The Center Square YT: The AI Boom's $40 Trillion Problem: Productivity Lag

Hey, welcome back to Everyday Economics. This is the show that helps you make sense of the economy going on around you. It's brought to you by the Center Square. I'm Chris Crute, publisher, and my partner is Dr. O. Orfeh Divangi, PhD economist. It's Monday, August 31st, 2026. Today, the biggest bet in this economy and the payoff that we're still struggling to find. We talked a little bit about this in our previous episode. It's difficult not to talk about AI and where we are with AI. Data centers, chips, and power are soaking up capital in an extraordinary way. Much of the money that's funding this is borrowed. Dr. O argues in his column, and certainly in some of the posts he's been leaving on LinkedIn, that productivity grows slow while that spending surge, which leaves a lot less room for error than some of the headlines you might be reading, suggests. Dr. O, good to be with you again, sir. Thanks for having me, or should I say welcome back? Thanks for showing up. I've been here. I've been here. You've been gone. For me, it's more like it. Okay. All right. Hey, man. You and I like to talk about AI, and we've been talking about AI, I think, for a lot longer than some people who try to cover this space. Let's define this thing first. Total factor productivity slowed from roughly 1.5% in 2024 to 0.8% in 2025. Before we go anywhere with that, what is that actually measuring? How much we can do with the same number of workers and machines? Yeah, okay. Make it real for the person who's trying to- If we have new technologies, we are able to go ahead and shift the production possibility frontier. We're not doing more stuff. We're creating more output with fewer inputs. That's what that means. Can we do more stuff with the same number of workers or fewer inputs than we were able to do in the past? That's what we mean by the shift in total factor productivity. Usually, new technologies like computers give us the ability to do more with less. AI could give us the ability to do more with less. Are we seeing that show up in the data yet? Not really. We're spending a lot to build this new infrastructure that's going to allow us to use AI. Despite all of that spending, we really haven't seen, at least at the aggregate level, we really haven't seen the lift in productivity that I think a lot of people are hoping to see. I think that makes sense. In the last episode, you talked about as a big construction project, like a road project or something like that. When it starts, it just doesn't seem to be getting anything accomplished because it's not open. Yeah, it's a bunch of spending. You get a bunch of workers, a bunch of materials trying to build this road, this highway. It's just a bunch of spending, but that spending doesn't translate into anything productive yet. You don't have people driving down the highway to go to work yet. You still have congestion on the other smaller roads. That's where we're at right now in the process. We're just building the highway. We're building the airports so that the business people, like yourself, can get to their work much faster than they were able to in the past. That's the thing, it's eventually. I guess everybody's waiting for the payoff. Let me ask you this because this continues to come up and it shows up in a significant way in the conversations that we have and I think in the financing part of it. Goldman Sachs estimates roughly a third of hyperscaler capital spending will be debt financed. That sits alongside that fast-growing layer of project finance, private credit, and things that are just not on the balance sheet. You wrote that the debt has to be serviced whether or not the bet pays off. Who's actually holding that risk right now? Unfortunately, everyone. We're making these massive bets on AI. If it doesn't pay off, we find ourselves with a whole lot of debt. It's a bunch of companies. It's you who holds their stock in your portfolio. Also, look, at the end of the day, we're betting on AI to lift a real income and help us carry roughly $40 trillion worth of government debt. If the economy doesn't grow and we have so much debt, then we're going to have to face either spending cuts or tax hikes. I'd rather not deal with either one of those bad scenarios and see incomes rise so we can cover the debt. The whole economy is on the hook. It's waiting for this AI miracle to lift real incomes and improve everybody's lives. Until we see the payoff, I'd say we're somewhat committed already. We might as well finish building the highways. You just don't opine. You look at the data, and you deliver the data, and then you try to explain what the data means. I appreciate that, and I know that our audience appreciates that. It's not just your opinion. You're not just thinking about these things as you'd like them to be or guessing. You've argued that stronger productivity is no longer upside for AI. It's the actual requirement for this to make any sense at all, for bringing down inflation without crushing demand, for raising real incomes, for making the debt load logical. If the payoff arrives late, say five or seven years from now, what would be the problem with that? How quickly does AI have to arrive and start to deliver for it to make sense in this economy? Well, look, if real rates stay as high as they are right now, the rest of this economy is barely moving. That's the problem. Yes, that's really the nexus of my question. If this doesn't do what it's supposed to do, it's a big investment on something that's not going to- Real rates will stay elevated and crush this economy. That's the problem. We really need to see the payoff start to show up, and I'd like to see it sooner rather than later. Let me ask you this last question. I think in the last seven, eight months, we've seen AI do things that AI couldn't do before. You had a conversation a year ago at this time about AI and AI capability. First of all, that might be a one-person conversation. The other person you're talking to might not even tinker with AI yet. Well, we're well past that. We're in an early adoption phase. How much faster does it have to move in the next year than it has in the past for it to really start to make a dent on the scorecard? That's a good question. I don't really have the answer to that, and I'm not going to opine like you suggested earlier. I don't know the answer to that. I think ultimately, all of these investments and all of this equipment just scares me a little bit. Here's why. The stuff also depreciates. It has a life cycle. We're going to have to keep more, put more and more and more into the stuff at a time where we're borrowing so much, pushing real rates higher, and the rest of this economy just isn't moving. You look at housing, you look at home bill, all the interest rate-sensitive parts of this economy are really stuck right now. They're really bouncing along the bottom. Really, that's where we are. We're not in an economy that's showing broad base strength and that would allow a Fed chair to talk about potentially raising interest rates. He didn't say that. The markets interpreted that as a potential rate hike, but ultimately, I think this is just not very serious at a time where huge parts of the economy are barely moving forward.