There does appear to have been a move recently by Open AI and Anthropic.
[Snorts] To embrace direct customer connectivity. You know, and Anthropic so one business model they could have, which they have had in the past, is we'll sell the token machine to a third party and that third party will turn it into an application. So think, Anthropic powering Cursor that people then used for programming. More recently, Anthropic's launched their own competitor to Cursor. So that this move to say I want to be directly connected to the customer is a sign, I think it's a pretty strong sign, that there was fear that being a model provider solely was as protectable as they wanted it to be.
Because people were switching out the models, sometimes with open source from China and things like that. But if you have the customer lock-in and contract, then you don't have to worry about that as much. So that's one thing we're seeing.
Do you think AI is a bubble?
Well, that's a very interesting question. I've said this before, but there's a wonderful book by Carlota Perez where she says that only real waves cause bubbles. So when the question's asked, there's this notion of, well, if it's a bubble, then AI is not real. And you get into this argument. No, what happens is when there are real waves, people get rich quick. That's already happening here. These huge secondaries at these AI companies. So people are stuffing billions of dollars in their pocket. When people get rich quick, it attracts interlopers and charlatans. And this, by the way, was true in the gold rush. Like this isn't a new concept. If people are getting rich quick, everyone rushes in, and that's what creates bubbles. So, yes, I'm seeing actions and activity that looks speculative, and that probably is over risk-seeking. And when will that correct? I don't know. It's a very big wave. There's real customers buying real products. There's people taking massive bets, the biggest bets ever taken in the venture capital world. And it may be that some form of correction doesn't happen until 5 years from now. But there will be one day because the speculative part is showing up.
And how do you think about investing through something like this for the average retail investor? Like what should they do?
Man, at this point, the odds that you're going to have some kind of insight around some AI company when they're raising money at the highest prices they ever have, I think it'd be foolhardy. And by the way, if you just had the index, you're pretty well exposed. Nvidia got so big as a percentage of the index. Like if you just had an S&P index, you have a pretty strong Nvidia position. Like and you certainly have Microsoft and Google, you're doing well. That back to Burton Malkiel, like that generic advice would have served you fine for the AI wave.
Now, do you think that AI is going to have the kind of job disruption that we're expecting? We're on the back right now of a bunch of layoffs, but you've got Marc Andreessen who's saying, 'No, no, no, this is not because of AI. This is them just using a PR excuse to right-size their organizations.' And other people saying, 'No, no, this is AI.'
Yeah. I think both are true. I mean, I think anyone that does a layoff today is going to write it in the 'Oh, we're just being smart. Like, why wouldn't you? Like, it's just such an easy take.' And I think Square was the one that kind of kicked this off and plenty of people pointed out that on an employee revenue dollar per employee basis, they were like three or four X many of their peers. So, highlighting the fact that they needed to get fit, regardless of whether AI might help or not. And so, I think you're going to see some of both. Yeah, I go back to what I said earlier. Whether you're a company or an individual, if you're worried about AI disruption, become the most AI-enabled version of yourself you can possibly be. If there are 30 paralegals at a law firm, and you're the one that knows most about how to use AI to do what a paralegal does, what are the odds you get fired? I think very low.
They're going to want to keep you around to run the AI tools.
Yeah. I try to get people to see that same thing is yes, it's entirely possible that the jobs market as we understand it now becomes very different in the future, but you want to be the last man standing. And so, if you're the most efficient because you are leveraging AI, you're going to be in a really good spot. Now, I do think there's one question around AI when I talk about what I see in the future, I at least in my own mind have to be very careful to delineate between AI as it stands now, which is a phenomenal tool, but it doesn't have good taste unto itself. It will routinely make things up, and so, you have to know how to catch it when it's getting into a problem. You have to know which answer is better than the other. But then there's where it could possibly, it's not there yet, but it could possibly get to the point where it becomes recursively self-learning and actually become super intelligent. Do you have any mental model for whether this is likely to become more intelligent than humans?
I'm not bought in yet to the notion that that will happen. I'm one of these big believers in that phrase, 'What is strong opinions loosely held?' So my strong opinion right now is that's not going to happen. I know that certain people believe that it will and I know other people that are skeptics on that. Regardless, like you have to know what it's capable of at the edge of your field. And the only way you're going to do that is by playing with it. I've met many mostly academicians for some reason who are skeptics and they got it to fail once and that's the memory they locked in their brain and then they quit testing it. And I said, 'Well, when did you do that?' And I said, 'Have you tried that experiment across four versions of the same AI bot because you can do that now. You can go back and use the older version.' And if it's getting better, you should be aware of that. Like does it still make that mistake or not anymore? And if it's not anymore, then you can't sit around and tell me that it doesn't work because this thing you did. They're definitely getting better. And so like knowing what's possible, certain fields are going to be more deterministic than others. I'm not saying to rely on everything it says a hundred percent of the time, but it can turbocharge what you do. And if you listen to different people explaining, especially small business people, go out in your community and talk to small business people that have figured out how this can help them and listen to the list of things they have it doing for them, it'll blow you away.
That's wild. When you look at the way that we could do AI that would be better for humanity, do you think open source or traditional proprietary companies?
I'm a big believer in open source for almost any technology. I'm probably out on some kind of wild edge of the curve where I actually think the patent system may be a waste of time and just get in the way. And part of it comes from being exposed to open source and how wonderful it is, but there's this great book by Matt Ridley called The Rational Optimist and he says most progress, and I love that word, progress, in the world has come from when ideas have sex. That's the phrase he uses. And the easy example to imagine is you go back to an agrarian society and a farmer walks into market and tells another farmer how he can double his crop output by doing this technique. That transfer cost nothing. That idea went through and then that other farmer goes home and his whole community prosperity improves. So, the teaching each other things that the other person knew is a free gift to increase prosperity. And to me, patents just get in the way of that. And the argument that someone would use against me making that argument is, 'Oh, well, what's the incentive to succeed?' But there is a litany of academicians that have developed new ideas without ever being paid for. Like there's no proof I don't think there's any proof whatsoever that progress would stop if there wasn't an immediate economic lock-in. You know, I think this may relate to regulatory capture of like a 17-year window. And open source is just another incredible data point against that argument in that people contribute tons of time and ideas to these projects with no immediate economic gain. And these technologies become building blocks for other entrepreneurs because they're free. Like it's just so like any entrepreneurial endeavor, if there's an open source alternative, they'll tilt towards it because it just makes their job easier and cheaper.
Mhm. Do you have a sense of why China is going open source? Is it just for the reasons that you said? Or do they see a way to leverage
The previous five-year plan, I think that's the 14th five-year plan, specifically mentioned open source. And I think what happened is for 20 years, they're accused by the West of being IP thieves, like flat out, like just straight up called that. And so, if you're in that place where you're accused of that and you see Linux and you see MySQL, this looks like a pretty good solution to your problem, right? Like the world has said this stuff's free. And so, they got behind these programs and in a big way. On a global scale, they got I mean, and this started 20 years ago. And so, I think it's a pretty good foil to the accusation of being an IP thief if you embrace this idea that IP should be free. And in AI in particular, there's a situation there where there's probably 10 deep-pocketed players who are all competing, and open source has become part of the competitive requirement [Snorts] and that's super powerful because they can all co-learn from one another. It's wildly powerful.
It's very interesting at the aggregate level, but I really just don't understand what is the thing that they use to build a moat to make sure that their company is successful.
So my firm Benchmark has had liquidity events on over 10 open source companies. Like you come up with ways to do it. It's just more competitive. There's a notion in economics called pure competition. You can also go find a Wikipedia page on that. And in pure competition marginal revenue equals marginal cost and that's what drives the most prosperity for everyone in the ecosystem for all the consumers that are out there because you can end up with the lowest price product and that's pure competition. Things like patents and lock-in and regulatory capture, these things prevent us from getting towards pure competition. And so you end up with products that are priced too high. You end up with margins that are ridiculously high. So I think it's bad for consumers to have that stuff. But you can build businesses in pure competition. Like they happen. And there have been tons and tons. DB, huge public company around an open source technology.
