Sam Lessin on Artificial Intelligence

Essays

From the Podcast

But it's just interesting. The question I think socially is gonna be... Like, this happened a little bit with the internet boom, but at this level, the richest people in America, all of a sudden, most of them, are gonna be people who secure the bag, so to speak, in this cycle. And then if you really think stock numbers goes down, right, a lot, pensioners are the holders on them. The biggest problem is just gonna be people are gonna be really pissed. Like, all of a sudden, the Montana people are gonna be like, "Not only does AI suck, but our economy is dominated by a bunch of people who didn't actually deliver any value." That's very different than, like, the Google and the Meta story, things like that, where you're like, "Okay, people did really well. They were super lucky, but they also did things that had permanent financial value." I think the, the, the thing is gonna be second and third order effects of all this in the coming years, how the society refactors itself and responds to, like, what's clearly, like, a major distortionary event.

— Sam Lessin, More or Less Ep 163

So the question is, there has to be a shock. You need something. You need either a better narrative which sucks the wind out of the old narrative, which can happen, but is super hard when the narrative of AI is everything. Like, it, what's, what's more than everything, right? Like, or you need, like, something to go very badly that then people start having to rationalize, which is unclear how that plays out. So I'm with you. I, I think it's very difficult to defend stock prices that you'll never see financially make sense in numbers, in profit. So you either believe that, like, we're just ha- holding onto the chip 'cause there's a one in three chance it pays out, which is very venture capital, by the way. But you kind of still value the chip 'cause it still might pay out. It's, like, not impossible, right? That it pays out.

— Sam Lessin, More or Less Ep 163

I do think there is something with the devices, though, Brit, to your point, with OpenAI, 'cause again, they're great narrative capitalists, is like, it's a classic buy the narrative, sell the news, right? Which, again, makes these things hard to be public, which is, like, the argument that they will never release a device is because it will only hurt their stock price if they're public. It's hard to imagine a world, like... Which is very, very different than, like, I think Elon pulling off some rocket. Like, he has the downside of rocket blow up.

— Sam Lessin, More or Less Ep 163

Which has been like hilariously down constantly 'cause it can't keep up with all the bot traffic. Here's the funny thing to me about that. It's a great example. Like if you said, "Hey, I'm gonna launch a GitHub competitor," that used to be like a hard thing to do. Now, it's like three people like sitting somewhere in Elon's office or something or some office around. They're just like, "Yeah, like we're just gonna like..." I just think that just changes the world so dramatically when all this stuff, you're just like, "I don't care." There's no software lock-in, right? Like it, it can be distribution. It can be anything about... Like you... But you're like, "Oh, I'm like angry that GitHub is down." It doesn't matter. It's so easy to replace.

— Sam Lessin, More or Less Ep 163

The thing that's wild to me about the moment we're in and like understanding this is I think they also launched xAI recently, like a code repo thing to compete with GitHub.

— Sam Lessin, More or Less Ep 163

I gotta actually start playing with these things again, 'cause I just like have gone down my own like crazy version of this, where I don't use any of the software. I feel like I should... If you guys think it's this good, I'll actually spend time going back to use some of the software.

— Sam Lessin, More or Less Ep 163

Which again, it's completely fine as a seed investment. Like we're used to this as seed investors, that like things have an incredible quarter, but then a shit quarter, and then a great quarter, and then this happens. That is what we're used to. And by the way, if you told me, "Hey, I have a company three years in. They're doing really well. Revenue's still super lumpy, hard to predict," makes all the sense in the world. The only difference here is that these things are at a scale that's never been seen before so quickly, and everyone wants to tell a going public projectable cash flow public market story about them that's really hard to tell, right? Because they're still, for any... in this industry and what's playing out, operating in a lot of ways like the biggest, you know, seed companies in the history of the world.

— Sam Lessin, More or Less Ep 163

Which by the way, again, like there's a bunch of people with like eighty dollar calls on that thing that have just made like insane money. But it is just really interesting this whole vibe, which by the way relates to Kalshi and all this gambling stuff, is like people are so deeply making money in trading stories that are not owning assets. Like it's not about owning Moderna. I don't think you wanna own Moderna, right? But you sure as hell can pay attention to a positive press thing that can swing the stock wildly quickly. That's how you make all the money.

— Sam Lessin, More or Less Ep 163

It's been a great trade for the last few years, and it might still be for a little while longer, and the irony is they've done a really good job of getting people to give them really cheap money and then buying infrastructure, which will have some residual value, right? But, like, I don't understand how things that are these lumpy, where, like, you can't be the winner and compound your winning as OpenAI just demonstrated, is, like, a thing you can own long term.

— Sam Lessin, More or Less Ep 163

If you look longitudinally over time, like the whole... So I actually did this comparison recently, and I'll pull up the data. We can stick it in the notes, the show notes. It's not a great story to have invested in. Now, it's true that you can make money. Like, you're, you know, you get there first pre-regulation, ride the IPO wave, you get out, you've done fine. But if you just held electricity from the beginning to now, it turns out it's like it's a great utility. And by the way, that's a regulated utility with regulatory capture that's hard to transport. AI is globally liquid with no borders and is easy to transport. So you're gonna get the lowest cost provider. There's no money to make.

— Sam Lessin, More or Less Ep 163

Well, and here's the thing. It might slow and then re-accelerate, which is a mess for any investor, right? Like, that's the problem is you can't price this stuff 'cause you don't know. Like, I was saying to someone, I was like, "Look, obviously anyone on this call would be an idiot to not have wanted to seed invest in Anthropic." Doesn't matter the dilution. For the people who did it, they're gonna make a ton of money. I would love to hold Anthropic stock right now. If you said, "Hey, Sam, do you wanna hold Anthropic st- stock, but you can't sell it for five years, what price do you pay?" I have no idea, right? I don't think anyone does 'cause the variance is just so massive on these things.

— Sam Lessin, More or Less Ep 163

You got a barbell. You're like, "Look, if this AI stuff works, we might as well have our hat in the ring here." Like, it doesn't mean they're not. And like, so I think they'll be able to raise capital and get out on a meme, but, like, these are gonna be stocks that, like, wildly hit and then wildly miss their quarters. You know, OpenAI, I think if you unwind the tr- the real story of OpenAI, which I give Sam Altman a lot of credit for, it's the greatest fundraising story in the history of the world, right? And to do that, right, the narrative was AI scary. It wasn't always consumer. It was enterprise, then it was consumer, then it was kind of, like, floating, figuring it out. It was get all the talent and tell the story there'll be one winner in AI, and it will be us 'cause we will compound and be so far ahead we're uncatchable. That's clearly not true. And Anthropic did a different strategy, which was just focus on the part that was working, which is, like, coding and business applications where people, like, there's a lot of spend, da, da, da, da. They're crushing it with that strategy. Will that last forever? Who know? I mean, like, technical people like Codex more now, right? But it doesn't actually matter. It doesn't matter who's in the lead right now. What matters is that no lead is secure, and projections don't mean anything, right? And so I think that's the real story is it's too easy to switch, leads don't matter, it's gonna be impossible to financially price any of this stuff, and I think OpenAI missing massively on quarter and getting so eclipsed is just... It's like if Uber got hit by a Lyft, right? That wasn't supposed to happen. It doesn't tell the story of, like, how growth is supposed to work in these things.

— Sam Lessin, More or Less Ep 163

That's what they've grafted to. They said, "Well, we're an internet company, and internet companies that are working grow like this." But it was-- that was a very, very different age. I mean, companies that could control their growth. Now, look, Facebook had a period when it stalled out and had to be restarted and whatever. Like, it's not perfect. But I think the real issue here is when people are like, "Well, the ARR this year is this, and we're gonna hit this number," it's like no one has any idea, right? And the fact that you can accelerate and then decelerate and then re-accelerate, these are things that are acting like early stage companies still trying to find ignition. But they're just so big and went so quickly that they were having these IPO-style conversations about things that aren't. Now, how does it all play out? I mean, like, look Anything where, like, you can accelerate and decelerate and miss is a huge problem for the whole industry, right? 'Cause it just means that, like, we don't know, right? And that doesn't mean you still don't buy it or, like, you know, you do the Kushner barbell thing, buy some sports teams that you pay att-

— Sam Lessin, More or Less Ep 163

Look, I did CNBC earlier today on this topic, and here's the basic way I would look at it. The real story here is the variance quarter to quarter and year to year on how you project these things is God knows what, right? It's, like, an enormous variance, right? And the problem is that's a problem for everyone. Like, people wanna pattern match Anthropic and OpenAI to the rollout of the internet and Google and Facebook.

— Sam Lessin, More or Less Ep 163

Although here's a funny thing. There's something very funny about, if you wanna talk about relativity. So Dario- In a lot of ways, like, is one of the most important people in AI, right?

— Sam Lessin, More or Less Ep 163

Tweets