AI Boom: Bubble Signs, But Opportunities Remain

I always ask myself, this is the question I always ask myself on individual stocks: am I more afraid to own this or to not own it? And it's a good test for yourself because it's very easy. I think we have all the characteristics of a bubble building in AI. Yeah, maybe why don't you explain what your framework for bubble real quick.

Well, a bubble, I mean, when I say a bubble, I mean, bubble is just a very shorthand term to say something's happening. And so what do we mean, shorthand term for what's happening? There's a massive amount of capital being allocated to an idea, right? And that idea is not crazy. I mean, railroads were not crazy. Housing was not crazy. Building the TMT bubble was not crazy. 

Building AI is not crazy, right? It makes a lot of sense. It's going to drive society forward. So we're seeing that allocation of capital. That's not enough to sell stocks. That's actually a reason to buy stocks, right? In the sense when people are underestimating the scale of the opportunity. The second thing you see in a bubble is you see capital markets frenzies. Yes. And we're beginning to see that, right? We're beginning to see. 

I saw a big deal, big private equity deal this morning for a data center. You're beginning to see. It feels like almost self-dealing. Like it's hard to even keep track of everyone. I mean, the self-dealing controversy around who's funding who. I mean, I there is an element of this where in order to reach the level of AI that you need, you need physical spending.

 And so the fact that Jensen at Nvidia or AMD or anybody is is not pushing for more capacity. So the thing about AI, you got to remember is the more so AI is a guessing machine and the more, and it's a guessing machine that we then train, right? 

We do reinforcement learning to train, and it's a guessing machine though, and the guessing of the machine gets better and better the more power and the more GPUs it has. The better at guessing it is, the better we can train it, the more useful it is.

So there's an element here where we need to have this physical spending so that we create the economic surplus that AI creates. So AI is, today AI is creating a pretty significant economic surplus.

So what I mean by that is we're getting more value from AI than we're paying for it. Much more in consumer and in enterprise. I think much more in the way I'm using AI 3 or 4 hours a day in my work and it's been tremendous. Like what T Rowe is doing with AI, I mean, we fundamentally believe will give us a competitive advantage against our, I love it, and we can see it, right?

We can see the competitive advantage, so there's a surplus today, and I think...the whole ecosystem from Sam Altman to Jensen, they want to push the surplus, and so they want to push this compute power that makes the guessing machine better, which leads us further.

So it's a little bit of that that's happening. And as long as you're seeing the improvements you're seeing, I think that's actually a net positive. So you see allocation of capital, you see capital markets frenzies, that's starting, and then you see oversupply. Which we're not there. We're not there. I mean, we're just simply not to oversupply. We are in undersupply.

And I think when we look at the volume growth in the last three months, right? And we are private investors as well in a lot of the companies, so we see a lot, we see a lot. The volume growth we're seeing is so tremendous, right, that we, and then it follows on in the last few weeks, I mean, last week, you know, Microsoft saying, you know, we're running out of capacity, right?

So we're seeing, and a lot of these builds that are being announced today are builds that bring capacity in the second-half of 2026 or 2027, because we have to actually build things, right, and lay copper cables, right? And so these things are, so when that... And oversupply.

So when you get huge allocation of capital, capital markets frenzy, and oversupply, run for the hills. A bubble is bursting. And so we are very, very focused on understanding that.

That's not today yet, right? So today, when I get back to what I said earlier, are you more afraid to participate in this, especially if you think you have an edge, which I think our platform, especially some of our mid-cap investors that are looking at ideas all over the world. I think we have some edge.