AI Buildout is Boom ‘Not Bubble,’ Says Jared Gross

I learned a new term as this paper came out, which is TLDR, which is too long. Didn't read. Oh, and so, uh, we are now required to provide quick summaries of all of these 55 page papers. So I'll try to do my best today to give you that. Uh, yeah. There is an analogy. You know, look, the, the when we look at past capital investment booms, what you see is two paths. One is high competition where these very capitalized businesses compete directly against one another. You you end up with widespread adoption, a lot of innovation, but a lot of disruption. The other path is closer to monopoly. You think about AT&T, Standard Oil. That path does not appear to be where we're heading. It appears that we are in a competitive environment where once the initial enthusiasm and the flood of capital into the buildout peaks, we will start to see greater levels of disruption, which is exactly what the railroads experience revealed. So the big takeaway for me, Jared, is that you're calling it a boom. Nowhere did I see the word bubble. No, it's not a bubble. I mean, a bubble would imply that there's sort of a fantasy, a level of of expectations. This is a real industry. This is a real economic engine that is going to transform the markets, the economy, society in a lot of ways. Now, that does not mean there are not overvalued companies. That does not mean there won't be firms that go out of business. But I is a massive capital investment boom on a scale and pace that we have never seen before. And, you know, the purpose of writing this paper essentially was to step back and look at it from the perspective of asset allocators who are confronted with this wave of capital that is overwhelming a lot of the traditional asset class choices that they have historically used to diversify their portfolios. How can investors use this I investment opportunity to their advantage? I mean, is diversification even a thought within that world you can diversify. So you can diversify, you know, across sort of themes. So one of the things we talk about in the paper is the sequencing. So we're in the build out phase. The benefits largely go to what is often referred to as the picks and shovels, the suppliers, the chip sellers, the data center builders, the energy providers. And that's very much what we're seeing now. That is largely a public markets phenomenon. Over time, what you will see is more of a, uh, evolution of winners and losers in the provision of AI to end users. That's going to be post IPO model builders and the hyperscalers. And then eventually the technology will disseminate and you're going to see widespread productivity gains globally. I rarely do this, but I have to ask is this going to be available like a McKinsey piece in J.P. Morgan Asset Management? People can go and find a PDF every student out there. I this is your first read for October. I can't say enough about it. Now you got a Lou Ranieri piece halfway through this and the different and tranches. I'm using it in a broad sense, folks, of the securitization and funding of all the debt that's required. You also have the advantage of Bob Michael and his team in fixed income. And J.P. Morgan, don't tell me this is going to be smooth. There's got to be losses taken somewhere, right? There have to be losers. Oh, I think very much so. I think it's almost impossible to imagine AI evolving without a lot of losers. You know, the reality is everyone is competing to win, but they can't all win. You know, right now what you have is this dominance of the suppliers, the picks and shovels. These are all companies that existed before I. They're making a lot of money now selling into AI. If I fades away at some point, they will continue to exist. Most likely. The real challenge for them is to judge how much capital investment they want to put into place. Now, to support the AI boom and what how much is too much longer term among the AI providers? The real question is, can they develop sustainable revenue streams that will support not only the cost of the build out, but the ongoing CapEx that's required to maintain the system? Okay. But I mean, you do this and it's over. It's out 55 pages, literally on a daily basis. The unit and price mix of that future revenue stream is being analyzed, changed, amended literally on a daily basis. Yeah. I mean, where where do you and your particularly your securities in an analysts feel we're going to be 12 months or 60 months from now on the revenue guess that we're all trying to make. Well, I think directionally the revenue stream is growing because I adoptions. Yeah. Okay. The unit price of compute is coming down. And that just capped out as well, because you have very low cost Chinese models that are hanging out in the background, you know, and they're starting to see more widespread adoption. They provide essentially a price ceiling for compute in many respects. Now there's always going to be a use case for the more expensive right, higher end model. So but okay, so do you as a journalist, do you see CDs, credit default swap markets and some of the lesser great credit spread? Widening is healthy and normal in our price discovery of where all this debt's going, given the revenue mystery that we have. Yeah. Well, one of the a lot of the debt is being issued by the hyperscalers. And so what you have to recognize is the initial conditions for the hyperscalers were an incredibly strong revenue stream and balance sheets that were essentially pristine to a degree that we've almost never seen before. So they have a lot of capacity to issue debt without really taking on a tremendous amount of credit risk. And within the investment grade market, there's been some hand-wringing about the rapid growth of IgG hyperscalers debt as a component of the market. But the truth is, you can easily just look at that as a source of diversification. This is a new sector of high quality issuers that is taking its place among other traditional borrowers. And so, you know, when you step back and look at the IG market, we don't see the level of credit spreads as being particularly concerning. Now there are individual names where spreads have widened materially. I think you have to be a little more careful in some specific cases. This is why you engage in active management and do the better research. You got to have the boots on the ground to know what you're looking at.