of years some of the usual suspects talking about — including myself — talking about how given the pricing, 10-year forward returns on assets or equities or whatever are going to be subnormal.

You can’t do much with that. It doesn’t help you to know that, to determine whether you should be buying or selling stocks, or whether you’re in a bubble or not. It comes down to expectations, positioning, and value has become less, to me, relevant. But for instance, semiconductor stocks on a price-to-forward-sales basis are extremely elevated. On a price-to-earnings basis, nope, they’re fine. They’re actually pretty, they look pretty cheap actually. So, you know, you can look at valuations and say — “Does it—” I think it is a helpful and necessary piece of information that contributes to understanding a bubble if valuations are crazy.

But it’s not... You can have a bubble without valuations being crazy. And that’s because valuations are based on expectations, based on earnings expectations. And if you have a step change of 50% earnings growth to 100% earnings growth, it shouldn’t surprise you that stocks rally a lot on that expectation change.

Turns out they rallied more than that — well more than that — which not only tells you that the recent expectation for ‘26 or ‘27 was higher, but all future expectations went up. And that to me is indicative of a market that expects semiconductors to continue to grow well past people’s normal horizons, which to me is a bubble indicator.

When people extrapolate recent historic growth with an order book that — by the way, their order books are nuts. They’re just unbelievably packed with orders. So they’re gonna deliver those earnings. I think they’re gonna deliver their earnings. But at some point things don’t go to the sky.

And so the question is, when people stop extrapolating this sort of earnings growth year over year over year... Well, if it delivers — listen, if we grow semiconductor earnings 100% a year for every year forever, just firstly imagine what happens to something if you double it for 10 years. It goes from one to a thousand. Anything’s possible. But this is where I come back to the point I was making earlier, which is where’s it coming from? For one, where’s the money coming from to buy all this stuff? Like, where’s the money coming from?

If we need to buy a thousand chips 10 years from now and we’re only buying one today, how are we paying for it?

Justin: Yeah.

Andy: What I mean by that is the chip companies are gonna make a lot of money. I get that. But I’m talking about the customers. How are the customers gonna make the money to do that?

Because the customers are getting disrupted. So if you look from a macro standpoint, I think it’s very possible. And, and again, it’s one of those funny things. If semiconductor earnings deliver such that their valuations — such that you can continue to get sizable returns owning semiconductor stocks, a bunch of other companies are gonna be really struggling.

And so the bubble may stick. We may grow into it. I don’t think there’s any chance we will, just to be clear. I’m not timing the market, but semiconductors could double. Got it. They aren’t gonna go up much more than that. And when they stop going — unless everything else gets crushed. And I don’t see how that’s possible. I don’t see how you can have continued semiconductor demand except for one thing: massive leverage. The only way you can get the rest of the economy to get its share of the pie and semiconductors to get their priced-in growth is through massive leverage — borrowed money from the future.

So far, we haven’t seen that. And more importantly, so far we haven’t seen the financing. I think one of the things that we miss when we think about all of this is where is the source of the rally from the semiconductors? The source is that people are buying a lot of chips. Who’s buying a lot of chips? Data centers and hyperscalers who need compute. Why do they need compute? Because frontier models are demanding compute. Now, are the clients of the frontier models demanding compute? Yes, partly. But also the frontier model builders are demanding compute because they wanna create the next thing, which will actually be the thing that changes the world.

LLMs are gonna do what they’re gonna do, but it’s the next thing that’s gonna — they need to continue to advance to get that to be really the productivity miracle. So you’ve got this chain, and people like to call it circular financing. I don’t. It’s pretty straightforward. Compute. It’s all about compute. We need a bunch of money to build compute.

And if you just think of it that simply — where is the source of money? It can come from revenue from those who already have installed compute. It can come from revenue that happens from the frontier models. Or it can come from borrowing and financing.

And so what do we know about that? Well, the first — what do we know? We know the CapEx promises are massive. Trillion-dollar CapEx year over year, which has doubled from the prior year, which doubled from the prior year. So we know there’s a tremendous amount of CapEx. The source of money has mostly been — there’s been a little bit of it coming from the frontier guys, but most of the source of the money has come from the hyperscalers. And they’ve handed that money to Nvidia, who has then handed it to others. But most of the money for the compute build is coming from the hyperscalers who are building the compute, and that compute needs funding.

And so the first source of funding was, okay, we’ve got plenty of free cash flow. Their customers, by giving them profits for their tools, allow them to spend that cash flow — instead of just putting it in the bank and accumulating cash — on CapEx. So that got used. They had a cash hoard already from years and years of being capital-light businesses. They spent that. Now they’re canceling their buybacks because they had been using cash flow to buy their stock back. Meta canceled theirs two quarters ago. Google canceled theirs this quarter. Amazon doesn’t really buy. Microsoft has shrunk its buyback for a while now, and they’ll likely cancel more.

So that takes a bit out of equities, just as a broad thing. And they’re issuing corporate debt — lots and lots of corporate debt. All of those things — that lack of stock buybacks, that additional corporate issuance — all is how this CapEx is going to be funded. So you have to ask yourself, who’s buying the corporate debt? Do they have the ability to buy that corporate debt, to lever up to buy that corporate debt? And what about the pricing? Because eventually all CapEx ends. When the cost of financing is in excess of the return on the CapEx, it usually happens after it’s already crossed over — classic business cycle.

But right now we’re in the phase of everybody needs compute, and everybody needs money to buy it. And so one of the things that’s a potential catalyst for a bubble top is that issuance being a headwind on assets. And so I’m looking with great curiosity about how the three frontier model guys are going to IPO. Just their proceeds alone rivals the proceeds of — even corrected for inflation, et cetera — rivals the proceeds of all 400 of the IPOs that happened during the tech bubble, during the four years of tech, all coming in three tranches this year. Corporate debt exploding higher in terms of issuance. That supply, to me, is the big risk.

If they get it done, if somehow the financial markets lever up and take on that debt, take on those IPOs, take on more risk assets relative to their capital, lever up, then you get a virtuous circle because the money does get spent on compute. And if it works, if it delivers the promise of AI, it will pay off as a positive ROI, which means everybody who invested, who lent money to these companies, will get a nice return.

So there is a very bull case — if without a hiccup, all of these financing, all of this CapEx gets funded by the market and the CapEx delivers ROI without a hiccup — you can have an extremely long run. Not... This, that’s how you grow out of a bubble. Like, bubbles can bubble and then consolidate for years as earnings grow without ever popping, and you can get back to health. That does happen. It’s just very unusual.

Justin: I think that was an excellent overview of looking at today’s environment, specifically taking this semiconductor question, taking my question, and really getting at the first principles under the surface — how you should think about this. Where does the money come from? Are they gonna get the return on the investment that they think they’re gonna get? All this stuff is very first-principles based.

Andy: And the consequences for everybody else—

Justin: Right. Those are important questions. Yeah, absolutely. I mean, I don’t know, and that’s exactly why we wanna have these discussions with you, Andy, about this and a number of different things in the future.

But just to sort of try to sum up — if I could explain it back to you, you know, at a high level — a bubble forms when something new comes to the market or the economy, whether it’s a new technology, a new financing sort of scheme or what have you, and then that kind of puts us in this possible or potential bubble regime.

And once you enter that bubble regime, you have escalations that take place, and they can take different forms, obviously. You have sort of like the FOMO, the neighbors over here getting rich. You have everybody talking about sort of how they’re making money in this bubble, and that sort of puts us in this bubble peaking phase, which is — to your point — hard to time, but there are signals that you can look to that say, “Okay, this looks like we are in sort of maybe the tail end of the bubble.”

Do I have that right? Did I miss anything?

Andy: Absolutely, and I think the key thing there is we’re somewhere along the way in this escalation and FOMO regime. And so the question that maybe we can address at another time is, what do you do about it as an investor?

Like, George Soros used to say — and I’m sure this has been misquoted — that when he sees a bubble, he rushes in to participate. I’m no George Soros, and I think I’m late to this being a bubble, so I’m not rushing in. But as an investor, you have to think about whether you want to... You know, the type of mistakes made during bubbles are the most costly. And so — and that’s the reason why the regime, identifying a regime that’s a bubble is important. Because you have to change your behavior. You have to resist keeping up with the Joneses.

Speaker: In the next episode with you, that’s what we’re gonna discuss. Our plan is to do this show with you monthly, but I think for this discussion and the next one, we wanna get this out sooner. So we’ll record that here in a couple days, and we’ll put these episodes out pretty close to each other.

Justin: This has been a great conversation.

Andy: That was great, guys. Thanks.