In this episode, Harvard Law's Eliza Martin and Ari Peskoe join me to unpack how data centers' skyrocketing electricity demand could leave ordinary customers subsidizing Big Tech's power bills. Most chilling is the potential alliance between utilities and tech giants that threatens to derail much-needed utility reforms while entrenching fossil-fueled infrastructure.
(PDF transcript)
(Active transcript)
Text transcript:
David Roberts
Alright, hello everyone. This is Volts for March 12, 2025, "Who is paying for all that data center power?" I'm your host, David Roberts. As everyone in energy world knows by now, data centers are coming and they're going to increase demand for electricity, by an amount that no one really knows but everyone seems to agree will be quite large. If the more enthusiastic forecasts bear out, data centers could go from consuming roughly 1% of US electricity to something more like 12% by 2030, quintupling the pace of demand growth. There are reasons for skepticism regarding those upper-end forecasts, but even the more modest ones are daunting.
In many regions and for many utilities, data centers not only represent the vast bulk of electricity demand growth in coming years; they could quickly become the vast bulk of electricity demand, period.
How is that going to affect the relationship between hyperscalers and utilities, and between utilities and their other customers? Who is going to pay for all the new electricity infrastructure required to connect and run all those data centers?
These questions are mostly being decided behind the scenes in obscure utility rate cases, but they deserve to be brought to light. To that end, I just read a great new paper called "Extracting Profits from the Public: How Utility Ratepayers Are Paying for Big Tech's Power."
I'm lucky enough to have the authors, Eliza Martin and Ari Peskoe, here with me to talk through it. Eliza is a legal fellow in the Environmental and Energy Law Program at Harvard Law School, and Ari — a repeat Volts guest! — runs that school's Electricity Law Initiative.
We're going to get into some nerdy stuff about ratemaking, but also some bigger and more fundamental questions about the future of electricity regulation in an age of AI.
With no further ado, Eliza Martin and Ari Peskoe, welcome to Volts. Thank you both so much for coming.
Eliza Martin
Thanks for having us.
Ari Peskoe
Thanks so much for having us.
David Roberts
This paper is just — I am the target audience for this paper. This paper is squarely targeted at my pleasure zones, but it's quite technical, so I just like just a little bit of background. I think the reason we're all talking about this is that the electricity system is changing very quickly in some very fundamental ways. The main thing is these new loads are coming online. And this is not like the past 20 years of loads coming online. It's not like a new house coming online. It's not even like a new business, even a new factory coming on.
These are enormous loads. Like when you have a half gigawatt, a gigawatt chunk of new load, that's just like a new thing for utilities, and we're having not just one of them, but a whole series of them. So basically, this rush of data centers is crashing into our conventional way of running utilities and regulating utilities and setting rates in such a way as to make a mess. And the danger in that mess is that ordinary ratepayers are going to end up stuck with the bill for a lot of the new expenses required by these big data centers.
And I think we can all agree that having ordinary ratepayers subsidize mega-jillion-dollar companies is not a good outcome. So that's why we're all worried about this. That's why we're talking about this. But I think to understand why and how that might happen requires a little bit of delving into utility business. So I think the place to start is: listeners may or may not know, you know, I think listeners, they know they pay their electricity utility a certain rate, a certain amount per unit of energy. What they may not know is that for any utility, there are multiple classes of ratepayers that pay different rates.
And the principle that is supposed to guide the setting of rates for different classes of customers is called "cost causation." So maybe, Eliza, we can just start there. Like, what is the principle of cost causation? What are utilities trying to do here?
Eliza Martin
Sure. So, cost causation is really just sort of the guiding principle in rate making, that consumer prices should align with the cost that the utility company incurs to provide service to that group of people. So, it's more expensive for utilities to provide service to certain groups of people. You know, ratepayers who are like, you know, normal homeowners, they're spread across a longer distance, so maybe there's more distribution infrastructure. So, the costs are just a little bit different for different groups of people. But, it's a very imperfect system.
Ari Peskoe
There's no objective method for allocating these costs generally. You know, set aside the data center issue. If a utility wants to build a new power plant because demand is growing, how are you going to split that cost among, let's say, residential, commercial, and industrial ratepayers? One way you might think about doing it is just, "Well, we'll just do it in proportion to how much energy each group is consuming." And that might seem like a fair way to do it, but you might also say, "Well, actually, what's really driving the need for this new power plant is that growth in peak demand," right?
"O n that hot summer day, that's really why we need this new power plant. And so, let's actually split these costs based on how much each of those ratepayer classes contributes to that peak." And there's no real objective way to say which way is right and which way is wrong. And so, you just have all these groups coming in, arguing in their own self-interest for why their way is sort of the objectively right way to do it. Even though it's just a sort of subjective determination.
David Roberts
Yeah, and just to underline the point you're making, it's not that there is some objective way that they're all lying about. There is no fact of the matter here. There is no right answer about cost. These are inherently subjective decisions.
Ari Peskoe
It's everyone coming in, arguing for their own self-interest.
David Roberts
Which is like, again, you can sort of make sense of in your head when you're talking about normal customers. Right. But again, if you're a multi-billion-dollar company with a gigawatt of new load coming, it is clearly going to have resources in that fight that other people in that fight don't have. Like, they're clearly going to be the 800-pound gorilla in that fight. So, the way utilities do this is they add up all the revenue they think they're going to need in their next rate period and then they get that approved by the public utility commission, and then they divide that up.
Who's going to pay what in that amount? These different ratepayer classes. So obviously, you have these different ratepayer classes sort of jockeying against one another, trying to lower their rates and increase the other rates for the other guy. And there's a lot of fuzziness, as you say, involved in that. So maybe just talk a little bit, Eliza, about what is the source of that fuzziness. I mean, there's not a, like a clear, you know, some answer you can say, "This class of customers costs the utility X amount." There's lots of fuzziness. So maybe just talk a little bit about what are some of the sources of uncertainty and fuzziness in that process.
Eliza Martin
So, I think that there is this initial fuzziness, as you mentioned, sort of about what the revenue requirement is for the utility company. This process is really utility-driven. So, the utility proposes how much money it needs to recover first. So that in and of itself is a contested issue. And then you have the actual, as you mentioned, the sort of discussion amongst the different groups of people about how much money they should be paying for their costs. So, industrial consumers have, you know, they'll get attorneys and everyone will comb through the utility filings and try to argue that their cost should be lower than, say, the normal consumer's cost.
But it's not an exact science. It's a really messy process that's based on competing approaches to cost causation. And it's a battle sort of in the filings about who should be paying what.
David Roberts
And as you say, PUCs, public utility commissions, who are hearing these cases, they might have their own preferences, too.
Eliza Martin
Public utility commissions are inherently a political body. So in some states, the regulators are appointed by the governor, and in other states, they're elected officials. So there's political pressure, whether that's from a governor or whether that's from, you know, the people that elect these regulators to favor their interests. So, you know, it's not a scientific process, for better or for worse.
David Roberts
Yes. Also, the one other thing you mentioned in the paper, the reason there's some fuzziness in this process is that, and here I have in my notes, it says "utilities lie." Which is just to say, when a bunch of people are arguing about how much they're costing the utility, that's pretty technical. There's lots of numbers involved, there's lots of information involved, and the PUCs are sort of reliant on utilities to tell them that information. And as you document in the paper several times, utilities have been caught basically just lying about the cost of various things. So that fuzzes up the process, too.
So then, alongside this process where you have ratepayer classes competing over who's going to get the lowest rates, you have these special contracts which are like — it's funny, I'm constantly coming across new stuff in the electricity world where I'm just like, "Can't believe it exists." I felt that way the first time I found out about utility holding companies. Like, we legally separate generators and transmission companies, and then we just allow one company to own them both. How is that separation?! Anyway, so, special contracts. So, Eliza, what is a special contract? If they already have this process, what is a special contract on top of this process?
Eliza Martin
That's an excellent question. Special contracts are really just statutorily created mechanisms that allow a utility company to request a deviation for a customer from the standard rate. So, oftentimes, they'll have to be approved by a public utility commission, although Mississippi recently changed that law for data center customers. So, it just allows a utility to offer terms of service to an individual consumer that otherwise aren't available for anyone else.
David Roberts
Again, that just seems crazy on its face that individual ratepayers, individual customers, are negotiating their own bespoke individual rates with utilities. So, when you say political pressure, why would there be political pressure to do these things? Like, what's the background there?
Eliza Martin
Oftentimes, it's that regulators don't want to be seen as the veto point for an economic development opportunity. So, if you know, a governor's publicized that there's a big data center that's coming to locate in this area of their state, then regulators don't want to be the reason why that economic opportunity or that data center decides to not locate there. So, there may be pressure to approve a special contract.
David Roberts
Yeah, lots of politicians want these things to come, and then, I guess, are probably pressuring their PUCs to make things easier on them to come. So these are the processes. You have ratepayer classes arguing over cost causation, and then sort of alongside that, orthogonally, you have special contracts whereby utilities are creating special rates for individual customers. So this is the background. This is the process that's going on. I want to discuss the many ways within that process that costs can get shifted to the public through shenanigans. But before we get to the shenanigans, it's worth making the point you make in the paper that just if rate cases just proceed in the normal way, that's going to involve some cost shifting. Because when new customers come on and you need some new infrastructure to serve the customers, new power lines, say, who pays for those new power lines?
What happens is the cost tends to get split. Maybe, Eliza, we could just walk through that before we get to the sort of shady stuff. Just like the normal way that new customers are treated and that new infrastructure is treated inherently involves some shifting of these new costs onto ratepayers. Can you walk through that a little bit?
Eliza Martin
Sure. So, just when a utility decides that or anticipates that there's going to be increased demand on its system, it just, you know, projects that it will have X amount of new capital costs for its infrastructure.
David Roberts
Yeah, and these are power lines, transformers, that kind of stuff.
Eliza Martin
And, you know, that's the bread and butter of utility profits because they enjoy an opportunity to earn, you know, a rate of return on that infrastructure. So, utilities have, you know, always had an incentive to pursue growth, and so they will just do that in their normal, you know, rate cases. They'll say that there's a lot of demand and that they need to build new infrastructure to meet that demand, and then they'll use the cost allocation formula that they already have approved by the Public Utility Commission to spread those costs across everyone else.
David Roberts
Yeah, we're going to get to those screwy cost allocation measures. But maybe just by way of preview here, it's just worth saying so. Like, you could say, and utilities have said in previous cases, like "X new big customers coming online in our system that's going to require us to build X new amount of transformers and power lines and whatever else that will cost X amount. But that new infrastructure will benefit everyone in our region, not just the new customer. And so, everybody should share the cost of that new infrastructure." Which makes sense if you're talking about adding normal, you know, like a subdivision or a commercial center or normal loads.
But when you're talking about adding a new gigawatt load to a utility area that maybe, you know, only has a couple of gigawatts of demand in the first place, then it seems crazy to share the cost of that across everybody. Then it's very obvious that it's the new customer that you're doing the new stuff for. So let's talk about some of the ways that costs are getting shifted. And one is through going through these special contracts. And it's crazy to me to read about Eliza, but like, these special contracts, as you say, they're often like, "Big new economic development is coming to our area. Let's do this!"
Right. So there's pressure on legislators, there's pressure on the PUC, pressure on the utility: "Let's do this. Everybody wants this. Let's do this." And so they do these special contracts outside of the normal process. And as you write in the paper, these special contracts get incredibly little scrutiny. So maybe just talk about some of the ways that the terms of these special contracts can serve to kind of shift costs to ratepayers. And why aren't they being scrutinized? I mean, I guess that, you know, we're back to the political pressure again. But it just seems crazy to me that some of the biggest, you know, these are some of the biggest decisions you're going to make around your electricity system.
To move those biggest decisions into a private process that bears no scrutiny seems crazy to me. But talk a little bit about what goes on in these special deals.
Eliza Martin
I mean, there's a whole plethora of procedural issues. I mean, when Ari and I were looking through the contracts, I mean, the terms are not publicly available. The utility will make a claim that there's proprietary price information and that the public can't see these pricing terms. But of course, that creates an issue because if there is a cost shift — so if the utility is offering a price that is lower than the utility's cost of service — to serve that ratepayer, then the problem is no one really knows until there is, you know, down the line there may be a rate case and the utility has a much higher revenue requirement.
So, you know, there's this initial claim for proprietary information, and then the regulators often just reflexively approve that. So, I did not run across a contract where the PUC rejected a claim for proprietary information.
David Roberts
Really? Not one?
Eliza Martin
No, I don't. Some were more redacted than others, but I did not see a contract that was totally available, which is unfortunate.
David Roberts
Yeah, it's wild. So again, like a publicly granted utility, monopoly utility, striking a deal with a private party that the public is not allowed to know what the deal is. Again, it seems crazy to me. Ari, maybe you can tackle this one. One of the kind of subtle things that's going on in here is that utilities, again, we're back to utility holding companies that own both sort of competitive businesses in competitive markets and regulated utilities, and they have some incentive to shift costs from the one to the other. So maybe explain how that works in the context of a special contract.
Ari Peskoe
Yeah, I mean, one reason we're skeptical about these deals is because of the long history of utilities exploiting their monopolies to benefit their competitive lines of business. And there's certainly right now a nationwide competition to attract these data centers because building the infrastructure can be a sort of lucrative business for the utilities.
David Roberts
Right. As Volts listeners are well aware. But I'll say it one more time, this is how utilities make money: They make money by spending money. So they want to build a lot of infrastructure. A new data center coming is an excuse to build a bunch of infrastructure. So they all want the data centers.
Ari Peskoe
Yeah, I mean, so utility regulators are supposed to watch out for these incentives and opportunities to exploit your captive customers that basically have to pay your bills. And so, what we're concerned about and what regulators ought to be more concerned about, and it's distressing that there's not more concern in these special contract proceedings, is whether or not utilities are, in fact, through these secret contracts, shifting costs of discounted rates to big tech companies onto other ratepayers.
David Roberts
Right. Maybe go into a little bit more about the distinction a utility has between its competitive and its non-competitive arms, and how this cost shifting works. Like, a little bit about the mechanics of what the cost shift is. Does that question even make sense?
Ari Peskoe
Well, so in this particular case, it's not really a — so there are instances where the utility does have an affiliated business and it can use ratepayer funds to prop up that business, right? So, for instance, the utility owns a power plant that's through one of its corporate affiliates, and it has that corporate affiliate sign a contract with the utility. The utility can sign at an inflated rate because it has captive ratepayers that will have to pay for that contract. And that's a way to sort of funnel money to this business, this power plant, that has to compete in the market.
Here, it's slightly different in that these are both sort of utility businesses. The data center is part of the utility's business, and these sort of captive ratepayers are part of the utility business as well. The distinction is, though, that there are these captive ratepayers that have to pay their bills, and then there's this utility also trying to attract these customers that have the ability to sort of go in all different parts of the country, these big data centers. And again, but it's the similar concern, which is that the captive ratepayers may be subsidizing the utility's sort of competitive lines of business.
And here, though, there's just no affiliate company, but it's the same kind of principle.
David Roberts
Right. Another way of shifting costs is this — call it a gap between federal and state regulations. So, FERC sort of tells the PUC to divide up transmission costs, and the PUC has this cost allocation formula. Talk a little bit, Eliza, about how those cost allocation formulas end up serving as a subsidy.
Eliza Martin
There's a couple of different issues. So, you know, first, FERC approves a cost allocation method for its RTOs. So, you know, like, if FERC approves its cost allocation for PJM, which splits certain costs and assigns them to the utilities, then each state will allocate the costs that are assigned by the RTO to ratepayer classes of every utility that it regulates. So the result of this is really that residential ratepayers who are not maybe causing an RTO to plan new transmission the way that the data center wants, that they're still bearing the cost for that. And transmission is complicated just because, you know, there are benefits that flow from that for all of us, but the way that the transmission is being built and for whom it's being built results in this cost shifting, potentially through FERC approved to, you know, PUC approved cost allocation.
Ari Peskoe
I would go a little stronger on this and just say that, you know, PJM's regional transmission plan was $5 billion last year. They said data center growth is one of the main drivers. PJM's role here is basically to bill each utility, but it's then up to the state regulators and the utility to figure out how those costs are divided among residential, commercial, etc. The bottom line is that in some states, we have residential ratepayers bearing two-thirds of the cost of this new transmission that is really being designed for data center growth. Now, there are some ancillary benefits that ratepayers do receive from this new transmission.
So, it's not like they're not getting any value out of it, but they're certainly not the cost causers of that new transmission. And I think this sort of approach to cost allocation is really on the presumption that we're building broadly beneficial projects. And that's simply just not the case when we have a few identifiable, very wealthy corporations that are driving a lot of this need.
David Roberts
Right. And just to repeat this point one more time, up until now, the assumption that new transmission infrastructure was broadly beneficial to all utility ratepayers more or less held. This is a new state of affairs that you get a gigawatt new customer for whom you have to build considerable new infrastructure. These assumptions that — it's not that this was necessarily a shady or corrupt practice, it's just not suited to the present circumstance, maybe you could say.
Ari Peskoe
That's right. I mean, it's the scale and the speed of the forecasted growth that certainly doesn't have a modern-day precedent, certainly not in the past several decades.
David Roberts
Right, right.
Ari Peskoe
And a lot of the approach to regional cost allocation has really just evolved relatively recently.
David Roberts
Yeah, and then another way PUCs allocate costs is by peak demand. As we mentioned earlier, a lot of infrastructure gets built just to meet these peaks of demand at certain times of day, certain times of year. And so, this is another thing that's just kind of jaw-dropping. Utilities charge big customers peak demand charges for their contribution to those peaks, but apparently, it's entirely possible for a data center to find out what period of time is the utility's peak, thereby what period of time the utility is going to use as a reference, and then it can just reduce its power use for that couple of hours and thereby reduce its demand charges all year.
Eliza, is that right? Am I getting that right?
Eliza Martin
Yes, basically. And it's, we've heard that there are companies that sort of do this, that help people forecast when peak demand is. So, it is a way that you can cost shift if the system is built based on a theory of peak demand.
David Roberts
Yeah, and it's one of the things if you're fiddling around with a couple of megawatts too. But again, if you're fiddling around with a gigawatt, that's going to make a substantial difference to the final outcome.
Ari Peskoe
Yeah. Again, it's this problem of scale because other energy users can also stop using utility-delivered power in order to reduce their demand charges. But the scale of this can potentially have a major impact. And I'd also just add that there are better ways to design demand charges that can mitigate this possibility. But it's really a matter of what the utilities and what regulators have approved, and there are some on the books that can allow for this just massive avoidance of costs. And the other aspect of this is that so many of these data centers have just this armada of diesel generators on site.
So, when this peak demand hour comes, they can keep operating by just turning on all these diesel generators, but they can save money in the process. So, the incentives here are just all unfortunate.
David Roberts
Yeah. And again, it's like one thing if a utility sort of misses a megawatt or two, but it's another thing if it's like the scale we're talking about. And the third way you mentioned about shifting costs, which I found really interesting, is you have some suspicion around colocation here and we're going to get into a different kind of collocation later when we talk about solutions to all this. But it seems like you all see some shenanigans going on under the banner of colocation currently. So Eliza, maybe just talk to us about how that works. What do you think data centers are trying to kind of get away with with this colocation strategy?
Eliza Martin
Well, colocation is, I mean, it's obviously top of mind given the FERC proceeding. But a general colocation arrangement, or the type that we sort of talk about in the paper, is when a data center would connect directly to an existing power plant behind the point of interconnection. And so, by delivering and taking power without using the transmission network, power plant owners and data centers argue that they should be exempt from paying any utility delivery fees or any grid services, basically.
David Roberts
Right, right. So, you got a data center right next to a nuclear plant. They're both behind the interconnection to the grid. The nuclear plant is feeding power directly into the data center. And so, both the nuclear plant and the data center are arguing to the utility, "We're not using your transmission infrastructure for this. It's just a direct connection. Therefore, we should not have to pay these fees and for the upkeep of the transmission charges." That's what they're trying to do.
Eliza Martin
Yes, and there's some discussion about whether or not that really is a feasible technical arrangement, like whether it's fully integrated with the grid or whether it's isolated. So, there are some technical issues around it and then just the regulatory issues, which is, "Is this allowed? Who regulates it?" That's sort of the ongoing proceeding.
David Roberts
Have we gotten PUC rulings on this yet? Like, are PUCs allowing it? What's the trend?
Ari Peskoe
Well, these transmission issues are under FERC's jurisdiction, but there is, to the extent we have a nuclear plant selling directly to a data center, that could be a retail sale that would be under state jurisdiction. So, this type of colocation is probably only allowed in states that have sort of opened up the system a little more to retail competition, which is primarily the Northeast and a couple of other states out there.
David Roberts
Right. I mean, and the thing that strikes me immediately, like if I were a PUC commissioner, I would be thinking, "The data center plopping down next to the nuclear plant is going to siphon off an enormous amount of the nuclear plant's power." And then wherever else that power was going is then not going to have power. And then you're going to have to build the infrastructure for that missing power. In other words, it seems like even if you're not directly mechanically using the transmission and system, it seems like you are still having a substantial effect on the transmission system such that you should pay. Am I — is that crazy?
Ari Peskoe
No, I mean, that's potentially the largest cost shift here is if this nuclear plant is no longer selling into the market. The market prices are set basically by supply and demand. And so, if a huge chunk of supply leaves the market, all else equal, prices are going to go up, at least in the short term. If the nuclear plants in PJM, for instance, can make more money by selling directly to data centers, well, then that's what they're going to do. And there's going to be a shortage, at least for a little while in PJM, and that's going to drive up prices for everybody else.
Eliza Martin
And just to add, you can end up with extending the life of other assets that maybe are dirtier because the price has increased so they can now bid in.
David Roberts
Right, right. You divert a bunch of your nuclear to a data center. Then you have a power shortage. Then you have a case for keeping your coal plant open, which is also a shenanigan that's going on. So then let's talk about what to do about all this. You know, and as we've said over and over again, this is in the utility world, relatively new. This just is a giant new thing that's happening. So, like, everybody's scrambling a little bit. And it seems to me like, just intuitively, the decisions and procedures that get established now are going to have a long tail.
And so, there's a lot of very important decisions being made now, up front. So, let's talk a little bit about solutions. One you mentioned, Eliza, the first one is just there should be some guidelines for PUCs reviewing special contracts. It's crazy to me that there aren't, but what do you mean by that?
Eliza Martin
I mean, in some states, like in Kentucky, there are some more specific guidelines or findings that regulators have to make before they can approve a special contract. So it's really, I think, more about showing your work, kind of, so making sure that, you know, the contract, the rate, you know, doesn't exceed the utility's cost to serve that customer. You know, limiting discounts to certain amounts of years, if there are any discounts. So being able to just sort of see what the utility is proposing doing, and there being a record of that, I think, is really what this comes down to.
But, I mean, ultimately, I think Ari and I are generally pretty skeptical of special contracts, and it ultimately comes down to whether or not the regulators have, you know, the ability, the will to challenge them. So for that reason, you know, we often think that maybe tariffs would be better instead of special contracts.
David Roberts
Yeah, yeah. The way this is framed, if you're going to use special contracts, at least there should be some guidelines for assessing them instead of them just being a black box, free-for-all.
Ari Peskoe
And I think that maybe the most important criteria that Kentucky has is that they only allow special contracts when there's spare capacity, because then maybe there's — you know, effectively that means we can only offer a discount when we're not going to have to add a lot of new infrastructure to the system. And that's just simply not typically the case these days that we have a lot of spare capacity around. So that could eliminate the possibility of a number of these contracts across the country if we put in that criteria.
David Roberts
Yeah, and you all make a note that, like, guidelines can only do so much. In the end, a lot of this just comes down to you needing good regulators. Like a lot of this just comes down to you needing empowered, informed, and good-judgment-having regulators. But as you say, special contracts in general seem shady. And I completely agree. It's amazing to me that they exist. It's hard for me to imagine a principled justification for their existence. It just seems like special deals for big customers who want to rush past the process, which, like, you know, I get why economic development officers like that.
But it seems unfair to the other people using the system. So, the second thing you counsel is just quit using these special contracts and shift to tariffs. And so, here we get to what I thought is a really interesting question that's being fought out now in utility proceedings all across the country, which is: should data centers have their own special tariff? Should they just be treated like other industrial customers have the same tariff that other industrial customers have, or should they have their own for some reason? And this battle is being fought in Ohio, Indiana. FERC weighed in a little bit.
What is the case for saying, "Yes, data centers are special and deserve a special tariff."? What is the argument for that side?
Eliza Martin
It's my understanding that the argument is really based on this idea that similar consumers, similar utility customers should be grouped together if they have similar usage. The cost of service for the utility company for providing service to these customers is the same. So, that's why you have sort of this ratepayer class or the consumer residential class. That's why you have an industrial class. There's an argument that data centers, because they are more energy intensive and run 24/7, are materially different from other industrial users of energy.
And FERC got into some more nitty-gritty issues around cryptocurrency customers that they can move, so there's greater stranded asset risk and stuff like that. So, utilities have a requirement to provide non-discriminatory service. The issue is whether or not they can discriminate for data center customers, basically to isolate them from other consumers.
Ari Peskoe
I wanted to pick up on one point there, which we didn't discuss earlier, which is possibly the biggest potential cost shift here, which is the stranded asset problem.
David Roberts
Yeah, yeah, yeah, let's talk about that.
Ari Peskoe
Yeah, I mean, you know what makes data centers potentially different from other large customers is the riskiness. So, we can imagine all sorts of reasons why this data center growth may fizzle out over the next few years.
David Roberts
Yeah, yeah. I mean, we should say these projections are very uncertain. No one knows.
Ari Peskoe
Right.
David Roberts
No one knows what's actually going to happen with data centers.
Ari Peskoe
And so, the concern is that utilities are going to start building all of this new transmission, power plants, et cetera. It's going to be billions of dollars of costs. And then, the data centers are going to try to cancel their contracts and disappear.
David Roberts
Yes. Because also, what's going on behind the scenes is that you've got multiple states, multiple utilities wooing these data centers. So, the data centers have every incentive in the world to play them off one another and to sort of flirt with each one. Right?
Ari Peskoe
Yeah, or just, you know, it turns out we don't need all these data centers, maybe. And so, one of the unique aspects that utilities benefit from their business model is the idea that they can socialize these risks to ratepayers. There's a history of utilities trying to pass on these costs of these stranded assets to ratepayers. So, one of the key issues in these tariff proceedings is how to ensure that data centers are on the hook for infrastructure costs that the utilities are incurring just to meet these data centers and preventing consumers from being on the hook.
David Roberts
It's one thing if you build 10 megawatts of infrastructure that you don't need, but again, it's a totally different thing if you build the infrastructure necessary to handle a gigawatt of new load and then the gigawatt doesn't show up. That's just a very — that's a different class of spending.
Ari Peskoe
I mean, so in this Ohio proceeding, the utility AEP argues that data centers are distinct because of the sort of newness and riskiness of this new type of customer.
David Roberts
Are PUCs weighing in on this? What are PUCs saying so far about this argument?
Ari Peskoe
We are on the edge of our seats, waiting for the Public Utility Commission of Ohio to weigh in on this one.
David Roberts
Interesting.
Ari Peskoe
Yeah, I mean, as you mentioned, FERC did weigh in and just said it was just one of those sorts of proceedings where FERC just said, "There's not enough evidence in this particular docket for us to make a decision in your favor." But they were not sort of uniformly opposed to data centers being different. They just wanted to see a stronger case made.
David Roberts
Right. This seems like such an important question, how this comes out. It seems incredibly significant. So, yeah, there's the whole argument about whether you have a special tariff for these things at all because of their scale, because of the riskiness of abandoned assets. And just because, like, you know, building a giant power line to serve a gigawatt load is not a sort of universally beneficial thing in the way that like previous infrastructure buildout was. It's like very bespoke and very specific. And it's not like you have another gigawatt load that can just slot in there if the data center doesn't show up.
So, in response to these dilemmas and this, you know, this question of whether there should be an individual tariff, a special tariff for this class is like, again, there's no capital R right answer, somewhat subjective. So, in response to sort of these concerns and these questions, y'all end up recommending or sort of coming down in favor of the idea of energy parks, which we discussed here on Volts a couple of months ago. But maybe Eliza, so maybe just tell us like what do we mean when we talk about energy parks and why is there an advantage here? Why does this sort of like clarify this question?
Eliza Martin
So, an energy park, I guess, is sort of also a form of co-location, but not the one that we were talking about earlier. But basically, the data center customer would bring generation assets and they would be totally isolated from the grid. So, you would avoid this sort of utility interconnection processes. You would be able, if you were a data center customer, you could choose what electricity was the cheapest or if you, you know, some of these big companies have commitments to get clean energy. So, you would be able to isolate yourself from the utility network to bring your own generation.
David Roberts
And then, if you're doing that, you're very clearly paying your own costs. Right. Like that's just, that's kind of like settles the question of cost shifting. Like if you're islanded from the utility, building your own generation and your own connections, that's a clear-cut case of like you're, you're paying your own costs and we're not paying for you. So, like, can that happen anywhere right now or are there regulatory circumstances required for that to work?
Eliza Martin
There are regulatory changes that need to happen, yeah. Really, there is just state protection of utility monopolies. So, state laws would need to be amended to allow these projects.
David Roberts
Right. So, if I'm behind the meter building a data center and then I build a giant solar panel field next to it, and the solar field is powering the data center, I am in the state's eyes selling power, which only utilities are allowed to do. Right. So, what you'd need in that case is, I guess, what we now call "retail competition." Is that the same thing?
Ari Peskoe
Yeah, I mean, I would just say we did not do a 50-state survey of this issue. This is one of those things where I can give the generic response of "it varies by state." And so, I think there are many states, particularly where we have the traditional vertically integrated utility that's built all the power plants, where they fiercely defend the scope of their monopolies, and this would most likely not be allowed. Then you have on the other end of the spectrum a state like Texas, where particularly if you're in some of the top parts of the state where there are rural cooperative utilities, you may actually be able to do this sort of arrangement today.
But I think, you know, in most of the country, we would likely have to change some of these laws that establish utility monopolies.
David Roberts
And that's a legislature thing, not just a PUC thing.
Ari Peskoe
Most likely, yeah. And you know, obviously, the utilities would generally not be a fan of this sort of change in state law.
David Roberts
Just because — well, explain what's the political economy here?
Ari Peskoe
Well, the utilities' monopolies are really their most valuable asset. It provides steady and perpetual profits, and so they generally fight against any effort to weaken their monopolies. There's a lighter version of an energy park which is just that the data center would be the utility customer, but the data center would be allowed or even required to contract itself for new generation. And that would certainly mitigate cost shifts because building power plants is quite expensive. And if you let the utility do it, it then has opportunities to push some of those costs to other consumers. But that would be sort of a compromise position that still a lot of utilities would oppose because they make a lot of money by building those power plants and want to have those opportunities for themselves.
David Roberts
From a kind of wonk perspective, it seems perfect, right? Because it's like a giant new customer, clearly paying its own freight, clearly paying for its own power, clearly paying for its own connections. It's very tidy from a wonky perspective. But it seems to me that the political economy is against it sort of at every turn. Like, is there a substantial, you know, constituency in favor of this solution?
Ari Peskoe
Yeah, I mean, I guess I would clarify that this lighter option is not foolproof for protecting consumers because it would still be a utility customer, presumably still paying for all the delivery charges. There'd still be some opportunities to play some funny games and shift some costs there. But we went through this about 30 years ago when several states did remove generation from what the utility does. These utility restructuring laws that several states passed, and one of the justifications, rationales for it, was that utilities in some parts of the country didn't have the financial strength to fund power plant development, so sort of getting it off their books.
And so, to the extent that there's just so much growth happening in some parts of the country, there maybe could be some argument for getting this off utility books. But I don't know if utilities would go for that these days.
David Roberts
One other solution you mention, which is also an intriguing and ongoing argument, is the idea that maybe new data centers should only be allowed to commence service if they commit to being flexible. So, there's a lot of argument right now about whether they can be flexible. If you ask the hyperscalers, they'll tell you, "No, we need 24/7 power, we need foolproof power all the time. It costs us billions to shut down," et cetera. But then there's been some new research. Tyler Norris and his group had a paper about it. There's been a lot of talk about maybe there is some flexibility.
So Eliza, maybe just talk about why would it be, what would you get out of this?
Eliza Martin
So, because as we sort of talked about, the system is built for peak demand. We spend a lot of the time during the year not reaching peak demand. So, there's theoretically a lot of assets that we've already built that could be used. So, that would just reduce the short-term investment that we need in infrastructure that, you know, we all are getting charged, would get charged for in order to bring on more data centers, to serve more data centers. But there's obviously, you know, there's the issues that you talked about, which is that we aren't sure whether or not data centers would really commit to this flexibility.
And then, utilities have resisted efforts to try to be more efficient in the past, whether that's with non-wire alternatives or GETs or other issues. So, I guess I'm a little bit skeptical that it could actually play out in practice. But, I think there is a way where regulators could require utilities to condition service to data centers on being flexible.
David Roberts
Yeah, and you cite a study in the paper that shows that if these giant data centers could just be a little bit flexible, just shut down for a few hours a year, it could save billions of dollars, which it seems to me like is somewhat in their interest. Because if it becomes easier for a utility to say, "Yes, go ahead and interconnect," then you could just get more data centers online. So, it seems like it's to their benefit too. Ari, do you have any perspective on the question of flexibility in data centers?
Ari Peskoe
Well, it eats into the utilities' profits because the point of flexibility is to reduce the new infrastructure we need, and that's how they're making their money. So, you know, as Eliza was talking about, that's why utilities have historically been, at best, lukewarm on these sort of flexibility initiatives like demand response.
David Roberts
They like high peaks. I guess it justifies their spending.
Ari Peskoe
But again, maybe we're in a situation where growth is escalating so much, so fast, that they have to start thinking outside the box. But our concern is that actually this data center growth is just causing them to just entrench all of their existing worst practices and just double down on their narratives and not really attacking these sort of opportunities to reduce costs for everyone. And instead, they just want to keep building more infrastructure.
David Roberts
Right. This is like an alcoholic, the booze was starting to run out and like, a new truck has showed up outside with barrels of whiskey on it. So well, we're getting near the end of time, so let's transition to that then. So those are a few solutions, which is like, I gotta say, and I know this is not, you know, your fault, but, like, none of these are really, like, super satisfying. All of them are a little bit piecemeal. All of them are going to work in some PUCs and not others. Some of them require legislative changes, some of them don't.
So, there's just a lot of, like, distributed work that needs to happen to prevent this from taking place, to prevent the public from getting stuck with these costs. But to sort of wrap up, like, the last section of the paper — so you go through a thing where you say that data centers don't necessarily pass the cost-benefit test because in addition to getting subsidized by utilities in all these various ways, they're getting, like, economic development grants, they're getting state grants, they're getting local municipal grants, they're getting, like, sales tax exemptions. It sort of reminds me of the sports stadium issue, right? Like, everybody thinks they want these things and everybody thinks they're great.
But, like, once you subtract out all the subsidies you spent to get them in your area, are they really paying out anymore? Anyway, you sort of argue that, like, they don't deserve these subsidies anyway. And all the sort of, like, national security arguments to the contrary are a little bit silly. But the one point I wanted to hit on, and Ari, I want to hear you talk about this before we're done. So, you say one of the dangers here is that subsidies to data centers could interfere with needed power sector reforms. There's a little couple of sentences I want to read here from the paper.
"As utilities wring profits from the public through special contract approvals, they may be developing a new alliance with Big Tech. Uniting utilities' influence-peddling experience with the deep pockets of Big Tech could further entrench utility control over the power sector." I will say, Ari, these are two of the most chilling sentences I have encountered in any PDF in years.
Ari Peskoe
Yeah, I'll take that as a compliment.
David Roberts
I mean, this is like, genuinely terrifying that basically, Big Tech's going to come along with giant bags of money and it's just all going to serve as an excuse for utilities to keep doing what they've been doing. Right? Like, in a sense, this is, they were nearing a situation where they were going to be forced to change their ways. But now, like, here comes all this new load, big bags of money, base load coming like, they're like, "Yes, we know how to do this. You build gas plants and expensive new transmission lines and get a rate of return. We're all over this." So, explain the political economy danger here, basically.
Ari Peskoe
Yeah, so we're seeing a lot of data centers being met with sort of low-hanging fruit in the utility world, which is big transmission projects and big gas-fired power plants, the sort of stuff the utilities have been making money off for generations. And as you know, Volts listeners know there's about 100 different ways we could be reforming the system to improve outcomes for consumers and for clean energy. The concern that we have is that that momentum is just being completely lost because of these shiny data centers that utilities absolutely cannot resist. They say, "The only way we can meet this urgent sudden need is by doing what we've always been doing."
Now, there's a long history of large energy users, like the big industrial consumers, kind of being watchdogs in utility rate cases because they have the biggest incentives to make sure the utility is not being wasteful. But, what if we have a world where the largest customers are now all getting on the system here in the 2020s through these special contracts?
David Roberts
Right. So, they then don't care if the utilities overcharge everyone else, right?
Ari Peskoe
Well, their biggest interest will be keeping those special deals going. And so, yeah, they may not be concerned with the rest of the system. And that's really the concern, is that they become allies of the utilities.
David Roberts
Not just allies, but like borderline owners. Like, if you're, you know, like some of these utilities are going to have a situation where the data centers are the bulk of their total load, which means that those customers are going to have incredible influence, just incredible power over the utility.
Ari Peskoe
Yeah, I mean, that's the concern here. And then that just entrenches the status quo, which has been the utility bias forever. Because this cost of service rate-making model has been working for them for a century.
David Roberts
And if they're allowed to use that basic model on data centers, they're not wrong. They will profit immensely because they are going to have to build a bunch of infrastructure. I mean, in other words, we're once again in a situation where utilities are acting against the public interest but are accurately acting in their own interests, right?
Ari Peskoe
Oh yeah, they're absolutely following the incentives. And it's, I think, a sort of inversion of what the big tech companies had been doing in this space. I mean, they had for years been advocating for increased competition, particularly in parts of the country where we don't have RTOs. But I think there's a real danger here that they just sort of switch gears and really focus on just getting these data centers connected through the utility processes and then protecting whatever special deals they get.
David Roberts
Whereas the alternative path that you all outlined, where data centers power themselves, then they become sort of a counterweight. Then they become almost competitors to utilities and then you have like a balance of powers keeping each other in check kind of thing.
Ari Peskoe
Yeah, I mean, that's our theory. I mean, I suppose you could spin other theories here as well. You could be really optimistic and think that somehow Big Tech with all of its technological powers will somehow modernize the utilities and bring them into the 21st century. I suppose you could come up with other theories. But our concern is the story that we lay out. It's a somewhat pessimistic view of how this could play out.
David Roberts
Well, I mean, it's hilarious hearing me argue in favor of pessimism. Of course, I'm always pessimistic, but to me, just the quantity. I mean, we've mentioned scale over and over again in this pod, but you can't mention it enough. It is a change of degree sufficient, I think, to amount to a change in kind here. What's happening, like what's happening around utilities, is fundamentally different. And right now, everybody is in panic mode, everybody's in growth, everybody's in beat China, everybody's in, "Oh my God, the other company is going to get there first." Just like to me, it looks like they're happily going to jettison their scruples about clean energy and good utilities.
You know what I mean? It just seems obvious to me. They're all just stampeding in one direction.
Ari Peskoe
It does seem like speed is the priority. And yeah, I'm not sure it seems like that's going to carry the day. But you know, this situation has changed pretty quickly over the last couple of years. So, you know, again, it's going back to the stranded cost issue. It's one reason why I think that has the potential to be the biggest driver of cost shifts is if this turns around completely the other direction in the next couple of years and then we have all these half-built transmission lines or something like that.
David Roberts
Oh my God.
Ari Peskoe
And then, consumers are left holding the bag.
David Roberts
Yes, well, this is, at the very least, something that, you know, if PUC commissioners are listening, I know some of them are. I guess it's ridiculous for me to tell them they need to be thinking about this since that's all any of them are thinking about, I'm sure. But, like, this is definitely the question of our time.
Ari Peskoe
Yeah, and go back to something that Eliza said earlier, which is just the value of good regulators here. Right?
David Roberts
Comes back again and again.
Ari Peskoe
Yeah, like, there's no silver bullet, as we've also heard many times on many, many Volts episodes. But there's also no replacement for just regulators doing their job really well and trying to protect consumers as best they can. And you know that that can be really hard when you have all this sort of background noise about national priorities and economic development.
David Roberts
Yeah. But again, I'll just reiterate that, like, decisions made now are going to compound over time. So they're very important that we get a lot of this right on the front end, I would think. All right. Well, this has been fascinating. Somewhat ominous, but fascinating. Eliza Martin, Ari Peskoe, thank you for your work, the paper, and thanks for coming on.
Eliza Martin
Thank you.
Ari Peskoe
Thanks for having us.
David Roberts
Thank you for listening to Volts. It takes a village to make this podcast work. Shout out, especially, to my super producer, Kyle McDonald, who makes me and my guests sound smart every week. And it is all supported entirely by listeners like you. So, if you value conversations like this, please consider joining our community of paid subscribers at volts.wtf. Or, leaving a nice review, or telling a friend about Volts. Or all three. Thanks so much, and I'll see you next time.
Share this post