When is a Bubble Not a Bubble?

Speaker 1

Hello, everybody. It is Friday, April the 5th, 2024.

Speaker 2

Another week in tech. Hasn't been a good

Speaker 1

Week for my own particular football team. They drew 1-1 at West Ham United in East London. Of course, West Ham, famous for their bubbles. But the bubble wasn't a bubble for Tottenham, at least when they went there this week. And a bubble isn't a bubble for Keith either. Keith Teer, the author of That Was The Week, his newsletter. I don't think he was watching the Spurs game. He's probably watching something else. Uh, but the title of his newsletter this week is when is a bubble, not a bubble. And Keith is joining us, not from East London, but from Palo Alto, just down the peninsula from me, Keith. Um, when is a bubble, not a bubble?

Speaker 2

Um, well, a bubble is not a bubble when you disagree with the people who say it is, let's say, let's put it that way. Um, A bubble is in the eye of the beholder, shall we say. What I talk about in my editorial this week is that bubble has become a politicized word. That is to say, for people who are pretty much skeptics on technology and innovation, whenever money is attracted to something new in Silicon Valley, the word bubble arises pretty much in the first day, definitely by the first week. And it lasts for several years until, um, inevitably there's some correction in the market and then the bubble has burst. And that narrative is basically a false narrative.

Speaker 1

I bet it is Lena Khan behind of the bubble.

Speaker 2

Lina Khan has nothing to do with the bubble. New listeners will not get the joke, but we talk about Lina Khan a lot.

Speaker 1

Keith has a particular love-hate relationship, more hate with Lina Khan than love. So even though you, Keith, you've acknowledged that, I mean, even if you're not a tech skeptic, sometimes the bubble is for real. I mean, there is some general frothiness about the AI economy, isn't there? Would you acknowledge that?

Speaker 2

Well, I think if you put it into normal language and say that large amounts of money are being attracted very early to a promising future technology and that the money being attracted is a lot larger than would be justified by the current value of these things, then yes, that would be true. But that to me is normal and rational behavior. It's people betting on the outcome earlier than the outcome is uh you know for certain uh and and i think in technology that almost always benefits the people making the bets so it so bubble implies they're doing something irrational um i don't think they are i think they're being super rational

Speaker 1

Well, I'm going to cheat a little bit. Usually we end on the X of the week, but let's begin there because your X of the week is an anti-bubble post by Mike Maples, who was at the Y Combinator event this week with Gary Tan and the Y Combinator team. And Maples was really blown away with all the AI there. So maybe talk about that. Were you there, Keith? It looks as if there was a lot of interesting stuff going on. There was this car outside that I actually saw in San Francisco. I don't know who makes this car, but it... It's visually astonishing. When I saw it, I almost crashed my own car. And then, of course, there's the obligatory dog-like robots and lots of sunlit rooms where excited entrepreneurs sat and contributed to the bubble. What happened at the Y Combinator event that so impressed Mike Maple?

Speaker 2

Well, firstly, standard practice is that something like 250 companies are chosen by Y Combinator to go through a training experience where they improve their story and often their business over about a three-month period. And they have a demo day at the end of it. So last week was the demo day. Of the 264, a very large number were AI companies. That is a Tesla car, by the way, just in case you're not sure.

Speaker 1

Is it a Tesla you can buy or is it a prototype?

Speaker 2

No, you can buy it. That's the finished look. It's meant to compete with the Ford F-150.

Speaker 1

Well, anyway, that was there and there were all the robots. I mean, that robot looks as if it could have existed 10 years ago. I don't know what was so revolutionary about that one.

Speaker 2

Probably mechanically it could have, but I bet you the software makes it do things it couldn't have done 10 years ago now. But anyway, the short story is Maple was impressed by the quality of the founders and their focus. And Mike is a principal at Floodgate Ventures. He makes investments, early stage investments.

Speaker 1

Yeah, and he wrote, I could sense greatness in every direction that I look. So Mike Maples certainly doesn't believe in the bubble, although the VCs wouldn't, would they?

Speaker 2

Well, they don't. You see, a bubble is good. That's what people don't understand. A bubble is good. It means that money is being attracted to innovation. Imagine if there was no bubble. it would imply that there's no opportunity.

Speaker 1

Yeah, and that's the analysis of a lot of tech historians. Carlotta Perez, I think, suggests that bubbles are essential in order to create the infrastructure for real innovation. So what happened...

Speaker 2

in ai this week keith in terms of making sense of what one of your your pieces was called by ed zittron was called bubble trouble yeah well he he's focused on a technicality um it's interesting he's the word bubble trouble but i i think he's more his real message is ai may be in trouble and and his argument is that lack of data to train the AI on, meaning that at some point, 100% of all the available data to train on will have been used. And at that point, how will the AI get cleverer?

Speaker 1

Which is a good point. And of course, you didn't cite some of these pieces, but as ongoing issues in the law courts between creatives and these new AI platforms about whether they should or shouldn't have the right to their creativity, to their art, to their music, to their writing. So the content or the stuff that Zitron's so worried about may dry up even before they've come to the end of

Speaker 2

I don't think there's a concept of it drying up. I actually think he's making a mistake. AI doesn't work like that. It doesn't need an endless stream of new data. It needs to have enough data to become clever. And once it becomes clever, it can pass data that it's never seen before. For example, I put code into ChatGPT that it's never seen before, but it's clever enough based on its learning to be able to edit my code and make it better so i i don't think ai requires an unlimited amount of data so you're you're uh uh an unrestrained optimist keith there's nothing about this ai craze that worries you at all i didn't call it a bubble i i would say All science contains the limits of its current knowledge. So, you know, if you look at the history of medical science, of air travel, at any given moment, the science contains within itself the limits of the science.

Speaker 1

That wasn't the question.

Speaker 2

I mean, last week we had a story about... So the answer is, of course, everyone should be worried. But the answer to being worried is more science, not less science.

Speaker 1

What about the, neither you or I as scientists were more interested in the market and startup opportunities. Last week, we had the story of Solomon, Mustafa Solomon and Reid Hoffman startup that essentially they gave away to Microsoft. Clearly they couldn't raise the money. Um, you, you mentioned, I don't know if it's you or perhaps one of the pieces that, um, the only real manifestation of, of, of bubble investment was Amazon's $4 billion investment in, uh, uh, what is it? Anthro, uh, anthropic. So, I mean, in the actual marketplace, is there a bubble? Are people bidding, um, irrationally what one famous economist would call irrational exuberance that I think he coined in the late 90s to define the the dot-com insanity which was a bubble whatever you say well a bubble is only a bubble at the end up until the end what's it before a little bubble no big bubble

Speaker 2

And by the way, it's always rational in my view. There is no such thing as irrational exuberance, except for the people who come too late and overpay just before prices are reset. And if you understand how long cycles work, this is a repetitive... I know, but let's use the late 90s as an example.

Speaker 1

One of your companies experienced the bubble phase, certainly more than one of mine. In the late 90s, there were many companies. The vast majority didn't survive. So is that the case? Let me just... Because this is an important question in terms of this AI economy. Of 100 AI startups, how many of them will, in your view, still be around in five years?

Speaker 2

It'll be shocking if the answer is more than 10%.

Speaker 2

So you're a bubble guy, Keith. No, I think that's entirely rational because the 10% that survive will make a lot more than 100% of the money invested.

Speaker 1

But that wasn't the question I asked. So let's go back to the 90s. If we were talking in 1997, 1998, it may only be one in 100 that would have survived in 2001. Are we not going to have what economists would call the correction in the AI economy?

Speaker 2

Yeah, but you see, the narrative is being framed as a pejorative, as if there's something wrong. The truth is massive amounts of value are being created. It's not destroyed. Value isn't being destroyed. It's being created. And a bubble implies the opposite. So a bubble is an ignorant word spoken by someone that doesn't understand economics.

Speaker 1

I'm not sure that's true. Let's go back to the 90s. There were a lot of Me Too dog food online stores, and none of them survived. Maybe one of them survived. That was a bubble in .com.

Speaker 2

You're kind of making things up, Andrew. There was one of those.

Speaker 1

There's only one everyone can remember. Was it called dogfood.com?

Speaker 2

yeah but there was lots that made it like i use um i i'm blanking on its name now but i get food for my dog from um the the pet place that's uh that was started way back and it's massive and profitable just like uh you know the shoe play the shoe but i don't think i think you'll miss again you're misreading what the bubble people would say that they're not

Speaker 1

claiming any sort of moral... Hold on. They're not claiming any moral observation. They're simply saying that of the 10 companies in... Of the 100 companies in AI today, only 5 or 10 of them are going to survive. Most of them will go away and there'll be a a climax or an anticlimax where the money runs out, and that's always the case, but there's no moral observation here. It's the nature of reality.

Speaker 2

The implication, Andrew, is that things are being overpriced. Now, if you want to really mathematically determine whether things are being overpriced, you have to measure the price from the amount of value that will be produced from the investments. And the amount of value that will be produced from these investments is many, many, many times more than the investments. So even if 90% of these companies fail, it isn't a bubble from the point of view of the people writing the checks. It's an investment. And it's a good investment, probably.

Speaker 1

I knew there was some signal rank ideology here. You're just trying to get people to invest in startups. I think we both know

Speaker 2

the truth about this well any any individual startup Andrew is a risk so if if you were to ask me you know is that investment in that company indicative of a bubble I might say yes but

Speaker 1

good investors don't only invest in one company venture capital right but i think the point of bubbles is that you're right of course about good investors people like mike maple who probably see these the best deals before everyone else but the point about bubbles is it brings everyone in so it's it's not just the good investors it's the bad investors that's the the old story of the the cab drive everybody in the late 90s knew that it was whether you want to call it a bubble or a craze or irrational exuberance, when cab drivers in New York or San Francisco started to talk to you about the value of Yahoo stock. So bubble, whatever you want to call it, exuberance or excitement, it attracts less professional investors. There's more money, there's more froth, and there are more casualties, startup casualties, investor casualties.

Speaker 2

No, if you look at the venture capital section of this week's newsletter, you would conclude that the opposite is happening, that there's a shrinking amount of money going into venture capital. The large numbers in AI are almost entirely driven by Amazon's investment. Yeah. So actually, individual investors are not piling in. And venture capital as a whole has got less money than it's had for many years. All the graphs are down in terms of amount raised and number of rounds. The average per round is flat to up at the early stage, but down at the late stage.

Speaker 1

Right, and this comes in, we've talked about this before, and you had another interesting piece from TechCrunch about big tech companies forming a new consortium to allay fears of AI job takeovers. One of the things I've observed, you haven't pushed back on this one, I think, is that it's much less of a kind of crazy startup environment in the mid-2020s than it was in the mid-1990s.

Speaker 1

Because you can't just start a company with a credit card and a couple of guys. Now you may not need the kind of billions of dollars that an anthropic require, but you've still got to have significant resources. So it lends itself less to... to startups and oddly enough, it's the big tech companies, the Googles, the Microsofts, the Amazons and the Apples of the world that are both big tech companies and also doing the starting up. Is that right?

Speaker 2

It's right, but there's an equally impressive counter trend. The counter trend is that because of AI, it's now possible for individuals to create companies without software or hardware. And one of the memes in the Valley is that in the next decade, we're going to see the first billion-dollar companies that have a single employee, the founder. And actually, Rex Woodbury's piece this week about... Weapons of mass... Production. Production. Kind of relate to that.