AI Is Adding to Inflation. What Does That Mean for the Fed?
This week on Fed Watch, ITR Economist and Speaker Lauren Saidel-Baker examines how artificial intelligence is influencing inflation, Federal Reserve policy, and the future of interest rates. The rapid expansion of AI and data centers is driving up electricity, construction, and technology costs. But could AI eventually help lower interest rates by improving productivity across the economy?
Key Episode Takeaways
- 00:00 – AI’s Impact on Inflation and Interest Rates
- 00:24 – Could the Fed Raise Rates 50 Basis Points?
- 02:05 – How Fed Rate Hikes Signal Inflation Policy
- 03:06 – AI and Data Centers Drive Higher Costs
- 05:52 – Could AI Productivity Lower Interest Rates?
- 07:53 – Upcoming PCE Inflation Data and December Fed Meeting
The below transcript is a literal translation of the podcast audio that has been machine generated by Adobe Podcast.
AI is raising inflation, but could it result in lower rates eventually? Hi, I’m Lauren Saidel-Baker, and thank you so much for joining us for this November 20th edition of Fed Watch. We’re going to do a deep dive into the AI impacts of Fed thinking and of interest rates. But before we get there, I want to turn to really the Fed communication.
We’ve been hearing more and more recently about the path of rates going forward. And even though the new Fed chair is showing less willingness to communicate, he’s made it very, very clear that he wants the Fed to be a little bit quieter than it had been under former Chair Powell. We’re getting different types of messaging and different comments from other members of the FOMC. In the news this week, you might have heard Waller was out there talking about the need for potentially even a larger rate increase in December. I want to be clear, our base case is still that that December meeting, much more likely than not, we will see 25-basis points of rate hikes. But until at least today, the odds in the market were showing slightly higher chances of a 50-basis point increase compared to holding rates steady, no change. So, what really would be the mechanism and what’s the motivation for a bigger increase all at once, rather than more of those smaller 25-basis-point-at-a-time increases throughout 2027, which again, is our base case expectation. Again, here it all comes down to communication. We have seen in past economic cycles, and typically downside cycles when the Fed is cutting rates and when you have a little bit more consternation across the economy, we see communication be used very intentionally, almost as a quasi tool of the Federal Reserve. It’s not just that they’re cutting rates, but they’re setting market expectations for more rate cuts to come. That’s meant to shore up confidence during those turbulent times.
So what does that mean on the other side, where we find ourselves today, where rates are actually increasing, and the goal, of course, of those rate increases is to decrease inflation expectations? Well, there is an argument to be made that one more larger rate increase, again a 50-basis point increase, here could give the signal not just that the Fed is raising rates today, but that they’re committed to that path. They really want inflation in their crosshairs and that we will see that further commitment. Again, this is all signaling, which is something that we have seen mixed messages from various committee members on. There are very much different theories and expectations from each of those individuals. So, we’ll be watching this very closely. I think that will be critical as we get into this December meeting and see not just if rates do, in fact, increase, but what the voting structure is like, whether that’s unanimous or if we have some outliers, some dissents in the committee.
But I want to get to our main topic today, AI, data centers. I know this is a story that we have been following very closely here at ITR Economics. I’m sure it is one that you are seeing just headline after headline about. And there are some very real overall economic implications, but especially implications for the Federal Reserve. Chair Warsh was at a summit earlier this week and actually gave some commentary about AI, artificial intelligence, or the other term you might have heard here is superintelligence. Really interrelated topics and just a different way of saying it. So, what are the economic impacts? How does AI and the current data center build out really impact rates? Well, it actually does it in two offsetting ways. In the near term, AI and data centers, this is increasing inflation. So not only do we have those data centers going up and sopping up more electricity, right, that demand pull on electrical prices. We’ve seen for several years now increasing electrical rates, even for the residential rates. But we also have things like the build out itself contributing to higher construction costs. So, there are some very deep pockets that are funding all of these data centers, and they’re willing to take that highest price because they want their data center built today. That means higher wages across the board as they are pulling on this rather limited construction labor availability. That translates through to things like construction materials, those critical components that they need. And we’re even starting to see higher prices in the tech itself. Over the long span of history, as technology gets better, as computers and processors and chips get faster and smaller, they get cheaper. But more recently, all of this build out is just causing so much demand in the sector that we’re actually seeing some types of computer prices increase, again, just purely due to that demand pull. So, there are a lot of estimates about exactly how much this AI and tech space can contribute to inflation. Some of the Fed’s own estimates hover in the 0.1% range of adding to PCE. Again, this is all very theoretical, but that is in the short term. Moving more into the medium term effects, we get this kind of add on effect that some of these data centers, maybe they’re not just funded by profitability and cash, but there could be borrowing to build out these facilities or for some of those add on effects down the line. So additional corporate borrowing, that is going to feed into the supply demand of the corporate bond market.
But longer term, that is the most nebulous. And that is our biggest question. Can AI in fact, raise productivity overall for the economy? Raising that productivity would ultimately reduce costs. It would lead to lower interest rates, all else equal. And here’s the real question for me. We’re seeing some early tentative signs of productivity raise due to AI, but we’re not seeing it clearly across the board, and we’re not seeing it as a steady and sustainable raise in productivity measures. I think a few things are really hampering that productivity gain. First of all is simply adoption. We, if we divide up industries sector-by-sector, we actually do see some still have very low total adoption rates for AI. There are some leaders, clearly financial services, information services, the tech sector, right? They will always be out at the forefront of this adoption. But other more legacy segments, if we get into things like mining, I think the actual lowest bucket was agriculture and forestry, which for a good reason, just doesn’t have too much interest in this space, and at least yet too much reliance. Once these adoption rates come up across the board, that’s going to show better potential for AI to translate to productivity. But the other thing we have to do here is on a company-by-company basis, and even a individual-by-individual basis, we just have to get better at using the tech. We have to get more comfortable with it, knowing where we can save time and be more efficient and where, well, maybe AI just can’t do that task for us quite yet. So, as we get better, as we see businesses stop just throwing money at the problem, buying licenses for all of their employees, but not having a clear path forward of how this technology is actually going to solve our problems, that’s going to be a step that I think differentiates the winners and the losers in this tech scale up.
A quick note for you all here, our Fed Watch viewers, we will be off next week for the Thanksgiving holiday. Wishing you all a very pleasant holiday. We will get some data next week with PCE, so expect us to talk about that the week after, again, I hope if you are waiting for those inflation numbers with bated breath that you will, you will find that mid-week when it comes out. But again, no Fed Watch episode next week. Wishing you all a very safe and happy Thanksgiving. And then we will be back for December and to see what actually happens in this next fed meeting, the much-awaited December Fed meeting. We hope you’ll join us again then right here on Fed Watch.