Liron Shapira
Host of Doom Debates, the #1 show for high-stakes debate about AI extinction risk.
“I believe in the Icarus curve. So we're Icarus right now and we're going up because we're flying closer and closer to the sun. And that's great. Flying closer to the sun is better, right? You get a better view. It's fun. But eventually your wings burn off and you just plunge into hell and it's game over.”
Doom Debates! · AI Research & Frontier Labs · October 2026
Shapira was answering the challenge that people who expect AI catastrophe should be betting against the stock market. His reply is that a doomer expects earnings to keep rising for some years and then go to zero. He says that when he plugged in those numbers he got a valuation close to where stocks trade today.
Transcript
I think that's right. Yes.
I mean, I can't vouch for it, but it sounds like what I've heard.
It honestly does make a lot of sense. People forget. They're like, no, stock is just a vibe. And I'm like, no, it really is. Like, there is a theory, like, look, if you own 100% of the company, you can take control and then route the cash flows to yourself, right? So that's kind of like the nuclear option. So therefore, from there, you can derive that the value of the stock is the value of its future cash flows. So, you know, it's a good theory. It holds up in many cases over the long term. So you take the U.S. stock market right now and you say, what are the future cash flows if you're an AI doomer like me? Well, I think I've mentioned before that I believe in the Icarus curve. So we're Icarus right now and we're going up because we're flying closer and closer to the sun. And that's great. Flying closer to the sun is better, right? You get a better view. It's fun. But eventually your wings burn off and you just plunge into hell and it's game over. Okay. So that's the cash flows, right? Where it gets like higher and higher than the plunges. So I substituted a number. I'm like, yep, the cash flows are going to grow like 30% per year, but then like in seven years, the AI is going to be uncontrollable and the cash flows are going to go to zero, what they call the terminal value. Like the terminal value is just zero. There's no residual value of these corporations. But if you plug in these numbers where it goes up and then down, funny enough, you can derive a price to earnings ratio, like how you're supposed to value the stocks today, which is actually remarkably similar to like the 20X forward-looking price to earnings that the stock market has today. So I tweeted about this a couple of weeks ago. I'm like, here you go, Tyler Cowan. Look, I did my analysis, my AI doomer analysis, and it's yielded a 20 price to earnings ratio. So from there, we can conclude that anybody who's putting money in the stock market at the current valuation of the stocks is consistent with AI doomerism because they think it's going to go up and then crash. So there you go. I've met your challenge.
And did you hear back?
So I didn't hear back on that. But then there's another twist to the story, which is my old friend Max Raden from college tweeted back at me and he said, well, what about interest rate? Have you considered the bond market? And I was like, so even though in equities, I feel like Tyler Cowan doesn't really have a leg to stand on because it's so easy to plug numbers in and be like, yep, today's numbers are consistent with AI Doom. They're also consistent with things going on as usual. So there's no Bayesian evidence there. But then when you look at interest rates, he's kind of got me. I realized like, you know, Cowen might actually have the better of me in this sense because I do actually think that I know something about real productivity, right? I do think that data centers, because of super intelligence, as long as there is a human economy, as long as we're not all dead yet, I do think that data centers are going to drive a lot of real productivity growth. And if that's the case, there's still some other macroeconomic factors that you have to throw into the mix. It's not trivial. But as far as I can tell, I think interest rates are going to go up because I think these data center companies are going to be like, look, give me your capital. Like save your money, stop spending, give us your money now. We're going to invest it. We're going to 10X it in a few years. So don't spend on anything else. And they're going to suck up demand for money, which is going to drive up interest rate. So Cowen got me thinking. I'm like, huh, you know what? I should go make a bet on high interest rates. So, you know, Cowen may actually come out on top in all this.
Huh. Yeah, the bond market is a little mysterious to me. I mean, a lot of factors, inflation weighs into it. Our treasury secretaries attempt to intimidate it, apparently unsuccessfully, intimidate the bond market. A lot going on. I mean, I do think short term, you could have a you could have a crash. Yeah, bubble could pop, but that doesn't mean AI itself is stopping anymore than the dot-com bubble popping in 2000 meant that the World Wide Web was screeching to a halt. But yeah, I don't, I don't profess to know much about interest rates. So I'll let you and you and Tyler place your bets. The one final, I know you got to go to the arrow. I want to say, on the one hand, I feel kind of petty and bad for getting into my personal grievance about the way Scott Alexander's blog treated my book. But on the other hand, it was actually useful full disclosure because it is quite possible that that colored my earlier critique of the way he of his Steve Pinker challenge. I mean, that's the way, that's the way the human mind works. And it's probably easier for me to rev up indignation about you can make
More from Doom Debates!
