Can I step on the scene? I'm ready to attack. Andre King, yeah, he's the one to bash, asking questions, no filter, just keeps on grinning. He's the man who's got the whole crowd spinning. Money, fame, success, it's all in his sight, till deep expose you with all his mind, no sugarcoating. He'll cut... Andre King, he's the one. Yeah, he's an expert.
Hello everybody, we're in Austin today, one of the boom towns in America. Although it's hard to tell whether we're living in America in a boom or bust. There's AI, of course, the markets are doing very well, and yet in other ways there seems to be some stagnation in the American economy. My guest today, Byrne Hobart, has been doing a lot of thinking about booms and busts and bubbles. He's the author, co-author of a new book, "Boom, Bubbles, and the End of Stagnation." He is a newsletter writer, an investor, and a man who spent a lot of time thinking about whether or not we live in a boom or a bust. So, Byrne, tell me: boom or bust?
So right now it certainly feels like a boom. I was in the Bay Area last week and people seemed happy. The company catered food was delicious, and the headquarters buildings were new and beautiful. And as you mentioned, stocks are up, funding is abundant — especially, well, funding is abundant if there's an AI story involved. It's a little bit trickier otherwise, but yeah, it feels very, very boom-like right now.
And yet you know that it's more complicated than that, don't you? That's what your book's all about.
Yeah, so in the book, I think one way to think about it is to think about different ways that an observer might react to the claim that we're in a bubble, or there was a bubble in whatever the thing was. And I think one thing you can do, if you are a very strong believer in the efficient market hypothesis, is you could say there's no such thing as a bubble, that we're always using our best efforts to estimate the value of different investments, and that if Pets.com or Webvan or a house in far exurban Las Vegas in 2006 has some price, that price is our best guess as to what it's worth, and that those prices fluctuate. And that doesn't seem to be entirely true; it doesn't quite comport with reality.
And then I think another, maybe more conventional view of bubbles is that they are this dark aberration where sometimes people just get a little ahead of themselves and stop thinking about fundamentals and get unmoored from reality and start making investments and making decisions that really are more driven by euphoria and irrationality than anything else. And then there have been some efforts to rehabilitate bubbles, where people point out that yes, we do invest a lot in some things, whether it's railways in 1830s and 1840s Britain or it's fiber optics in late '90s US, but we end up getting something useful out of them. That the later stage of the Industrial Revolution would have been much more challenging to execute if the UK and the US did not have large internal railroad networks to make it easy to transport goods around. That something like YouTube or Netflix would have been much harder to build had we not overbuilt in the past. And there is some truth to that.
But the counterargument there is: if these things were worth building, we could have built them. We could have built them when it made sense rather than before it made sense. We probably would have saved some time, we probably would have saved some money doing that, and wouldn't have overbuilt. And our view in the book is that there is something special about bubbles, and that what they do is they are a coordination mechanism that tells people that now is the only time in history when certain things will get built. It's not just that we build it sooner than expected, but that it only gets built in the first place because of this collective euphoria.
So AI is a very fun example of this because you have a lot of different inputs into AI. You need people who are researching and thinking about the algorithms — that's not especially capital-intensive. You know, having someone alone in a room with a whiteboard and a computer, fairly cheap. But then you actually need the data to train them, you need the hardware that they will run on, and you need to actually do these fairly expensive training runs. And here's where the coordination comes in. So when Nvidia is trying to think about how much should they invest in the next cutting-edge GPU design, one of the questions they ask is: what will OpenAI and Anthropic and Gemini and xAI and all these companies — what will they want, what will they need? And then when those companies are deciding what they're going to do next, how powerful their next model is, they're asking: well, what is Nvidia going to do? How many chips will be available? How powerful will these chips be? How power-efficient will these chips be? Etc.
And so both sides have essentially locked into an expectation that AI is improving very rapidly, and therefore it makes sense to invest a lot in it. And had they not mutually convinced themselves of this, and had they not convinced investors that there would be returns at some point, we would just see much, much slower progress, and would potentially not get to the point where we have models that are powerful enough that they give us some indication that AI models can get much more powerful than that.
Byrne, I'm sure you're familiar with the work of Carlota Perez. Yes. And please correct me if I'm wrong. My understanding of what she argues — she's an influential thinker and writer — is that you always need essentially waste. Booms or bubbles are essential because out of bubbles you get foundational growth. Is that a theme that you pick up in your book?
Yes, absolutely. And I think it is somewhat challenging to get just truly comfortable with the idea of waste, of the idea that not only are we putting more money and more time into these sectors, but we're also expecting that a lot of that money will be wasted, a lot of that time will be wasted. A lot of people spent a lot of their careers on things like pricing subprime mortgage-backed securities, or selling those mortgages, or building houses that people didn't really need to live in, or didn't need to live in for a very long time. So there is a lot of waste in these bubbles, but there is also the creation of new technologies that just wouldn't have existed either at all, or would have taken much, much longer.
And in the book we do have case studies that talk about private-sector bubbles — that's a lot of the later ones — but a lot of the earlier case studies are things like the Manhattan Project and the Apollo program. So these are government-funded, but they have a lot of the same characteristics, where a lot of capital and a lot of talent gets dedicated to some new idea, some untested idea, and it really only makes sense if people believe that one, it's possible, and two, it's going to happen right now. So if you look at the Manhattan Project, for example, you really — if you want to build a nuclear bomb and it's the early 1940s, you need to do two things: one is you need to design a bomb that actually works, rather than just notice that it's hypothetically, theoretically possible, and two, you need enough fissile material to actually build one.
And we didn't know going into it which material would necessarily work, and we didn't know which process for extracting the right materials would work. And so what we did, which was a very bold, kind of crazy move, was just launch a bunch of — we did multiple designs in parallel and built multiple enormous factories that were using different techniques to separate U-235 from U-238 to get the right kind of plutonium, etc. And some of these did turn out to be wasteful in the sense that the process didn't really work. Enough of them worked that there was enough material for one prototype and then two working bombs. So it did end up working, but there was just some waste. We had to know that there was going to be waste prospectively because we were sort of overinvesting.
But the alternative to that is you get a really good design for the bomb whose fissile material you don't get, and you also get a lot of fissile material and you don't have a design that works for it. So the only way to be very confident that you'll actually get the outcome that you want on the timeline that you care about is to do a bunch of things in parallel and expect some of them to be wasteful. And had we done it in sequence, had we said, well, we'll try different — you know, we'll try gaseous diffusion and then we'll try something else and then we'll try something else, and then once we have enough material then we'll start designing a bomb — well, suddenly your timeline starts stretching out so far that the urgency to actually build this doesn't exist.
And so I've imagined this hypothetical of congressional hearings in the late 1940s about the Manhattan boondoggle, where we spent just staggering sums of money on a project that didn't actually yield any improvement in the war effort, and that money could have gone to conventional things. I think that's a real hypothetical possibility in a world without bubble dynamics, and that's just not the world we live in. We live in a world where we can coordinate, we can agree that it is acceptable to waste, because waste is a sign of a sense of urgency, and it's a sign of just sufficiently abundant capital, sufficient commitment to the end goal that we have high confidence we can get it done.
But on the other hand, of course, the Manhattan Project and the development of a nuclear bomb was essential. There was no choice involved, like with AI or the internet or electric vehicles. There was the fear that the Germans were also building a bomb, so there was no rational analysis of is this a good investment or not. It was essential, wasn't it?
Well, there was debate over whether or not the Germans were actually developing the bomb. And there was an incident in which the US intelligence services actually recruited a baseball player who also happened to be a very nerdy baseball player and just sort of one of those generalist jacks-of-all-trades. They sent him to Switzerland to listen to a lecture by Heisenberg and try to infer from the contents of that lecture whether or not Germany was actively working towards building the bomb. And the idea was, based on the outcome of that lecture, he would or would not be assassinated by this American agent in a neutral country. And it turns out that the lecture was a little bit more boring, a little bit more bland, and did not give any indication whatsoever.
Yes, who was the baseball player?
I believe his name was Moe Berg.
Yeah, Moe Berg. So how do publicly financed booms and privately financed booms differ? We can think perhaps of the Apollo project versus what Elon Musk is doing now at SpaceX. Are they comparable?
