Can the US Grid Keep Up with the Demand for AI?

Steve Odland: Welcome to C-Suite Perspectives, a signature series by The Conference Board. I'm Steve Odland from The Conference Board and the host of this series. And in today's conversation, we're going to talk about an important and fast-changing subject: the issues around electricity demand, power prices, supply, and how it impacts the economy.

Joining me today is Alex Heil, senior economist at The Conference Board's Economy, Strategy & Finance Center. Alex, welcome to the program.

Alex Heil: Hello, Steve. Thanks for having me.

Steve Odland: Alex, you and I have talked over time about all things related to power, the grid, electricity, and so forth. But this is such a fast-moving subject, there's new stuff going on here. But maybe you could start by just giving us a 30,000-foot view of where do we stand in the United States with the level of power capacity versus current and forecasted demand.

Alex Heil: Yeah. Thanks, Steve. Yeah, and we have talked about this issue several times already, but there isn't a day that goes by that another new news article comes out that tries to capture a new angle. So just to set the scene here for our listeners, I think it's important to recognize that over the last couple decades or so, there have been several important inflection points.

So the first one really was in 2007, the iPhone was introduced. People started using devices more and more, and there was real concern at the time that, in the US, electricity demand would just grow off the charts with all these computers, the data centers supporting some of the underlying processing and activity. This would all just completely overwhelm capacity. Interestingly, that's not what happened for the next almost 20 years, until we're now in 2024, 2025.

And the answer in the story there is really one of efficiency. There's efficiencies, not only light bulbs, but also devices have become more efficient. And as a result of that, electricity demand really flatlined for essentially 20 years. Now we're at the point where data centers are being constructed, artificial intelligence is really ramping up. What used to be a couple percentage points of electricity demand from AI data centers might go up to 5, 7, 9, 10 over the next decade or so. So at this point, I think analysts are very concerned that going forward now, this is going to be a real crush when it comes to the US grid's ability to support that increase in demand.

Steve Odland: Yeah, and I think your background is really important and helpful, and of course, you referenced 2007, but there's never any starting point or endpoint to these kinds of things. It started before then with a lot of conservation—turn down your thermostats to a lower temperature.

Alex Heil: Of course, yeah.

Steve Odland: And then we went to compact fluorescent light bulbs, and then it evolved into LED light bulbs. So each step, and then all of the regulations around appliance energy consumption, including water, which is a different issue. But all of these things have contributed to a reduction in demand at the same time the economy is growing and growing into that supply that was freed up. And you've written about this.

It's important for people to understand that because we've gotten complacent about increasing the supply and paying for that increase, essentially, because we've had this dividend of conservation. We're really towards the end of the easy, low-hanging fruit conservation.

Alex Heil: I think it's important, and you make important points there. I think, basically what we've now seen over the last couple decades, we've seen the load overall stay relatively constant. And we started to switch, essentially, and alter the composition that makes up that supply side.

So we had a decline in coal-generated power, increase in natural gas renewables. Solar was nonexistent in 2005, now it makes up a relatively large share of electricity capacity already, with a projection and the expectation that's going to increase. But it's not only that the composition now is going to continue to be changing, but it's also that the top end is going to start to rise, and I think that's very different from what we used to see.

Steve Odland: Yeah. And even though personal computers began to be adopted in the '80s and really went through widespread adoption through the '90s, that power consumption—using the internet and the development of the internet—is now ramped up on steroids. Because the internet is evolving to introduce all of these AI tools, and that consumes exponentially more power. And hence, the data centers that are being built and that power consumption. So all of this, you've described it, I think, as an exponential curve.

Alex Heil: That's very true. And just to give another point of reference, overall electrification and decarbonization in both the building and the transportation sectors, they play a large role here, as well.

So just take electric vehicles. If we converted every vehicle on the road currently to an electric vehicle. So everything that currently has a gas tank turns into one with a plug, essentially. We would increase the total demand on our electricity grid by about 25%. This is meaningful in a world where decarbonization and electrification is increasing.

And this is not just a data center issue, but there are certainly other demands also placed on what is a fraying and, in some parts of the country, deteriorating electricity grid.

Steve Odland: And then you have the move against natural gas use in the home, both in terms of furnace and cooking, because of the emissions in the home, the carbon monoxide and other toxins that are emitted from the burning of that gas. And so there is the electrification of buildings moving away from gas. But all of this, it puts pressure on the grid, but it also makes us risky, as we're so dependent on a sole source of power or of energy use, aren't we?

Alex Heil: I think New York City, just to point out one example, with its Local Law 97, the local ordinance that basically is trying to accomplish the decarbonization and electrification of the building stock in the five boroughs of New York City. In the discussion of that policy, a lot of the issues that came up were what analysts were referring to as the winter peak. So it's not only that there's an electricity peak in the summer, when temperatures peak, but also if everything is electrified in the home, in businesses, and all the buildings, then also the heating is electrified, right through heat pumps and so forth.

And that means on cold days, there also might be a spike in electricity demand. And so there is a risk when there are outages, there's risks that in very cold temperatures, certainly, there are harmful effects for residents and business owners. So I think, yeah, a new type of electricity system and electricity use system brings with itself new operations, a new way of operating that system, as well as new risks. I think that's very true.

Steve Odland: Yeah. At least when you have energy, spread around and diversified into gas, solar, coal, gasoline, natural gas, diesel, all of these various energy forms, if one goes down, you can keep the economy rolling. And we were talking at 30,000 feet here, the entire US economy, but if you literally had the entire economy electrified, wow. Somebody could, a competing power could, power meaning an adversary, geopolitical adversary, could jam the electricity system or do a cyber warfare. I mean, they could take down the whole economy.

Alex Heil: Yeah. I think that, like I said, I think there's certainly concerns like that. I think it also brings up the importance of storage, electricity storage. Some of that is still relatively short duration, so that may get you through a four-hour period, especially in the shoulders around the electricity peak during the day, when you're talking about solar. And so there are different ways one needs to start thinking about providing power to users.

And, similar to investment portfolios, diversification is a good strategy. I think we need to balance diversification with some of the carbon removal and carbon reduction goals, as well as just the operational risk, and making sure that, in the overall scheme of things, we're adopting cost-effective, low-carbon sources of power. And that requires a new way of thinking and a new way of operating and managing the electricity grid.

Steve Odland: Yeah. And not to get off track, but you've been very articulate on many of these adjacent subjects, including the need for nuclear development as a baseload source of energy, but also carbon sequestration in order to meet our overall carbon goals.

Alex Heil: Yeah. Both of those, the data that I usually cite around nuclear is the average nuclear reactor in the US is 40 years old, has maybe now an extended design life of 60. So by the time you get to 2050, the US without additions is out of the nuclear business. So I think that there are all kinds of constraints, and the share of nuclear, percentage of the overall electricity capacity in this country has not moved. We've added a couple plants, but certainly that sector has suffered from cost overruns, cost delays. The most recent plans that were added have not exactly been exemplary case studies in cost containment.

And I think that's something that the nuclear industry is still suffering from. And in a world where we don't have the designs that are essentially uniform across the entire country, but every project is a unique challenge in itself, it's hard to see how the costs are going to come down.

And, yeah, you mentioned the carbon removal, carbon sequestration. That is something that has certainly been developed, but it's still, the scalability is still something that is, there's a big question mark around this. And I think we've captured that in our research, and frankly, we've captured this in prior conversations, right? We've talked about some of the ins and outs of carbon sequestration, and if you really want to do this at the gigaton level of emissions, that is something that is going to require another level of investment and effort to get there.

Steve Odland: And also energy because it's countered. The reason I bring it up is because you've got multiple goals here. And if you want to get to your carbon goals, you actually have to produce even more energy. But speaking of which, you mentioned the 25% increase if we electrified all transportation, but what is your demand forecast over the next decade or so?

Alex Heil: I think the honest answer is, it's really hard to say. It used to be the case that electricity demand in total in the country grew at half a percent per year, something like that. That has ramped up over the last couple years. So when I say ramped up, it's doubled in terms of the growth rate, there's potential that it might grow even more.

I think it's important, though, to keep in mind that as of right now, 2025, the load placed on the system by data centers feeding large language models, AI, and so forth is still relatively small. We're talking 3.5%, something like that. We're not in a case of, let's say Ireland, where nearly a quarter of the electricity load right now it's because of data centers. We're a very different world still, but if the 3.5 or 4%, if the total electricity load that comes from data centers, if that's going to double, if that's going to triple, if we're talking 10%, that is going to mean that the overall demand for electricity is probably going to, anywhere around a percent, 2% per year.

And that's may not sound like much, but that's a huge difference from what we had. And that's going to require a whole other level of planning for capacity, especially when we consider that these increases are not uniformly spread all over the country, right? They appear in regional pockets, where there are data centers in Virginia and other parts of the world as Texas, that's a growing market. There's much more growth contained in some of those markets, and that is going to bring very concentrated effects both on the capacity constraints, as well as prices that consumers and businesses experience.

Steve Odland: We're talking about energy supply, demand, and where are we going in the US? We're going to take a short break and be right back.

Welcome back to C-Suite Perspectives. I'm your host, Steve Odland, from The Conference Board, and I'm joined today by Alex Heil, senior economist at The Conference Board. Alex, before the break, we were talking about your demand forecast, and you were saying 1-2%, and you mentioned that it doesn't sound like a lot, but that's big. It's a lot of energy, and we haven't been adopting a regular practice over the past couple of decades of expanding our production or our distribution because we've had this conservation dividend.

Where do you see conservation going? Is there another step, big step that we could do to save energy in various places.

Alex Heil: I think, if we're labeling this energy in terms of primary energy, right? So everything at the very even front end of gas and so forth that is used, I think there are a lot of opportunities still that come from electrification, when we're talking about overall efficiencies. So moving from furnaces to heat pumps, just to give an example, creates much more efficient energy use conditions. It will mean that we're using more electricity and we, before the break, talked about some of those, the issues that come along with that. From an energy use overall, that means greater efficiency, but it's a different system that we certainly need to manage.

I think another sort of the large-scale opportunities where there is a lot to be gained is on the transmission and distribution side. Our grid is relatively old. Some parts of the country, the infrastructure is decades, near century old, something like that. It's literally a grid that as we have learned, as a result of some of the climate impacts, is failing in parts of the country.

But there are new technologies available that, for instance, make even existing transmission lines more efficient. And what is called usually in that context reconductoring is equipping transmission lines with new materials that are more efficient, that are better capable of transporting electricity over longer distances. And I think there are efficiencies to be gained there. Is that going to be enough? No, probably not to meet the demand over the next 20 years or so, but it's certainly something that is going to take the sting out of certain markets, and I think that is certainly true.

And on the generation side, I think, we have a lot of opportunity to have some of the low-cost renewable power sources, solar in particular, to be added to the grid. Obviously, where it's suitable from a perspective of just the location being friendly for solar and being economically feasible. But I think there are low cost, there are efficiency ways that we can add capacity, as well.

But overall, I think we are still grappling, we as an energy analyst community are still grappling with this notion of what will the future look like once we're talking about electricity demand that is actually going up at a rate three times as quickly as a decade ago. That is still to be determined.

Steve Odland: Now this distribution point, you made many points here and really important. Just for our listeners, you have to dial back to your high school physics classes. The resistance is measured, electrical resistance is measured in ohms. And there is this resistance, or think of it as friction in the system, from the transmission point to the end user point. And part of that is the material that's used to conduct that transmission. Part of it is distance. And so over distance, it's lost.

And so your points, Alex, on using new materials and what you call regeneration, which is like a booster along the line, essentially is, it takes investment, but it will help get more of what is produced to the end user. But I think you've also talked about fewer Very large mega plants and more distributed generation, too, which would help, right?

Alex Heil: Yeah, I think the distribution of electricity generation is absolutely critical. I think it'll make a lot of sense to have more distributed solar, combined with storage, in many parts of the country, but also, there's this whole notion of, let's say, microgrids. So a campus location, a business park, they may manage the electricity demand and operation within that area, within those buildings, within those facilities on an almost enclosed basis that is separate from the grid overall—obviously connected, but there's separate management and operation that takes place.

I think there's real room for some of those innovations in order to make sure that the overall system is not too dependent on a huge few large connections that, if they go down, supply's at risk for hundreds, thousands, or millions of people.

Steve Odland: But even with these technologies, you've talked about the need to invest in generation. There also needs to be investment in the distribution, not only in these techniques that you're talking about, but also capacity, correct?

Alex Heil: Yes. And simply, it's just replacement of existing lines, replacement of infrastructure that is currently there that, of course, you're going to improve and upgrade capacity of those lines. But at the very basic level, you're really replacing one line with another that obviously meets modern engineering standards.

But I think a lot of that will have to happen, especially when one considers that also electricity is flowing between different, in the US, between different states, depending on usage on generation, on overcapacity in certain states and constraints and shortages in others. So the investment in those lines, that is going to be critical. And frankly, from what is the cost overall, if that's an important question to ask, as well. That's going to be one of the main cost drivers.

Steve Odland: Yeah. And then there's the whole hardening of the system so that you have greater uptime and reliability. Now, part of that, if you think about the California fires, part of the problem with the fires was created by outdated generation and the interference with dry brush and trees and so forth. So there's a need to harden the system.

Hurricanes also, or large storms, tornadoes, whether they take out these systems, and so we need to move them underground. We need to use these new technologies. But it's essentially replacing what has gone in over the past 150 years, which is not easy, number one. Number two, there are land-use issues. You can't just bury these things everywhere. Three, it's trillions of dollars, and it takes time. So all of this is necessary over the period of time at the same time that we're trying to expand the system to meet increased demand.

So the combined needs of generation and distribution and upgrading, from an investment standpoint, Alex, is just enormous. And that's at a time when the states and federal governments are running deficits. So it's a huge challenge.

Alex Heil: Right, that's very true. If I just pick up on one point that you made, so it is certainly true that these, let's say transmission distribution lines, they require a lot of investment. So do generating capacity, new generating sources as a result of this uptick in demand, in part from data centers and in part from electrification.

One big question in all of this is these incremental costs, how should they be allocated to users? So there's, based on the economics, you can consider that the person who has caused the increase in demand should be responsible for a large share of those costs. But this is easier said than done in many circumstances. But you really have to grapple with this question of how some of these tremendous increases in cost that, not in all cases benefit all users of the system, how should they be assigned? In terms of the connection to a data center, the nuclear power plant that is now being run for longer and requires more investment because otherwise there would be a shortage of electricity in a certain region. Who is going to pay for this?

Because what we've also seen is that electricity rates, they have been increasing fairly dramatically, in some parts of the country more dramatically than others. And businesses and households are feeling the pinch. So I think going forward, just from a proper economic cost/causation perspective, that will be a whole other issue to wrestle with.

Steve Odland: And these rate increases have been substantial. I think we've seen what, 30 to 50% rate increases in California alone due to some of these issues. Now, normally, as you invest in supply in excess of demand, pricing comes down. But because that investment in supply is so expensive. It needs to be spread over a larger user base, that isn't necessarily the case that the prices would come down. So it feels like, almost regardless of supply and demand balances, prices have to go up.

Alex Heil: I think initially, maybe a year and a half ago, when this debate around data centers and the demand on electricity capacity first really came up in the news, a lot of the utility folks were talking about this from exactly the perspective that you just mentioned, which is we have more demand, and so we need to invest in more capacity, but that can be spread over the large number of users. So the marginal cost might actually be beneficial for everybody. It's not going to drive up rates too much.

That discussion has morphed a little bit because what data now more recently discovered seem to suggest is that some of the wholesale costs, especially around data centers, have gone up orders of magnitude faster than what we're seeing for the average prices in the electricity grids overall. And so there clearly are these cost pressures that are emerging.

It's too early to say that this is going to be the long-term trend, because there's some overinvestment, there's a lot of activity right now. We'll have to see how this all looks once the dust settles. But I think that these cost pressures are real, and they will be felt by businesses and households that, in effect, had very little to do with some of the investments in the way that they are made in the first place.

Steve Odland: Yeah, and we've been talking largely from a consumer standpoint, where pricing is basically given to them without a lot of competition or any choices. Businesses are a little different. How should businesses be planning for this? What should they be doing?

Alex Heil: I think businesses, there are several strategies, there's a whole host of strategies, it's a complicated field. I think managing their own load profiles to avoid peaks is helpful when negotiating rates with utilities. I think smoothing out, and that's as a result of potentially utilizing storage, that's as a result of moving operations and shifting peaks that way, that can be very helpful. I think there's room, and we mentioned this earlier, for some of the distributed generation resources. So low-cost solar and storage for large businesses where it makes sense can be a good strategy to become at least partially independent and resilient. And I think that's a reasonable strategy, as well.

But maybe above all, you really have to be aware of what's going on in your region, in your grid region. It's where the grid is managed in the United States, in these different grid regions. And then more locally, I think important considerations there are really, what is the trend going to be, and what does that mean? Even if I'm just a relatively midsize business or larger business, and I have in the past have just been buying these electricity rates, these electricity products that were offered from the utility. But now there might be the time when active management anticipation and planning is really replacing the "lean back and wait" approach.

Steve Odland: Any final thoughts on this whole area of electricity and power generation?

Alex Heil: I think it's a fascinating area to study. Things have changed a lot over the last 24 months. I think we are living now in a world where gas has taken on importance because of the large share that it contributes in the power sector, in particular. but there are constraints and shortages that have now materialized. A) it takes a while to add gas plants, generating plants to the grid. The turbines, they are all in back order. It takes a while to get those. That's measured in years, not in months, in most cases. So that's been identified in the past as one of the main fuels in this electricity story.

But I do think that with the falling cost of some of the renewables, solar in particular, and storage—the cost of storage has come down tremendously. There's a real role that that particular resource can play to provide electricity to companies and consumers in a whole wide range of sectors in the market. We, as of the end of this year, we're marking the day that we will have been paid, our solar system that we have on our residential home will have paid for itself in just under six years, which is actually faster than was anticipated. And that's in northern New Jersey, which arguably is not the most solar-friendly part of the country.

But I think there are opportunities there. It really just takes a careful consideration of cost, of efficiencies, of capacity and how it can be utilized. It's a very nuanced story.

Steve Odland: Yeah. And the length of home ownership is close to 12 years. With the reduction is cost, the payout is higher, and hopefully this will make it more financially doable for US households.

Dr. Alex Heil, thanks for being with us today.

Alex Heil: Thanks so much for having me, Steve.

Steve Odland: And thanks to all of you for listening in to C-Suite Perspectives. I'm Steve Odland, and this series has been brought to you by The Conference Board.