Kush Patel: The way I think about it is that the grid has always been designed for large generators and small loads are added over time slowly, right? And now we're adding large loads very quickly that use power 24/7, and that also have pretty rapidly depreciating assets as well.
What I look for is around are, folks making sure that they're incorporating that uncertainty, innovating around tariffs, contracts, call it the human elements of the equation? As we know, things move slowly in the electric sector, especially on the utility side. I think we're seeing some innovation there.
Teri Viswanath: That’s Kush Patel, he’s a senior partner at E3, Energy and Environmental Economics. It’s a North American consultancy that does a lot of public sector work in the electricity markets. What he’s talking about is real-time problem solving for the planning period of the here and now. Within the next five years, power supply is going to remain tight, as we’ll discuss. Taking stock of this moment is what today’s conversation is all about.
Hello, I’m Teri Viswanath, the energy economist at CoBank and your co-host of Power Plays. As always, I’m joined by my colleague and co-host, Tamra Reynolds, a managing director here at CoBank. Hello Tamra.
Tamra Reynolds: Hey Teri. Following up on the conversation we had back in December with S&P Global on rising costs, and then more recently, with the Department of Energy’s Lawrence Berkeley Labs on their expectations of data center growth, we wanted to better understand how other consumers might be impacted by these fast-emerging mega-loads and rising costs. And really, how to prepare for the toughest planning cycle ahead. Here’s our conversation with Kush.
Kush, the reason Teri and I wanted to have a good conversation with you today were really about two important reports that you guys released last year. The most recent one, of course, discusses the customer impact of data center load growth. And the other was a discussion that focused around the long-term viability of building infrastructure for these large-load customers. So, that kind of where we wanted to start the conversation. Let's start off with maybe a discussion around some of the work that you did for the Virginia Joint Legislative Audit and Review Commission around the customer implications of data center load.
Patel: Yes, happy to speak to the report and the analysis. I think that's pretty interesting, just given that Virginia is the world's No. 1 data center market. Our study specifically looked at both Dominion, the investor-owned utility, as well as some of the public utilities there, NOVEC and Mecklenburg. I think one of the interesting takeaways from that analysis is that it was really focused on constrained demand growth. For most of the audience that might be relatively intuitive.
But I think for all the folks out there who are not really paying attention or even think about power markets or electricity, I think that was a pretty powerful concept that was established during the study that there is a limit to the amount you can interconnect per year.
You can see that in the report that I think the maximum that was ever added from a generation perspective in Virginia was 2.2 gigawatts per year. On the higher large-load cases, we're looking at north of three gigawatts per year for the next 15 to 20 years.
That is just a totally different mindset, infrastructure investment that's needed, not just in any single year, but continuously. I think that was really an eye opening for some folks and stakeholders to be able to see that quantification at that level. A really important finding was that data centers to date hadn't really impacted other customer classes, so there hadn't been any cost shifts or subsidization. I think intuitively that makes sense. Folks are designing rates and allocating costs appropriately.
But again, you have to do the analysis and the work to show that math. I think we did that in the study. I think the other important secondary conclusion to that was that, yes, historically, there's been no issues with subsidization. If anything, there has been a benefit for lots of reasons. But then going forward, you're going to have a much different environment where there's no low hanging fruit. You have to build all this new infrastructure. Much of that infrastructure is going to be more expensive than the existing infrastructure that's a little bit older, depreciated, et cetera, lower cost.
And then what do you do in that environment where growth is imposing costs, not just on the folks that are causing that growth, but also for other customers? I think the debate that's happening in Virginia right now is around balancing those two different things of supporting the industry, deciding on what is a reasonable level of growth. Then how do you protect other rate payers or hold them harmless if that's the goal?
Then I think the third piece of it as well was that we really looked into the clean energy policies of the state. Again, no one had put all those three pieces together of what do you need to do to accommodate different levels of data center growth? What does that mean for other customers? Then how do we square that with our clean energy policy? We ran lots of cases that looked at the policy without and with it. Then another interesting finding was that the policy, while very important in the long term, 15, 20 years, in the short term it's not as important just because you're just constrained on what you can build.
Reynolds: From a build-out perspective, how do you support some of these things and how efficient can data centers become when you think about technology changes on their side? In particular, the news about the Chinese AI technology DeepSeek that came out last week, that was really forcing the idea of greater efficiency when you think about the capacity to run.
When you think about that kind of development and when you think about this other report that you guys referenced and put out, load growth is here to stay, but our data centers? Let's talk about the development of those and how you guys are thinking about efficiency disruptions that may happen as we continue to make the data center side more efficient overall and what the long-term trajectory is.
Patel: From our perspective, it was really important just to introduce the concept that there could be very big efficiency gains. Especially if power constraints are real, you can only interconnect so much, that's going to be a powerful incentive for all of these companies to be able to reduce energy usage to be more efficient, outside of just the economics of trying to have a cheaper, better product.
I think you're just understanding that everyone's operating under a pretty high level of uncertainty is pretty important. I think that's okay. You can still make good decisions under uncertainty. The problem is going to be if you're just waiting for more certainty, I think you might be waiting for a long time because it's going to be up and down, more usage, less usage, more efficient, plateau efficiency. I think that's just something that's probably here to stay at least for a while.
I think from our perspective, you just want to make sure you're running enough scenario sensitivities, where you're trying to capture potentially the future states of the world where what happens if every data center is twice as efficient over time? Does that really meaningfully change a forecast and your decisions that you're making in the next three to five, or 10 years, or 20 years, depending on your planning horizons?
Viswanath: So, a lot of uncertainty, but I think the one point that really was brought home, especially in this study in Virginia, is back to your point, 2.2 gigawatts of supply. That's what we can bring on and we're looking at demand that is growing in excess of that three GWs of demand annually for as far as the eye can see. One of the challenges, not having enough power supply or maybe the right quality of power supply. Also just, it's a very different paradigm. When we think about that large buildup that occurred post-war development, large buildup in the 1980s perhaps, it was a very different marketplace.
We have a lot more constraints in terms of what, how, and where we build. So, I'd like to just hear a little bit more. One, is the power supply story, understanding the renewables development we had was largely on the back of being able to develop a transmission to get the supply to where we need to get it. This is a very different marketplace. Talk to us a little bit about the power supply situation and the signposts on how development might unfold ahead.
Patel: I think the era of things being cheap and easy unfortunately seems behind us, The scale that we're talking about, not just in Virginia, but in various markets, really rivals and surpasses, post-World War II development. It's like the BPA or TVA, just giant federal initiatives to bring tens of gigawatts of resources online to supply millions of customers and industry across the U.S. That's the level that we're talking about and more, with load growth percentage points we haven't seen in many decades, but then we're a much bigger system, many more people.
The quantum is more than we've ever done before, actually, even if the growth percentage is similar to what we saw in the '60s and '70s and '80s. Just the sheer number of megawatts that we're talking about have just never been done before. SO, I think that's a challenge. In some ways it's really causing strain across the U.S. with our existing power system, with our existing structures. Markets and regulators were never designed to manage, 2, 3 gigawatts a year of new load and new resources continuously. That's not what things are designed for.
This is something that we said in our white paper that that's a big “traffic jam” and you're going to maybe exacerbate issues that have been present but not really known or really impactful because it hadn't seen this kind of growth before.
It is driving a lot of innovation and reform. Like most things in life, transitions are pretty messy. We're right in the middle of it now. Hopefully then, when we have successive waves of load growth or more things, and we've gone through this, so hopefully the next round or two or three will be a little bit easier. I do think to the question around, do we have the right power supply? Do we have the right things? It's going to be all of the above.
I think Virginia is a bit unique, just given its size and importance in the data center world, but most markets are going to be trending toward more growth over time for various reasons. And then, transmission is always going to be difficult to do, but you're going to do that. Building new renewables onshore will have its own challenges. Offshore has its challenges. Building new gas pipeline capacity and getting new turbines also have challenges. So, I think there's not going to be a simple answer to any of these. It's going to be a mix.
As always, we're a regional electricity system and each region will have its own particular mix of resources, but it's not going to be a silver bullet with any one particular type of technology. I would encourage the listeners and everybody else to just keep an open mind of what is really possible in the next several years, especially. Then that limits your choices to some extent, but also it allows you to think a little bit more creatively as well.
Reynolds: There's some interesting developments that probably are worth discussing. The first one is the possibility that PJM might allow non-intermittent projects to jump the queue. The other is the recent announcement by NextEra on their focus on natural gas fired and nuclear power, as you just mentioned, the importance of that. Help us figure out how this might change the next build cycle in the U.S. and what are people thinking about?
Patel: Yes, great data points, something that we follow quite a bit and very much in line with our Virginia study where you build a lot of renewables, storage, but you also have to build a little bit of gas to make sure you maintain reliability and to be able to meet the energy needs. I think that's very much in line with our analysis that we've done within Virginia and elsewhere.
I think it becomes a really interesting question around how each market wants to be able to manage the affordability, reliability, questions, which is, as always, very important as well as cost. I don't think there's going to be a way where you can just build one single type of resource. I hate to use the term all of the above, but it's pretty clear that is pretty much all of the above, especially when there's constraints on, like I said, nuke deployment, things like that, and also the costs.
NextEra and their investor relations report basically said you have wind and solar and storage being those costs, energy and capacity resource with gas filling in the gap in the next several years. Then post 2030, post 2035, you may start talking about newer technologies like SMRs and other things that are being developed, and lots of technology and resources are being deployed. I personally largely agree with that. I think there's only so much you can do in the next one to two years.
Again, like PJM trying to think about ways where you can fast track some certain projects. There's lots of controversy with that proposal, but ultimately, you're trying to solve your near-term problem. Then there's lots of focus on the post 2035 world, I would say, like what we have, all these new and exciting technologies that can hopefully scale at affordable prices. There's really that messy middle and in the next 10 years where maybe there's not as much focus for obvious reasons because it is very difficult. That I call the donut-hole or the next 10 years where it gets very difficult potentially.
Call it 2025 to 2035, that's where you start talking about the “art of the possible”, like what can we actually deploy. Maybe there's a utility or a set of stakeholders that really want a new gas turbine and they do gas plants, but that's going to take five years to order it from the OEM, get all the technical and other specs, get the permitting for gas pipeline expansion, but then what could be built in the next year or two, that might just be renewables and storage. And that's going to be important for both costs and just to be able to meet your demands.
Maybe for some of the listeners who are maybe smaller utilities, just to be aware of that context that you might be also competing against a NextEra of the world and others who are also going out and placing these orders and doing those things. You might want to think about different technologies or different things that can be done in the next few years.
Viswanath: I guess the coincidence of a lot of these events occurring at the same time, right? The aging of the grid occurring in a big build cycle, a marketplace that's very tight. It is going to bring some developments we haven't had on the drawing board before. I think super helpful context. You did talk about this interesting environment where it was an interesting graphic, as I'm thinking about this, because it's almost like the “field of dreams” for developers
Why are we seeing this energy cycle, the likes of which you probably have not seen since post-war development?
Patel: I think the last 20 years have been, not a aberration, but lots of different confluence of events that kept our load relatively flat on a national level. Ultimately, we've been flat because we had unprecedented efficiency gains from lighting, from new stricter building codes and the appliance efficiency standards, things like that. We saw, unfortunately, a lot of loss of manufacturing across the U.S., call it load destruction or load degrowth that occurred.
Then a lot of our economy shifted to more services that required less energy. All of those things are very big things that happen, and our load stayed flat. It didn't really go down, but it didn't really grow. That's really the only time in our history where that's happened. I think we're in some ways returning to the norm where we saw growth, continuously.
Again, there's multiple sources of that growth. Will AI continue? All the graphs are always up and to the right. I don't trust that as a forecaster because that's not usually how life works in the real world. That being said, there's also other things that are going on, like you mentioned, manufacturing, electric vehicles, even electrification of the oil and gas fields, Permian, places like that, just using electricity more and more as a source of fuel, as a source of energy.
If America continues to be a huge economic engine relative to the peers in the rest of the world, then you'll see a lot more foreign investment. Lots of reasons why it still makes sense for America to continue to grow our electricity sector. So, not to say it can't pause or even go back a little bit, but you're looking at a 15, 20-year cycle here. Yes, I've heard some people call it like a “power super cycle.” I'm not sure if I'm personally ready to call that, but it does seem like we're now shifting. We've been in third or second gear for a while, and we're now in fourth gear, and we'll see if we can get to fifth or sixth. That's the big question in my mind.
Reynolds: The E3 study we just discussed, found that if current trends continue without any constraints on growth, demand in Virginia would more than triple relative to today’s levels. Three-fold growth places significant pressure on system planners’ ability to build sufficient generation transmission, and distribution infrastructure to maintain a reliable system. But the preparation we do today, in terms of innovative planning, will set us up for success for what Kush tells us with be the early innings for the next growth cycle.
Viswanath: Kush and the E3 team are really thought leaders in the industry and I appreciate his insights. I hope all of you will catch up with us next month as we visit with San Isabel’s Ryan Elarton and Tri-States’ Reg Rudolph as they explore innovative power supply solutions…but this time, at the distribution level. These two executives are thoughtful collaborators, which each other and their broader co-op communities, I can’t imagine a better duo to challenge us on thinking out of the box.
Reynolds: I hope you’ll join us then and goodbye for now.
