How will data centers pay for power?

Right now, connecting a data center to the grid works like Texas hog season: no defined season, no bag limits, first-come-first-served, file as many interconnection requests as you want. Travis Kavulla’s recent essay in American Affairs argues the power industry needs something closer to deer season, with defined rounds, allocation rules, and prices that reflect what grid access actually costs.

The mechanism he favors is an open season, borrowed from interstate gas pipeline regulation. Rather than processing interconnection requests on a rolling basis, a grid operator would design an engineering plan reflecting realistic demand and tender it to the market in a structured bidding process. Winning bidders receive transferable property rights to grid access, comparable to water rights or spectrum licenses, rather than vague regulatory permission that reverts to the grid operator if a project fails. That difference matters for financing: a transferable property right holds residual value even if a data center company does not survive the artificial intelligence boom.

Kavulla has worked as a utility regulator at the Montana Public Service Commission, on market design at California ISO, in policy at R Street, and in regulatory affairs at NRG. He now leads policy at Base Power and teaches at the University of Chicago.

The conversation also covers three other mechanisms:

* Data center prepayments for discrete capital costs are, in Kavulla’s framing, the most direct fix for protecting existing ratepayers. Utilities resist them because prepayments erode the rate base growth that regulated utilities depend on for earnings.

* Transmission service agreements (contracts requiring upfront financial commitment before interconnecting), widely adopted across the eastern United States, rank a distant third. They base the commitment on average embedded rates rather than actual incremental cost, which overcharges some projects and undercharges others.

* Bring your own generation addresses a separate problem: demand growing faster than supply pulls the clearing price higher for all customers. Data centers that source their own capacity or pay for flexibility elsewhere ease that pressure. Utah and West Virginia are among the first states opening pathways for large loads to do this.

On ERCOT, host Joshua Rhodes frames the gap: the batch zero process rations grid access but does not price it. Kavulla affirms the distinction and argues cost-of-service regulation has been stretched past its breaking point. How Texas resolves these issues will shape its next interconnection rules and what current ratepayers carry as load climbs.

Timestamps

* 00:00 - Introduction & Travis Kavulla

* 01:59 - The Essay’s Core Premise

* 03:45 - Open Season: From Hog Season to Deer Season

* 08:07 - What Actually Gets Auctioned

* 09:45 - Data Centers Fronting Capital, and Why Utilities Resist

* 13:37 - Transmission Service Agreements and Their Flaw

* 17:41 - Do Data Centers Raise or Lower Rates?

* 22:55 - The ComEd and Dominion Problem

* 25:14 - Bring Your Own Generation

* 26:53 - Reforming Monopoly States: Utah and West Virginia

* 30:51 - Batch Zero vs. Open Season in ERCOT

* 35:43 - ERCOT’s Flexibility Tools and Speed to Power

* 42:08 - Which Idea Has the Best Shot

Resources

People & Organizations

* Joshua Rhodes (LinkedIn)

* Webber Energy Group (Website - LinkedIn)

* IdeaSmiths (Website - LinkedIn)

* Travis Kavulla (LinkedIn)

* Base Power Company (Website)

* American Affairs (Author page)

* Energy Capital (Website - LinkedIn - YouTube)

* Texas Energy & Power (Substack)

The Essay at the Center of This Episode

* How Will Data Centers Pay for Power? — Travis Kavulla, American Affairs (the essay this conversation is built around)

Studies, Cases & References Discussed

* Factors Influencing Recent Trends in Retail Electricity Prices — the Lawrence Berkeley National Lab / Brattle study on load growth and rates, explained by a Brattle researcher (the North Dakota “headroom” finding Kavulla references)

* FERC Rejects the Amazon–Talen Co-location Agreement at Susquehanna — the co-location dispute Kavulla cites

* FERC Upholds the Amazon–Talen Rejection on Rehearing — the March 2026 follow-up

* Kavulla Outlines His Data Center Interconnection Proposals — RTO Insider coverage summarizing the open season, prepayment, TSA, and BYOG ideas

Related Energy Capital Episodes

* How Texas Decides Which Data Centers to Connect — Tiffany Wu on ERCOT’s Batch Zero process

* Who Pays for the New Grid with Pablo Vegas — ERCOT’s CEO on load growth and who pays

* Texas Growth Is Running Into Power Grid Limits with Katie Coleman — large load interconnection and cost allocation

Transcript

Joshua Rhodes: Hey everyone and welcome to another episode of the Energy Capital Podcast. I’m really excited to have Travis Kavulla here to talk about an essay that he recently wrote in American Affairs, where he’s the energy editor about how data centers will pay for power. It’s a really hot topic right now, and I’m excited to dig into that and how it kind of ties in to what ERCOT is doing for data centers. Before that, if you’re in the energy nerd space like you would be if you were listening to this podcast, you probably know Travis already, but He’s got a pretty impressive background that I’m going to go ahead and go through now. So Travis went straight from Harvard to the University of Cambridge, which I found funny that you went from one Cambridge to another, I guess. I guess that was easy for the well, I guess postcard you’re going from Massachusetts to the UK, but still.

Travis Kavulla: The imitator to the original.

Joshua Rhodes: Yeah, fair enough. Should I call this New Cambridge? But that’s not a thing. That’s not what we call those. Anyways. After that, you spent some time in the media with National Review and some freelance work in the UK, East Africa, and the US. But then I did not go back and look to see what you wrote for National Review. But at this point it looks like you kind of pivoted into energy and in particular the regulatory space around energy, which is what we’re gonna talk about today. But you started out as a board member for the Western Electricity Coordinating Council before becoming a commissioner at the Montana Public Service Commission, then jumping over to California to work on the energy and balance market with CAISO, the California independent system operator. Then going to R Street, where you’re the director of energy and environmental policy. And then I think your longest period of time, and probably most people know you as the VP of regulatory affairs at NRG. NRG is big in Texas, and so that brought you into Texas quite a bit. And so we were very happy to have you here. Currently a lecturer at the University of Chicago and the head of policy at Base Power, also based in Texas and Austin actually, Travis Kavulla. Welcome to the Energy Capital Podcast.

Travis Kavulla: Thank you for having me, Josh. Great to be here.

Joshua Rhodes: Yeah, so we wanted to bring in and talk about this essay where you’re talking about of how data centers should pay for power. And really this essay, we’ll link to it in the show notes. It’s brilliant. And I think it’s getting read pretty widely, including I saw on LinkedIn from FERC Commissioner David Rosner. So that’s great. One thing I was gonna say when I was going through your list of things is that I’ve already put my money in Polymarket that one day you’ll be a FERC commissioner. So I don’t know. It’s a pretty good bet right now. Just kidding. Not not the case. But at least your work’s getting read by them, so that’s great. So and I’ll go ahead and try to lay out the premise of your essay, and you can tell me where I get this wrong. But essentially, it’s like that the US should stop treating hyperscale data centers demand as just another low class to be included inside traditional utility rate making, with a core idea that scarce grid access should be explicitly allocated and priced, and new data center loads should pay its incremental cost, and that new generation, the bring your own generation, should be essentially required. Other than, you know, socialize through the utility rates. Did I get that roughly right?

Travis Kavulla: You got it down and you didn’t even need ten thousand words, Josh. Yeah. I mean the basic premise here is that the typical utility business model, which involves estimating what load is going to be on the system in the future and beginning to make capital outlays in advance of that load showing up in a lack of knowledge about whether it will appear, is not a good fit, given the amount of uncertainty over which loads are going to show up, or in terms of cost allocation when you have obviously an inflationary price environment. Where serving the next unit of demand is so costly. So it’s compounding risk and costs that need to be better tied back to the future users of the system. That’s the fundamental insight.

Joshua Rhodes: That totally makes sense. I mean, you know, I’ve been steeped in ERCOT for a long time and I haven’t believed these large load numbers for years, right? And they just keep getting bigger and bigger and bigger and bigger. And I know in other regions that we’ll get to like PJM are also, you know, experiencing these large increases. And so one of the things you do is you kind of argue for a few different ways around if we’re going to try to connect as many of these large loads as possible, you argue around a few different concepts. One of them is like an open season for access. And I just wanted to translate this into Texas Speak just for the audience. So open season. So right now it feels like that connecting large loads is kind of like hog season, where there is no season. And it’s just anyone all the time, as much as you can, please, like first come, first serve, whatever. But kind of moving more to like a deer season, where there’s an actual like defined season and bag limits in terms of you can’t have all of them, you can only have a certain amount of them. Did I get that right?

Travis Kavulla: That’s right. Yeah. I’m from Montana and I’m trying to think of the parallel to hogs. I’m glad we’re not cursed by that particular infestation up where I’m from. But that’s right. Having a defined process and season where people both put in requests, and then I guess if we’re going to layer on another hunting metaphor, it’s sort of like how certain jurisdictions reserve a few out of state licenses that can be auctioned off or paid for by a premium to the hunting authority. Well, normal users of the hunt, people who are lucky enough to have a Texas driver’s license, operate by a somewhat different set of rules.

Joshua Rhodes: That makes sense. And so that’s kind of your like your preferred method. So could you talk through kind of how that might look in practice? Like how might that be set up? Let’s just take a step back and kind of talk about what is the problem it’s trying to solve. Like why do we need an open season for grid allocation right now?

Travis Kavulla: Yeah. So right now the power grid, both on the large load side and the generator side, works on kind of a first come, first served basis. And if you want to interconnect a new data center or develop a new power plant, you file an interconnection request and you take a spot in line. And the people who file these requests realize that there’s a lot of uncertainty involved about whether the project will ever get developed. There’s a lot of uncertainty about the price will be quoted to interconnect to the system. And in view of those uncertainties, you begin filing more and more and more requests, hoping that one or two of them pan out. And it leads to this sort of gold rush mentality, which is where we are today, with interconnection queues that express way more demand and way more power generation than are ever realistically going to be connected to the grid. So an alternative approach is an open season. They’re used actually in the natural gas pipeline industry. Okay. And there, you know, the pipelines will assess the amount of incremental interest in shipping gas on pipelines that exists in the market. They’ll then take back those informal expressions of interest and they’ll take it upon themselves to design an engineering plan, you know, which sort of reflects their best judgment of the actual demand that can be served within a reasonable amount of time, within a reasonable amount. Of capacity expansion. And then they tender that plan back out to the participating public, the market participants, in what is called an open season. And it’s left to their bidding behavior to determine whether or not the pipeline is going to have adequate revenues or more to the point, what kind of optimal configuration of off takers results in the highest net present value relative to the project’s costs. And then that configuration is the thing that subscribes the pipeline. So it’s a process by which you both efficiently size an expansion of a grid as well as creating a cost allocation mechanism to pay for it in one fell swoop. But think of it as a utility planning exercise that’s subject to a strong market based check.

Joshua Rhodes: Okay. Sticking with the natural gas side of things, is that more in the interstate pipelines? In Texas we’re always having to distinguish between interstate and intrastate because there’s wildly different, shall I say, regulatory constructs that govern those two. So where is that happening?

Travis Kavulla: There are, and this is what I’m describing as the FERC procedures for interstate gas pipelines. Though certain intrastate uses them as well, as well as facilities like gas storage.

Joshua Rhodes: Okay. So in your ideal design for an open season, in the electricity sector, can you tell me what is the actual unit that is being sold or auctioned? Like what is interconnection priority, withdrawal rights, substation capacity, transmission upgrade rights? What is the actual thing that would be auctioned off?

Travis Kavulla: Here, what I have in mind and subject to permutations would be a denominated quantity of withdrawal that you would be able to say, Listen, I participate in this open season, and as a result of my successful bid, I have the ability to withdraw four hundred megawatts of power at or about this location. But it ties closely together to the concept of interconnection priority as well. Because with an optimal plan in mind, it becomes the sort of engineering and construction plan basis on which the utility proceeds. And an important part of the economic logic is defining the likely timing of utility infrastructure’s construction in order to facilitate the bidding process.

Joshua Rhodes: Okay. There’s a lot of parallels here to kind of the ERCOT batch zero process and other types of things, which I’m hoping to get to a little bit later, although there’s differences here. I just want to put a mental pin in that to come back. ‘Cause there’s like roughly four mechanisms that you talked about. So there’s open season, kind of an auction based process. It sounds like a point of interconnection size, right? Yeah. It means like I can take this much, maybe I can also push that much. I don’t know. There’s permutations, like you said. The second I don’t know if it’s second best or if there’s really an order, but like the data centers would front capital to build this stuff. But utilities might be resistant against that. Why would that be the case?

Travis Kavulla: Yeah, so you’re right. The second one I propose in the essay, the second approach to paying for incremental grid costs is just a data center which knows enough to flash a wallet to raise their hand and say, Listen, you know, utility, you’ve studied my interconnection. You’ve estimated it’ll be these three facilities that need to be constructed at a cost of three hundred million dollars. We, the data center, and maybe even you, the utility, think there are probably some spillover benefits to other customers of the grid by virtue of you constructing those. But nevertheless, we’re just going to offer to prepay that amount of money. And you know, utilities don’t like this candidly because the utility business model since the industry’s inception in the United States at least has been really a spend more, make more type of business model. I mean, we’re all familiar with the fact that TND utilities are, you know, legally instituted monopolies, but that’s sort of a short way of expressing what they really have and really want a monopoly on. Which is the opportunity to invest capital in their system to the exclusion of others because it is those capital expenditure opportunities that create the so-called rate base on which utilities earn a return. And if you have customers prepaying for capital expenditures the utility would make, that actually erodes their opportunities for earnings growth, which for them is a bad thing. I mean, any other business in the American economy would be tickled pink. At their customers saying to them, listen, we just want you to write a check right now to pay for all of the capital needs that you’re going to incur to serve us. But in the bizarre world of utility incentives, that incentive is exactly the opposite in utilities. So in order to make this work politically, it might not be, even though it is sort of the first best in terms of protecting other customers, I would say it may be necessary to continue to allow utilities to have that investment opportunity. Nevertheless measure the discrete incremental costs and then convert those costs into some kind of data center specific facility charge that they would pay as well. So I note, you know, there’s a proposal that Microsoft put in front of the Public Utilities Commission of Nevada that kind of closely tracks this idea. It just coincidentally kind of came out the week after I wrote my essay. But I think those guys have sort of an inspired approach. The the trade-off here is that in an open season In a data center rich environment like ERCOT, you know, you would be able to actually plan efficiencies and economies of scale around infrastructure build. So the prepayment approach probably makes more sense in jurisdictions like Nevada, where it’s in all likelihood ones, two Z ad hoc stuff, as opposed to larger utility systems that are seeing a ton of data center development and which should be planned for more jointly.

Joshua Rhodes: Yeah. And this is probably gonna show my ignorance when it comes to more kind of regulated entities. One of the things that popped into my head is like in this kind of front end capital, the if the utility wants to spend CapEx because it make a regulated rate of return and that’s what they’re going after, I mean, could the data center offer like a zero percent loan to them such that then they could spin the money, get the regulated rate of return, and then does that math even math?

Travis Kavulla: Well, usually, I mean, the issue right is that these are cost of service regulated industries. And usually if the utility accesses a loan at a particular stated interest rate, then the regulator will use that for the purposes of rate making. Uh-huh. I mean, the utility, if it is going to invest its capital, should raise capital in a economically efficient way and try to tie off the cost and risk of that financing facility to the particular customer it’s spending on behalf of, but All of this conversation ties back to the kind of Looney Tunes way we regulate the utility industry in this country, which is cost of service, which spawns a bunch of weird incentives, including this one.

Joshua Rhodes: You already mentioned the Microsoft deal, which I assume they were inspired by your essay since it came out after. But in my head, that was more around like kind of the third thing, the the transmission service agreement. I was actually having a little bit of difficulty kind of pulling these two apart. So what’s the difference in the fronting capital versus like the third argument there, the transmission service agreement? I know you’ve been critical of a bunch of them, but maybe there’s one or two that are popping up that look all right to you.

Travis Kavulla: Yeah. So the Transmission Service Agreement is something that utilities in the eastern United States have been prolifically using. And basically there, it’s a way to get data centers to pay up big table stakes before they’re interconnected to the grid. So it does, like the other two, it eliminates some of the gold rush that has occurred on load interconnection queue. It’s big failing, I think, is that TSAs to date. Have all been based on the kind of backward looking average embedded transmission rate. And so let’s imagine that I build a data center on an electrically robust brownfield, maybe where a big industry has, you know, retired a manufacturing plant and which may not need a lot of capital upgrades to the transmission system. I mean it’s, you know, 500 megawatt data center. And then imagine I build the same 500 megawatt data center. In a green field, that requires hundreds and hundreds of millions of dollars of utility infrastructure spending to accommodate. Under a TSA arrangement, I would be quoted the same take or pay collateral requirement for either of those projects. It’s tied, in other words, to the size of my project and not to the actual incremental cost to the utility to serve that project. So it fulfills one ambition, which is winnowing out kind of wheat from the chaff. Yeah. But it fails to send an efficient price signal. And it may not be adequate in the case of all projects to guarantee a sufficient amount of revenue to actually cover the incremental cost of the build. So I’d put it in a pretty distant third place, but it’s better than nothing. It’s better than returning to the status quo ante, where utilities are just outlaying capital based on pure speculation about where we’re going and not expecting anything. Upfront in the form of commitments from data centers to take service from the grid.

Joshua Rhodes: No, that makes sense. I mean, I’ve historically when I was looking in other areas, I think correct me if I’m wrong, one of the first take or pay kind of contracts, it’s kind of similar here, I think, was in Ohio around data centers. And it sounded good to me, I guess, until I read your essay where you talked about how I guess the problem there is they’re using these average embedded rates, like they’re historical looking, but it looks like the future is gonna be different than the past. So can you explain like how would looking to the past for this in the current day and age not be adequate? I guess. Can pull that thread a little bit further.

Travis Kavulla: Right. Yeah. It’s really as simple as the fact that the cost of service implied by the rates you pay today is all based on infrastructure that was costed out over the past several decades. And that infrastructure costs less than new infrastructure does today. It might not be the case if we were not in an environment where poles and wires and switch gear and copper and everything else weren’t rapidly inflating in price. But that is where we are. There’s a lot of macroeconomic inflation. There’s even more inflation in this sector, specifically because of the demand growth and the inability of supply chains across the sector to keep pace. And so when you have people pay an average embedded cost rate for the service, you’re not actually acknowledging the trend that building something new is more expensive than, you know, the thing you own free and clear. And at a basic consumer level, you could. Extrapolate this. But most people, most businesses, you know, wanna make use of capital assets as long as they can because they realize the new thing that they turn around and have to buy is, you know, going to cost a lot more money.

Joshua Rhodes: Yeah, which is I mean, I think this brings us to there’s been a lot of studies that have been coming out recently about the impact of large loads and data centers and particularly on electricity rates. I remember about a year or a year and a half ago or something, like it was in the span of one week I got contacted about being a part of like three or four of like these large I mean, everybody was trying to figure out what is gonna be the impact, what is gonna be the impact, what is gonna be the impact. You know, it really depends on, you know, where you’re drawing the box around that. You know, there’s a Berkeley lab study showing low growth can sometimes reduce prices when they point to places like North Dakota and places that had headroom to move into, and you’re increasing utilization rates and that can, you know, reduce the per unit cost. And then you’ve got VJM’s market monitor saying, you know, the data center low is materially increasing, you know, capacity auction revenues and other other types of things. So it’s like, is everybody right or is everybody wrong?

Travis Kavulla: Yeah. Both of those things can be correct, right? And in the essay I try to separate the industry into two parts, you know, sort of the regulated grid costs that are still cost service price regulated by utility commissions, and then the more commoditized supply of power that works on the basis of marginal cost pricing in economics. And to just spend a moment to talk about each of those, I mean the regulated grid costs, it’s a pretty simple division problem, right? The utility sector typically has a bunch of fixed costs of infrastructure divided by the amount of throughput on that system. And that division problem spits out a quotient, which is the rate that you pay. Right. Now, if you have a lot of headroom in a system and you can fit additional volumes into the denominator without triggering a growth in the fixed cost numerator, then rates will go down. Right. Right. And that’s exactly what happened in North Dakota. In that Lawrence Berkeley lab study, which is really good. But you need the conditions present that allowed that to happen in North Dakota, which had been overbuilt infrastructure from a previous oil boom. And so you could end up fitting a lot of megawatts, a lot of new megawatts under the hood of that earlier vintage capital spending. And meanwhile, on the commodity side for power generation, it is what the market monitor in PJM says, which is that. If demand rises more quickly than supply, then the kind of classic curves of where the supply and demand intercept point occurs is going to shift to the right. It’s going to end up using more expensive supply in order to furnish power to the grid. And in commodity markets, and specifically in electricity markets that are governed by sort of a logic of uniform clearing price, that last megawatt. Needed the most extensive megawatt needed to serve incremental supply is the thing that the price equilibrates around. And that’s the same in gasoline, in eggs, in other commodities. And so until supply catches up with demand, you end up exposing everyone to higher priced supply costs on the commodity side, which gets into the other recommendation of the paper, which is bring your own generation, a requirement. Basically that new loads furnish additional capacity to try to reestablish the equilibrium in the commodity market.

Joshua Rhodes: Yeah, so I want to get to BYOG here in a second. There’s a couple I want to camp out here for just another question or two. You know, one of the strongest official critique of the current TSAs, maybe that Commissioner Judy Chang’s warning that Commands agreements will can still roll these transmission upgrade costs into formula rates for everyone else. If they have these transmission service agreements, like how is it getting snuck in, I guess? We kind of talked about in embedded rates and other types. Is that just what it is? Like how can they still get kind of sucked in here?

Travis Kavulla: Yeah, in I mean, in a very literal way, when a utility in PJM goes to build out new infrastructure to serve data centers or anyone else for that matter, that infrastructure just goes right into its rate base, the costs of it. Yeah. And that rate base, you know, is broadly socialized to the consuming public. And utilities that have adopted TSAs are simply hoping that the incremental revenue pledged, but based on the kind of socialized average rate from a new data center will be adequate to defray the cost and the risk of the incremental capital spending. And the record that Commissioner Chang is sort of referring to out of Commonwealth Edison, Illinois’ large load tariff, contains a lot of data that suggests that half of data center projects probably impose incremental cost requirements that well exceed the revenue that’s pledged out of a TSA. And half of the projects actually are located in places where the utility wouldn’t have to build out as much incremental infrastructure. And there the TSA probably overcollateralizes the obligation from the data center. Whereas a TSA or a prepayment arrangement that was predicated on tying off the incremental cost of capital spending to particular data centers is a more elegant way of dealing with this particular problem.

Joshua Rhodes: You also mentioned Dominion’s Virginia’s Dominion structure. Like, you know, you kind of criticize it for using kind of these embedded cost issues. Like, is there anything different between, you know, Comed and what’s going on in Dominion?

Travis Kavulla: It’s very similar. And the one difference is that Dominion is not a fully restructured utility like Comed is. Comed doesn’t have a role in furnishing supply to customers. So Dominion has taken the TSA idea and tried to extend it to power generation. And basically says, listen, there’s going to be a take or pay arrangement where data centers at sixty percent minimum of their contracted load has to furnish revenue associated with the prevailing generation rate that we charge all of our customers. And that actually is even more problematic than doing it on the transmission side, I think, because candidly it would be easier to tie off incremental capital additions on the supply side to incremental causers of the cost. There’s one other thing that is a real bother about Dominion, and I talk about it extensively in the essay, but I won’t get into it here. Virginia Is an interesting market in the sense that it has some degree of retail competition where large users of the system can select a third-party provider to provide them power. Yeah. In what Dominion has proposed, and the Virginia Commission approved it, any customer taking service in the Dominion service territory would have to pay this toll to them for generation costs. Even if you’re being provided power capacity and supply by a completely different third party. You would also have to pay this VIG to Dominion. So ironically, in the name of customer protection, they’ve actually used it as an opportunity to re-monopolize their system. And they’re basically saying, hey, in the name of consumer protection, we’re going to be the ones to serve every megawatt hour of demand needs related to the data center. Whereas ironically, the easier way to protect consumers would be to insist that third-party suppliers who don’t have a captive ratepayer base. To backstop their business deals instead would be the people serving them. So again, it’s topsy turvy in D in Virginia and Dominion, and I don’t think they’ve landed in a particularly good place, despite many advertisements to the contrary.

Joshua Rhodes: Okay. So that kind of brings us o we’ve been dancing around BYOG, bring your own generation here for a little bit. That’s the fourth thing you kinda bring up in the essay. Tell me what you mean by bring your own generation. I mean I maybe it seems obvious, but tell me what you mean by it and then why is it not possible currently in a lot of places?

Travis Kavulla: Yeah, it really is pretty straightforward. If I’m a one hundred megawatt firm demand in a new data center, I should be furnishing the system with a hundred megawatts of capacity that has energy production roughly matched to my load factor or consumption of energy. And my doing that both promotes investment and generation in a time of uncertainty around generation investment opportunities. And it also is re-establishes that equilibrium we were talking about. And, you know, this ordinarily is one of those things that would not be a good idea. We would expect and want commodities to trade in a homogeneous way, in an undifferentiated way. But I think circumstances have kind of overcome that moment. And you can look to the PJM market, which operates a capacity market, where, you know, the cost of new entry clearly exceeds the price cap. Now in the capacity market. And so the design of that market has made it a foregone conclusion that new capacity is going to have to be added to the system if it is needed through something other than this normal market. Not that the capacity market is particularly normal, but that kind of all signs then point in the direction of bring your own generation, bring your own capacity, call it what you will, in order to solve that problem.

Joshua Rhodes: Again, this is going to be a bit dangerous. I’m always down in, you know, steeped in Texas. And so getting into, you know, more regulated areas is always dangerous for me. But in a monopoly supplied state, what’s the minimum legal reform needed for that? In my head, my brilliant idea was what if we just do a reverse PURPA? Right. Whereas like if you can generators of a certain size or a certain size or below, if they can come in below the utility’s cost of service, like the utility has to buy the power from them. If a third party generator could come in at lower than the cost of the utility, like what if the data center could take that power instead? It’s kind of a reverse concept. Does that even make any sense whatsoever?

Travis Kavulla: Yeah, I’d have to think about that one more, Josh. It’s an interesting idea. And utilities, vertically integrated utilities, should be open to third party procurements, especially if those third parties end up bearing, you know, the development risk and the risk of their projects, you know, being in or out of the money relative to the market. I would say what I’m proposing here is almost more simple, which is in a majority of states today, I as a customer am literally forbidden from buying power generation from a third-party supplier. I have to go through the utility. And that utility will tend to serve me out of sort of an undifferentiated system of power generation assets at these sort of average embedded cost rates. So a way of getting it right for new large loads is not only to expect them to kind of offset the incremental costs of generation to supply them, but require them to do so. And it is probably simplest, or if not simplest, then at least sort of the most attractive on a policy basis to have some kind of modest opening of retail competition in some of those states to facilitate that happening. And there have been a couple of states kind of dipping their toes in that water lately.

Joshua Rhodes: Yeah, and I mean, I want to ask about Utah and and West Virginia. I know there are some states that have they may not have full retail competition for every single rate class, but some have for commercial, not residential, other places like that. And so maybe there are some avenues there. But it does seem like there are some cracks kind of forming in that towards the BYOG. Utah is allowing large loads to contract with large scale generation providers. West Virginia, just next to Virginia, has some micro grid, which is kind of funny to call these data centers microgrids because, you know, they’re the size of some macro grids, here nor there, but some framework that carves out these special service pathways. So are we squeezing the balloon and this is where it’s kind of popping out here and like in other places? What do you think about these approaches for these states?

Travis Kavulla: I think they’re great approaches. And you know, this is a new topic for a lot of these states. So the state laws that we’ve seen written about this are not necessarily the most elegant. They’ll need some redoes and iterations. It’s a step in the right direction. And again, what’s really at stake is do you really want your regulated utility with obligations serving legacy customers and their balance sheet to be used chasing data center load? Or as a first principle. Do you expect data centers to come up with their own deals with third party providers and have the costs and risks of those deals exist wholly within those contractual relationships between two parties that do not have a captive customer base to use as a plaything to chase data center growth? So I think that’s kind of what’s happening. You do see electric cooperatives and smaller utilities. Be pretty positive on this stuff because they don’t want that stuff on their balance sheet, right? They realize that it’s a big risk to try to serve a data center that might be as big as the rest of your customer base, but they don’t want to get in the way and they’re happy to facilitate grid service to those entities. But they are the people who are kind of at the vanguard of expecting BYOG as a requirement just because they don’t want to have much to do with that side of the business, which comes with its own set of risks.

Joshua Rhodes: Yeah, no, totally. I mean, co ops in particular, I think a lot of them are member owned. And so it’s like if you end up in impacting like the rest of the rate base, you probably know them or see them at the corner at the diner or whatever. Okay. I want to bring this back to Texas. This is the Energy Capital Podcast. This is like steeped kind of in Texas. And we’re talking about paying for data centers. We’re talking about allocation of infrastructure. I mean, Texas is also going through this process, right? And I’m very actively going through this process. And by now I’ve had at least one, maybe two podcasts have come out. By the time this one drops on ERCOT’s batch zero process. And so there’s a bit of a difference here. Batch zero is kind of like this large load backlog triage system. You know, while your open season is more of an economic allocation system, they converge on rationing, but they’re diverging on price formation. Should ERCOT be open to doing? I think the batch zero is just trying to fix like a big problem right now that we’ve got. But a batch zero indicates maybe there’s gonna be a batch one, batch two. Batch three, batch next, I don’t know what we’ll call it. Should ERCOT consider doing this open season process instead?

Travis Kavulla: They certainly should. I think a lot of the stakeholder conversations that have existed in the context of the batch zero process almost gesture to why they should. Okay. Because most of the lobbying that I’ve seen around batch zero has been people trying to get into batch zero. Okay. Yeah. By having administrative criteria drawn just so that includes them on the right side of the line to get into the batch and excludes others. And that unfortunately, you know. If only there were a social science that used pricing to try to make those determinations rather than the judgment of RTO bureaucrats about project readiness. I mean, having more of an economic signal for conferring interconnection rights gets RTOs out of the game of deciding who’s more ready than the other people in order to get into the batch. There’s arbitrariness in all of these designs, but Judgments about project readiness are historically have been easy to game and are really difficult to do well in terms of where to draw the line. That being said, I’m very sympathetic with the problem that ERCOT faces and sort of like the other con sort of like the discussion about TSAs, going from a first in line, first and right ad hoc process, gold rush type of thing, to actually trying to group projects into batches. Is a positive incremental reform. I just don’t know if it is meets the moment given what we’re seeing on the grid.

Joshua Rhodes: Yeah, no, I mean it seems like they’re calling it this transitional thing. It’s like, you know, we went from the first come, first serve. There’s some folks that were already in that process. I also like, you know, sympathetic towards like it’s a big, sticky process. There’s trillions of dollars at risk here. I don’t know, whatever we’re valuing all the tokens are or at. And that’s always been the hard thing for me to figure out. So what are the gating criteria? What are the inclusion criteria? And then there’s gating criteria. But my understanding of the way the batch zero process is if you don’t meet The inclusion criteria, you’re not included. And then if you don’t like the allocation you get after the first study, then you’re just kind of dropped out of the system and you forego like 80% of like the cost you paid to get into the whole process. And so it’s kind of like use it or lose it. But I believe one of the concepts you have for open season is that you could sell that right. You could sell that place. Like it’s place in the queue or what it’s a place in the batch, but you could transfer that if you needed to. Is that right?

Travis Kavulla: Yeah, that’s right. I mean, the benefit of an open season, I think, is that it really does confer an actual property right to the use of a public network. Yeah. And in that sense, it’s sort of like a water right or shippers’ rights on natural gas pipelines or the people who buy spectrum out of the FCC’s electromagnetic spectrum auctions. And that’s valuable because ultimately in all technological revolutions, right, we see a lot of also rans like. Everyone and their grandpa wants to be AI and a data center company. And a lot of these people are not going to be around at the end of the day. Even a lot of the big names might not be around at the end of the day. Yeah. And so actually having a property right that in the event of a business model going away, in the event of a default, can be residual value to the estate makes everything a lot more financeable. And that differs that kind of right. Differs from simply having like vague regulatory permission to use power at or about a certain place. And if you go away, that right is sort of recouped to the house, which gets to determine how to reallocate it. So yeah, creating rights and property in exchange for value, it is really not a mind-blowing insight, but it is something that is really not present in the power grid.

Joshua Rhodes: Yeah. No, that it sounds like you worked for a place like R Street or something at some point in your career. I love those guys. I mean, one of the things also that ERCOT is trying to do is like increase flexibility for data centers by offering a couple different approaches. And ERCOT is doing this, they’re kind of borrowing from two approaches they already have. They’ve got this provisional controllable load resource that says, okay, we’ll give you a certain amount of point of interconnection. And if the grid can support it, we’ll let you take more, but from time to time. We’re gonna enforce your point of interconnection. So maybe, you know, your point of interconnection is 200. When the grid’s not stressed, you can go up to 400. But when we tell you to go back to 200, you better go back to 200. And that’s controlling the load. And then there’s also kind of this, they’re calling it the BYOG SFL, like self-limiting feature. I don’t know if the acronyms are killing me here. There’s two approaches that basically enforce like a point of interconnection limit. And then there’s a couple different ways of dealing with it. One, you can have your own generation. Behind there to keep your computers running or keep your GPUs running, or, you know, you can turn them down. You talked about flexibility also in your essay. Is ERCOT ahead on these processes? Like would these fit well elsewhere too?

Travis Kavulla: Yeah, they would. And a real credit to ERCOT for considering these. And we’ve seen similar developments in how FERC resolved a dispute between Amazon and Talen for the co-location of resources, though the policy implications are broader than that, as well as in the Southwest Power Pool. But yeah, the bottom line is if you are not demanding firm service from the grid. Then what that really means is that you are not necessarily imposing the same incremental capital expenditure costs on the grid. You’re especially not doing that, you know, if you’re bringing adequate generation to kind of net out your impact onto the grid at a particular location. Or for that matter, and here I put in a plug for base, you know, paying for other people’s flexibility at or around that node or elsewhere on the system to furnish flexibility or capacity. So All of doing any of those things should almost get you like to the very front of the line. If there’s going to be a batch zero, maybe it’s batch minus one. But people in that batch really should be the people who are not requiring long lead time incremental capital expenditures. And really the regulators and RTOs looking at this issue deserve some credit for thinking about that as a speed-to-power solution. Yeah. And also to the degree it is tied back to the purchase of. Battery aggregations or bringing on a new gas power plant, even in a restructured market, conditioning the get for speed to power interconnection on the give of bringing new generation capacity that usually would not be considered in tandem with a new large load for the purposes of grid interconnection is a really positive step. And if people read the whole essay, they can kind of see in the economic logic of the open season how I’ve tried to make sure that. People who are flexible get good treatment under these systems, even if the system is intended to resolve in favor of incremental capital expenditures and the assignment to those costs.

Joshua Rhodes: Yeah, no, mean I think flexibility is key. You know, the more flexible things are, the more megawatts that can get onto the system. They’ll be operating more of the time, right? Which can help the utilization factors and other stuff of infrastructure to try to drive that stuff down. You brought up the Amazon Talen thing. I was gonna say a couple of years ago, I was also in a room with some big generator developers who were talking about, you know, a data center wanting to take, you know, one of their units off the system and they were getting pushback. Regulators and they’re like, well maybe we’ll do it flexibly, but the data center doesn’t want to be flexible. And I was like, well does the data center have to be flexible? Can you get that flexibility elsewhere? Does it maybe close by? So I was going to say, does that BYOG have to be on site or that can that

Travis Kavulla: Yeah, it should not have to be on site. In fact, you know, I understand why the co-location debate has occurred, but it’s weird to expect these customers who have the highest value of loss load of seemingly any customer on the planet to be the ones from whom we’re sourcing power interruptions. Like that doesn’t make a lot of sense. There should once again be a market for this and they should be able to pay other people to be flexible on their behalf. And the same thing goes for co-located generators. Typically, there are a lot of good reasons why, you know, you would not want your load tied into all of the weird outages and contingencies that happen in power generators, which are finicky machines. And so being able to kind of broadly source your supply, even while expecting to pay incremental cost associated with new sources of that flexibility or capacity. You know, is kind of a North Star that we need to live by.

Joshua Rhodes: Yeah, I mean, I think that’s one of the great things about the way that base does things too is like, okay, if you’re going to interconnect a lot of things and you can move a lot of little things to look like a big thing, that can be useful. And you’ve already basically got the interconnection at every single house or commercial or business, whatever it is, you know, maybe we can move faster there. And you’re doing great things in the ADER and the ERCOT programs. So just kind of tie this back, like, you know, Texas is trying to build this plane even as it’s well taken off. Like in the state is you were building an ecosystem of around, you know, speed to power, but it’s conditioned on, you know, financial commitment and emergency flexibility. Is that necessary but not in your eyes necessary but not sufficient without that price discovery?

Travis Kavulla: It’s a good way of putting it, Josh. I mean, so far, we’ve probably tried to extend really beyond the breaking point some of the traditional mores of cost of service regulation and this belief that a utility in the current moment can plausibly serve all of the possible loads that are supposedly coming onto the system. And if we really want to get serious about speed to power and efficiently integrating. Compute onto the system, there does need to be some price discovery for those loads that are kind of the highest value, the most flexible, the ones that are able to be built and served by the grid in the fastest possible timeline. And that’s fundamentally what some of the ideas that I propose are in search of.

Joshua Rhodes: All so if you had to choose one of your ideas, I think it would probably be an open season. But like if you had to take one of your ideas, what is the most likely one that would act that could get some traction? Just given the regulatory landscape that exists, like who’s your second favorite child here?

Travis Kavulla: Yeah. I mean, I’m just hard pressed to say that if I’m a data center and I get quoted a number that reflects all of the incremental costs of serving me on the system. And the data center says, That’s great. Let me write you a check right now. And by the way, I’ll also pay the existing rates to offset the costs of everything else you’ve invested in over the years. Yeah. The answer should be a thousand times yes. I don’t see it despite the utility business model being devoted to spending capital on behalf of their customers, it solves instantly the problem of cost shifting and speed to power in a way that should be really attractive. And the downside really is shouldn’t we take a beat and try to plan these systems more efficiently so that the joint costs can support more joint uses? But again, in a small system for a one off project that generally should be an ECS, but I’m holding out my help for open season.

Joshua Rhodes: Yeah, I mean shut up and take my money, like you know, type situation, which yeah, I think a lot. Travis Kavulla, thank you for coming on the Energy Capital Podcast.

Travis Kavulla: Thank you, Josh.

Joshua Rhodes: Thanks for listening to the Energy Capital Podcast. If today’s conversation helped you make better sense of how the energy system actually works, share the episode with a colleague and hit follow on your podcast app. You can find us on Apple Podcasts, Spotify, and all the usual platforms. For deeper analysis and context each week, subscribe to Texas Energy and Power at texasenergyandpower.com. That’s where you’ll find every episode, every article, and our latest updates. We’re also on LinkedIn, X, and YouTube. Where we share clips, insights, and ongoing commentary on energy policy, markets, and the grid. Before we go, a quick note. The views expressed on this podcast are my own and do not represent the official positions of the University of Texas, IdeaSmiths, Austin Energy, or Columbia University. A big thanks to Nate Peavey, our producer. I’m Joshua Rhodes. Thanks for listening, and we’ll see you next time.