How PJM Is Grappling With Data Center Power Demand

PJM Interconnection, the nation’s largest grid operator, is preparing to file a wide-ranging proposal with federal regulators aimed at managing the rapid growth of electricity demand, including AI-driven data centers. The plan stands out as one of the first comprehensive efforts by a grid operator to address surging load from new technologies while maintaining system reliability and limiting cost impacts on consumers.

The proposal arrives at a moment when the electric grid is under growing stress. Tightening power supply-demand balances, high-profile grid failures, and a series of narrowly avoided outages have raised concerns about whether the power system can continue to meet demand reliably. At the same time, those pressures have increasingly shown up in electricity prices, which have increased sharply in many areas.

PJM’s proposal tries to answer a question grid operators across the country are now facing: how to say “yes” to large new loads without turning reliability into a gamble or costs into an afterthought. The plan lays out a structured approach to integrating data centers and other large loads, with an eye toward keeping commitments realistic and aligning responsibility with impact.

Abe Silverman is an assistant research scholar with the Ralph O’Connor Sustainable Energy Institute at Johns Hopkins University and a former general counsel to the New Jersey Board of Public Utilities. Tom Rutigliano is senior advocate for climate and energy at the Natural Resources Defense Council, where his work focuses on PJM.

Full Transcript

Andy Stone: Welcome to the Energy Policy Now podcast from the Kleinman Center for Energy Policy at the University of Pennsylvania. I’m Andy Stone. Over the past several years, the nation’s electric grid has increasingly found itself in the headlines as a system long taken for granted has been pushed closer to its limits. This recent focus on the grid has been driven in part by high-profile grid failures and a series of narrowly avoided outages, and has increasingly raised questions around the future reliability of the electricity system. This grid stress, along with factors including inflation and fuel costs, has shown up in electricity prices, which have noticeably trended upwards since the COVID pandemic.

And into this already operationally and economically strained system has come the AI boom, bringing a rapid proliferation of new data centers, many capable of consuming as much electricity as a small city. Grid operators are now grappling with how to accommodate this surge in demand while keeping the system reliable and affordable and while accounting for the grid’s significant environmental footprint.

On today’s podcast, we’re going to take a look at a plan outlined in January by the nation’s largest grid operator, PJM Interconnection, to address the challenge of integrating large new sources of demand, including data centers, onto its system. The announcement is notable not only because of PJM’s size and importance, but because it is a particularly comprehensive effort by a grid operator to confront the intertwined challenges of reliability, affordability, and the rapid growth of new technologies.

In today’s discussion, we’ll walk through the elements of the solution that PJM is expected to file with federal regulators, try to bring its complexities down to earth and explore what the plan could ultimately mean for consumers. Joining me are two experts with deep experience in PJM and the policy debates surrounding rising electricity demand. Abe Silverman is an assistant research scholar with the Ralph O’Connor Sustainable Energy Institute at Johns Hopkins University, and former general counsel for the New Jersey Board of Public Utilities. Tom Rutigliano is Senior Advocate for climate and energy at the Natural Resources Defense Council, where his work focuses on PJM. Both participated in the policy discussions leading up to PJM announcement of its proposed solution last month. Abe and Tom, welcome to the podcast.

Abe Silverman: Oh, it’s great to be here. Thank you.

Tom Rutigliano: Good morning, Andy.

Stone: We’re going to be looking today at PJM plan to accommodate rapidly growing electricity demand, most notably from AI data centers. But before we get into the details of the plan, I’d like for you to start us out— and I’m going to ask you, Tom, to do this— by giving us a quick introduction to the range of challenges that these new large loads present. And I wonder if you could explain how they present a challenge that the market really wasn’t designed to address.

Rutigliano: Yeah. This is really happening at just about the worst time it could have at PJM. If you look back as recently as 2020, PJM had plenty of surplus power, almost an embarrassing amount of surplus power. Some people accuse them of keeping too many old power plants running. And things have really gone downhill in the last five years. Or, let’s say, tightened up, not downhill, even without the data centers.

Starting around 2020, PJM’s interconnection queue— the process of connecting new power plants to the system— seized up. And the rate that new power plants can come online has slowed to a trickle. And then at the same time, we saw gas really displacing coal, which in a lot of ways is a good thing. The gas plants are more economical and cleaner. But what we didn’t realize is that we were running up some tech debt, because gas plants are not as reliable as we hoped in the winter. And in around 2023, PJM finally did the accounting for that, and realized that their system didn’t have the surplus it needed. So we’re in a place now where PJM is suddenly seeing this huge new demand for electricity, right when their supplies were pretty tight to begin with.

Stone: So what makes these new large loads like AI data centers fundamentally different from the other sources of demand and demand growth that have been typical in the past, Tom?

Rutigliano: There, it’s size and speed. You know, as you mentioned, a data center can use as much power as a large city. We’re seeing 100 megawatt up to gigawatt loads, which are just huge. And that they come on so fast. You know, in a way, we’re seeing one of the country’s fastest-moving industries colliding with one of its slowest-moving industries. The electricity sector is used to predicting things five, seven, ten years out, and planning accordingly. You can’t build power plants in a year or two.

And then on top of that, you’re seeing these huge demands for electricity, really, for the benefit of a single corporation. You know, electricity load growth, it’s generally considered to benefit everyone about equally. This year you buy an air conditioner next year, I buy a water heater and so on. And so it makes sense. We’ve built market structures to spread those costs out to the whole public. But now, when you have new power plants being built basically for the benefit of a single company, a lot of the market structures break down, and we all have to think more about both, how do we get these things built much faster than we’re used to? And how do we figure out who pays that’s fair?

Stone: So you’re getting, these new data centers bring a lot of demand, a lot of strain on the system, and the question is who should be paying for, I guess, the upgrades to the system to manage this new demand that really is concentrated in one company, say?

Rutigliano: Yeah, exactly.

Stone: So Abe, I want to turn to you. Concern around how to address these new large loads has been growing for several years, and I guess it really took off— I’m kind of guessing here, you know— when ChatGPT was introduced in the fall of 2022. But it feels like the urgency around the issue, around dealing with these new large loads, really spiked— at least in the case of PJM— when new demand, or forecasts of this new demand, led to dramatic increases in the capacity costs in the market. And we’re talking about billions of dollars of new capacity costs coming out of PJM’s most recent capacity auctions. And much of that will ultimately be paid by consumers. Can you briefly explain, what is a capacity auction? And then tell us what has happened with the prices in these auctions.

Silverman: Oh, I see, you’re giving me the easy questions. All right. [Laughter]

Stone: The easy stuff.

Silverman: The capacity market is this tool that PJM has designed that’s supposed to look into the future and send price signals to make sure that we have the right amount of power plants in the right places on the grid, three years in the future. And the idea has always been, “Hey, this is great. We’re going to look into the future a little bit. And if we need to build a new power plant, say, in a load pocket, like the Baltimore area, we can come in. We’ll send a price signal that’s on a locational basis, and that price signal will drive investors to put money into building a new power plant. That was the theory.

And really, what’s happened over the last couple of years— as Tom mentioned, we’ve gone from this very substantial surplus where we had too much generation in the PJM market, and these capacity auction prices were extremely low. Then over the last couple of years, as we’ve started adding in more and more of this load, the price has shot up. And not, not just by a little bit. So we often talk about what the cost of capacity is to the entire market. For 2023 and 2024, the total capacity bill for all consumers in PJM was about $2.2 billion. Still a lot of money. But then in one year, that number jumped from $2 billion to $14 billion and change.

And, you know, you just stop there and think about that for a second. That is an enormous increase. And a lot of customers throughout the PJM region all of a sudden saw this in their electric bills. Many of you are probably in the same boat that I am, where my bill went up 20% in one year. And that’s really unprecedented.

And then what’s happened since then? So we had this first sort of jolt to the system, starting in June first, 2025. And the problem is that we’re still adding even more demand to the system, and prices are continuing to go up. And without some sort of regulatory intervention, we’re going to double our capacity bills again. And just to put some numbers on what Tom was talking about in terms of the scale of the problem—  you know, the peak load in the Baltimore Gas and Electric area is around 6.5 gigawatts.

Stone: And Baltimore Gas Electric as in PJM, is that right?

Silverman: Yes, exactly, in Maryland. And the entirety of the city of Baltimore uses about six and a half gigawatts of electricity on a peak day. That is the amount of new data center load we’re adding to the grid every year between now and 2030. So you start thinking about, what does that mean? That means that we need something like five new nuclear power plants added to the grid every year just to keep even. And, of course, for all the reasons Tom talked about and that we’re going to get into, the system is nowhere ready to send that kind of price signal, to drive that type of investment. And in the meantime, consumers are really seeing these higher bills, largely, frankly, associated with the new data center load.

Stone: Yeah. I think the market monitor in PJM has said that the bulk of these new capacity costs are based upon the expected new AI data center load. Is that right?

Silverman: Absolutely. And you know, the number that just just sort of stunned me— the three- year total bill for capacity across all the 65 million consumers in the PJM region was about $40 billion. And the IMM came in and said about half of that was attributable to data centers. So just an enormous burden on Mom and Pop consumers.

And I think this sort of gets to the fundamental challenge— frankly, for all of us, but particularly for state and federal policymakers—how do you protect Grandma from these bill increases driven by the data centers? And then how do you ensure that the data centers are ready to step up and pay their own way, in a way that’s also consistent with the economic growth and tax policy that a lot of states have, where they’re very interested in attracting data centers? So there’s this really knife-edge balance between driving the kind of economic growth that a lot of our state policy makers want, and protecting existing consumers from having to subsidize big tech’s electricity bill.

Stone: Well, you just mentioned the states in this, right? And so there has been a broad political response, in particular to these recent capacity price spikes. And we’re seeing increasing attention from the governors of states that are in the PJM market. And we’re even seeing some alignment between the White House and blue state governors, again, that are in PJM, over electricity prices. So Abe, can you tell us a little bit about this political response?

Silverman: Yeah, this has been fascinating. Two weeks ago, we started the week with President Trump, of all people, tweeting out that that data center shouldn’t increase the electric bills for all Americans. Which, I frankly, did not expect to see. And then the next day, Microsoft came in and made a commitment, saying, “Hey, we don’t think our data centers should drive bill increases for everyday Americans either.” And then later on in the week, you had every governor in PJM coming together— which, to the best of my knowledge, is the first time this has ever happened. You had all the governors coming together with a set of principles, key among them that retail consumers, existing retail consumers, should be insulated or ring fenced from the financial impacts of new data center load. And then later in the same day— you know, this is all in the place of a week— you had the PJM Board of Managers. I know we’re going to talk about what exactly that means, but we have this group that runs PJM coming out and putting forward their own plan, and talking about, how do we protect reliability and affordability for consumers in this very complicated investment climate and very complicated reliability picture?

Stone: Well, you mentioned, obviously, the White House came out concerned about consumer costs. The governors in PJM. And I just want to point out, PJM is an extremely diverse market, and that’s one of the challenges in running the market. I mean, on one end of the political spectrum, you’ve got West Virginia, and the other end, you’ve got New Jersey. And they’re all in the same market, and they’re all aligned on this right?

Silverman: Absolutely. And I want to, really take a step back and talk about the bipartisan nature of the concern here. I mean, it was kind of amazing. There was this big White House press conference where you had several governors, including Governor Shapiro, up there on a stage with Doug Burgum and Chris Wright, the interior secretary, and the energy secretary, respectively, all signing on to the same set of principles. Talking about how PJM needs to take immediate action to protect existing retail customers against these cost increases. I mean, that is crazy. I mean, I can’t think of the last time that you had that kind of bipartisan response to the issue. And I’ll say there’s a wonderful article— we’re recording this on February 2— in The Wall Street Journal today talking about how rural America is falling out of love with data centers.

And so, you know, I think we’re seeing this kind of fascinating both cultural and financial shift. As well as from the business side. You know, with people like Microsoft and Equinix, a large data center operator, and now Google coming in and talking about their willingness to pay their own way. Now, of course, exactly what everybody thinks is their fair share— sagas will be written about that. And there’ll be a lot of implementation details there. But it’s fascinating to watch the zeitgeist shift from a “data centers at any cost,” to “data centers, so long as we’re not raising affordability concerns for regular Americans.” And that shift— I mean, you know, you get a bit of whiplash. That has all happened in the last six months.

Stone: A lot is going on. And against the backdrop that we’ve been discussing in the six-month timeframe that you just brought up there, Abe, I want to point out that in early fall— I think August or so of 2025— PJM initiated a process in which it engaged the stakeholders in its market, and that includes power plant and transmission line owners and consumer advocates and many others, to come up with a plan to accommodate all of this new data center demand, while keeping the grid reliable, the lights on, and keeping an eye on costs. And that process, which was called “the critical issue fast path,” or CIFP process, led to 12 different proposals from the stakeholders, none of which were approved by the stakeholders in a vote that was held this last November, two and a half months ago. As a result, PJM’s board of directors has put forth its own plan and did that in a letter on January the 16th. So I want to ask you, Abe, what does this outcome tell us about the nature of the problem that PJM is trying to solve, which you’ve already started to get into, and about the limits of stakeholders in the market to actually solve the problem?

Silverman: Yeah, yeah. So I’ll even go a little bit back in time, you know, to— gosh, almost a year ago— when the first capacity market clear came out. And everybody sort of had an “oh heck” moment. Because we were all of a sudden in an affordability crisis, almost immediately.

Stone: That’s when it spiked, right? The capacity prices spiked.

Silverman: When it first spiked, right. Exactly. And so we had politicians and state regulators coming out and seeing these higher prices, and really, frankly, getting very concerned. I’m here in New Jersey. We had Governor Sherrill, who got elected largely on the basis of an affordability platform. The same thing in Virginia with Governor Spanberger. And so you had this incredible focus on PJM for the first time in a long time, as these prices spiked.

PJM is run by two groups. They have a CEO and staff, and then they have what they call their board of managers. Think about it as just a normal corporate board. So you’re right. In August 2025, which feels like a lifetime ago, but is actually six months, the PJM board initiated this critical issue fast path, or as we lovingly refer to it, the SIFP, and basically put out a 10-page paper explaining the concerns that they had. And chief among them, actually, for the PJM board of managers wasn’t so much affordability, though that’s a byproduct of all this. But they were concerned about reliability. Because they were looking at the numbers and saying, “Oh my goodness. We are not bringing on enough new supply to serve the city of Baltimore every year between now and 2030, which is what you know our projections are suggesting.” So they were very concerned that reliability was going to suffer.

And so the PJM board and PJM staff then came out with a series of proposals. And this is, I think, what we’re going to talk about for the rest of the time today, where they laid out a bunch of different criteria and said, “Okay, stakeholders let us know what you think.” And boy, did stakeholders let them know what they thought. So the original PJM proposal required data centers to turn off when the grid got stressed. And the data centers and the generators did not like that. And PJM staff—  so the part of the house that’s run by the CEO— immediately retreated. And then Tom and I and a bunch of other people spent the next several months sitting down there in Norristown, arguing with other stakeholders and pointing out, “Hey, listen. This current system that we have set up is not sustainable. Data centers are going to have to either bring their own generation or agree to be flexible, or we’re going to have to come up with some other mechanism.”

And so PJM— you’re right. They eventually took all the different proposals to a vote. And I have to say, I don’t think any of us were super surprised that none of the proposals got a majority vote. That’s pretty typical for a complicated issue like this, where the generators have their financial incentives, and the data centers have their financial incentives, and so does everybody else.

So this sort of landed back in the laps of the PJM board of managers, which I think is where we all thought it would go, in early December. And originally they said, “Okay, we’re going to have a decision for you and tell you what we plan to do by the end of December.” And then that slipped into January. And as you mentioned, they came out with this ten, 15-page memo explaining how they were going to propose to move forward over the next couple of months to rein in the costs and really address the reliability issues. And now we are back where we started six months ago, back in the same rooms, with the same cast of characters, talking and going through all these options and trying to develop proposals. And Tom has been very involved in that.

And so again, we’re going to be making our case to the PJM Board of Managers. Us and everybody else in the energy universe. And that they will then make a decision, hopefully in the next couple of months. At which point they then have to file it at FERC, and we get to do it all over again before the Federal Energy Regulatory Commission. So it’s a very complicated process. But, I think that to me, the biggest through line is just how positions have evolved over the last couple of months. That we really have gone from this—data centers just saying, “No, no, no, we like the current system,” to I think everybody, including the PJM board of managers, recognizing that the status quo simply is not acceptable, politically, or morally. And again, we see that reflected in this unprecedented level of bipartisan interest in the issue, from the White House and from the governors.

Stone: Well, as you mentioned a few moments ago, both you and Tom were involved in some of the proposals that were submitted and voted on by PJM stakeholders in November. And I want to hear what both of your priorities are in just a moment. But before we go to that, I want to look at this letter that you just talked about, the letter that came out on January the 16th, which was that actual 14-page— I looked at it— 14-page directional letter from PJM’s board. And just to reiterate what’s going on here, PJM’s board came out with this guidance letter, a directional letter. It says, “Guys.” These are PJM staff. “These are the elements that we want in this solution. Now you go ahead and work out the details, and then you’re going to file that proposed solution to this whole AI data center load issue with the FERC.”

Okay, so let’s go ahead and take— Tom, I want to jump to you here now. I wonder if you could give us kind of an executive summary of what’s in that board letter, the direction that it gives to PJM. Because we’re going to dive into the details of this, and I think listeners probably need a bit of a heads up on what we’re going to see here. Could you give us that first high-level look at what’s there?

Rutigliano: Yeah, absolutely. The board took some pretty broad stroke actions here. The first thing they do is they put the data center developers on notice that at least in the short term, PJM is not going to be in the business of building power plants for them. And it makes that explicit, and it explains how PJM is going to maintain reliability if there’s more demand than supply, if there’s not enough electricity to go around.

And then, looking a little further into the future, it sets up a couple of different things to look at where the money for power plants comes from, and how can we make sure that we’ve got the kind of investable environment that lets banks finance power plants and lets developers build them? Of course, there’s a lot of details. It’s a 14-page memo. They also fixed the load forecast to make sure we’re not buying stuff we don’t need. But the big thing comes down to those two things. For the moment, PJM is not building power plants for the data centers, and let’s figure out how to get those things paid for.

Stone: Okay. So, we’re going to get into the details in that letter, the direction that PJM’s board gives to PJM staff. But before that, I also mentioned that both of you were involved in the process of coming up with some proposals to deal with all this. And Tom, let’s start with you. I want to ask, what were your top priorities going into that CIFP, the critical issue fast path process? And do they align or not with what you’re seeing in the board’s direction?

Rutigliano: Yeah. Well, Natural Resources Defense Council, we’re an environmental organization. And we were going into that to make sure that this new demand from the data centers doesn’t roll back the state environmental policies that we’ve seen. You’ve got a number of states in PJM that have very ambitious decarbonization goals. We wanted to make sure that those could stay on track even as things move forward with the data centers. And more broadly, we make sure this isn’t a carbon disaster. That we don’t see a lot of investment in fossil fuel plants that might put the region on the wrong energy trajectory for the next— easily could be 30 years.

Stone: Abe, what were your priorities for this?

Silverman: My priorities were making sure that we actually had an investable system, where the PJM market could send the appropriate price signals and operate in an economically efficient manner, while also— and I think this is really key— protecting customers from the higher prices associated with the data center load growth. And absolutely— I mean, I think what Tom talked about is really important. I came at it from a bit more of an economics perspective than a clean perspective.

But at the same time, it’s also really important that we design a system of investment that allows states to signal their own priorities in driving investment in this next generation of generation resources. I mean, it’s important to recognize we are talking about— take your consultant report of choice, but anywhere from one to $4 trillion of investment in our electric grid over the next decade. I mean, that’s just an amazing amount of money. We are basically talking, in some utilities, about doubling the amount of electricity that they’re carrying on their lines because of the data center load growth.

So we need to get this right. We both need to make sure that these costs are being allocated appropriately, and that existing retail customers are protected from increases, but also that for regions that have policies promoting clean energy, we need to make sure that that $4 trillion is being directed in a manner that’s consistent with state law and everything else. And that we don’t just sort of reflexively lock in, as Tom mentioned, an entire generation of natural gas power plants and the emissions that go along with that.

Because, remember, a power plant built today is going to continue operating— my daughter’s eight. She’ll be mid-career by the time those power plants retire. You know, we’re just making a huge decision about climate and emissions, as we think about where this $4 trillion of money is going to get spent. And I’m a realist, right? There will be some amount of new natural gas investment. But we also need to make sure that we’re leaving room and utilizing that capital in a way that’s consistent with laws in places like Illinois and New Jersey and Maryland that require some level of clean. So how do we do that in an economically efficient manner? How do we send those price signals? How do we bring on the new load in a responsible manner?

Rutigliano: One thing— I guess I’ll call it a nice thing that worked out about this process, is that the environmental and affordability concerns pulled in very much the same direction. You know, NRDC, we ended up jointly putting out a proposal with over 50 state legislators who were overwhelmingly driven by keeping power affordable for their rate payers. Both for the sake of the environment and keeping bills down, we agreed that these data centers need to be out of PJM’s market and sort of in their own playground. Because what we saw is that if you keep the data centers in PJM’s market, first, it drives prices up to the roof and keeps them there for a very long time. And it most likely ends up in a bunch of new fossil fuel plants getting built in the western Pennsylvania, Ohio area.

Stone: Tom, when you talk about keeping them in or out of the market, can you explain a little bit more clearly what you mean by that?

Rutigliano: Yeah, of course. So the capacity market, which Abe mentioned earlier, is how PJM lines up enough power plants that they make sure they have sufficient supply all the time. Right now, that’s one market for the whole region. And the data centers, along with everyone else, are in there. And so as long as the data centers are driving up demand in the capacity market, that raises the price of capacity for everybody. Instead, you want some structure where lining up the electricity for the general public is done in one space, and then figuring out how to get new power plants for the data centers is done in another space, without them affecting each other’s prices as much.

Stone: I want to hit on one other topic here that just came to mind, before we actually go through, step by step, what’s in the proposal here. But you know, you’re mentioning these costs, and these costs falling on consumers. And I also want to point out here that data center demand, where data centers have been cited, is actually quite local. There’s certain areas within PJM and within the country broadly that are seeing most of the influx of these new data centers. But that’s starting to expand into some rural areas where these data centers really do represent outsized new demand that really is exponentially more of a challenge. Abe, I wonder if you might just kind of fill in what that means.

Silverman: Andy, I’d say there are outsized demand everywhere, to start with. A gigawatt data center is huge regardless of where it’s located. But, you know, it has been really interesting to see where data centers are locating. There are primary data center markets. And the biggest of those, frankly, in the world, is Northern Virginia. Data Center Alley, the OG of data center load growth.

But we’re also seeing that now expand into other markets. Columbus, Ohio, and outside of Philadelphia, are both huge areas where data centers really want to locate. And so what we’re seeing is there are utilities, like PPL Utilities, which are west of Philadelphia, where they are now anticipating that they will see 100% increase in the amount of electricity they’re going to sell to customers, almost entirely attributable to data centers. Columbus, Ohio, the AEP area, same thing. Indiana is seeing a massive surge. Illinois, ComEd, the area around Chicago, is also seeing massive increases.

So we have these data centers, frankly, springing up everywhere. And they tend to be— outside of that, of that Northern Virginia cluster, which I think has to do with the federal government and sort of where a lot of data center investments were first made— they tend to go to places where there’s relatively cheap land and where the transmission infrastructure is really strong. So it’s actually not like they’re locating in cities so much. They are going directly to rural areas.

And I think I mentioned it earlier, but maybe it’s a good time to reiterate, which is, there is a lot of rural communities who are stepping up now and saying, “We appreciate the economic development, we appreciate the tax revenue. Those are both important things. But we also are very concerned about the way of life, the rural way of life that we enjoy. And we’re not entirely sure that these data centers are consistent with that.” And that encompasses both water issues. It encompasses the electricity price increases we’re talking about. And it’s also that data centers are ugly. And, you know, I think a lot of communities are looking at this and saying, “I don’t know if we want 1000 acres of new low slung warehouses that are totally nondescript.”

So it’s this interesting political coalition that’s emerging, where there’s a lot of people who I think are culturally, inherently suspect of data centers and AI, but are also extremely concerned about the local land use and the affordability issues and the way of life issues associated with with data center growth. And so I think that’s also driving this kind of weird political coalition where, you know, even people like Governor DeSantis, down in Florida, are looking at this and saying, “Hey, we want to be able to have state sovereignty over over data center locations. And we don’t necessarily want Washington telling us we have to bend over backwards to to facilitate new data centers.”

Stone: So, drum roll please. We are now, after all this context setting, looking what’s going on, actually going to look at the elements that are in this board letter. So let’s walk through pieces of the framework. One of the central themes in the board’s letter is uncertainty. Tom, I’m gonna direct this one to you. And that uncertainty is particularly around how much AI-driven electricity demand is gonna actually materialize. And I wanna note here that the last episode of this podcast took a deep dive into the problem of forecasting future demand. So if anybody wants to take a deep dive into that, please listen to the last episode. But for the sake of our conversation today, I want to go over this again and ask you, Tom, what changes is the board suggesting around the process for forecasting these new launch loads, and why is that important?

Rutigliano: The board told PJM staff to come up with a much more rigorous way of forecasting how many data centers they think will be built, or how much electricity they’re going to use. You can think of it as being about risk, right? If a data center project ends up not coming true, who’s left footing the bill? And if PJM forecasts that there’s going to be a huge demand for electricity and enters into contracts with power plants to supply it, the public’s left holding that bill, whether the data center shows up or not.

So you know, I think we’d said about this three year period, looking at $47 billion in capacity bills for data centers, right? Something worse than that would be a $47 billion capacity bill for a bunch of empty lots where people said they were going to build data centers. So the board put some discipline on that, in terms of how much earnest money a data center has to put up, things like that, to convince their utility that they’re going to be real.

But we do feel they didn’t go far enough. Our position is that if PJM is forecasting data centers at all, we’ve already lost. Because that implies some transfer of risk to the public. The only entity that should be taking on the risk if a data center gets built or not is the data center developer. And so our feeling is that, again, they should be bringing their own generation, not in any sense offloading that burden to PJM and everyone else.

Stone: One of the most visible and debated elements of the board’s proposal is this idea of bringing your own generation, or BYOG, which you just noted. So what is BYOG, and why might it be an attractive solution, and to whom?

Rutigliano: BYOG is something that we advocated pretty robustly for. It’s exactly what it says. It’s that PJM isn’t going to line up the power supply to make sure the data center can have 24/7 power. You know, the data centers can connect to the grid, and they can buy the power on an as- available basis. But— and this is a critical part of it— when power starts to get short, the data centers are told that they’re the ones who need to stop using power from the grid. And that the way a data center can get guaranteed grid power is by adding enough new generation to pull their own weight. So that’s the Bring Your Own Generation capacity. Basically, they’re welcome to buy the surplus, but if they want to guarantee it, they’ve got to create new supply.

Stone: So these new loads, like the data centers that would bring their own generation, would also, under the Board’s guidance, be able to connect more quickly to the grid. And that’s critically important to the tech companies, because they need to scale up rapidly. So tell us about what they call this expedited interconnection track for these loads.

Rutigliano: Yeah. This is a problematic area. So, you know, I think I mentioned at the top that PJM’s interconnection queue has been in bad shape, and there’s a real long delay getting new power plants even ready to start construction. So what this EIT does is that power plants that show up paired with a data center— you know, the data center says, “Okay, here’s the power point I want to build to supply myself.” They, one way or another, cut to the head of the line on the interconnection queue. Which is something that’s sort of tempting, but really, it’s not going to solve the problem, and the board needs to do better than that.

PJM has already tried a fast track once. They had one that was executed at the beginning of 2025 that they call it “the RRI.” And what we’re seeing is that the problems with interconnection go way beyond the queue. The administrative problems with the queue are only the surface. And that half the projects that are coming out of the queue are being told they need to wait four to seven years for major transmission system upgrades.

Stone: And that queue is the queue that you wait to get actually connected to the PJM system. Approved to connect, and then connect, right?

Rutigliano: Not even connected. Just approved to connect, and told how much it’s going to cost. So it’s the engineering studies to say, “Can the grid support you?” Or, “What upgrades do we need to support you?” And then at the end of that process, that’s when you simply get a bill and a timeline, and the developer has about a month to decide if they want to move forward the project or not.

So what this EIT is, is a idea of skipping the queue, so it potentially shaves maybe 12 months off the project development time. But what we’re seeing is that even fast track projects are dropping out at the same rate as everyone else. Because we’ve been neglecting the transmission system for decades, and there really is no quick fix to that. So the EIT politically looks good, but I’m afraid that it’s just driving faster into the same brick wall.

Stone: You mentioned a few minutes ago, you said the word “curtail” or “curtailment.” And there’s another key element of the board’s direction that’s known as connect and manage. And this concept of curtailment is elemental to that. And I’m sorry, this all gets very complicated, but that’s the whole point here. We’re trying to kind of make sense of this. Can you tell us, what is connect and manage? What is the relationship to curtailment? What is curtailment? And how might they also be used to accelerate the pace at which these new large loads, such as, again, AI data centers, can connect?

Rutigliano: Connect and manage is actually an idea that PJM is taking from down in Texas, where I’m sure a lot of your listeners know, they have tremendous amounts of wind right? Texas is the nation’s leader in wind power. And they were looking at similar problems to PJM, that they can’t upgrade the grid fast enough to support all this wind. But what they did is, they came up with this “connect and manage” option, where they tell the power product developer, “Okay, build your wind turbine. Connect it to the grid. And whenever the grid can’t take the electricity from you, we’re just going to tell you to shut down.” And that’s worked really well. It’s a big component of why Texas has been able to build so much new power so quickly.

So PJM took that idea and said, “Let’s apply it to the data centers.” To the demand side of the equation. Which says, “Okay, data centers. You can connect to the grid. But when there’s not enough power to go around, we’re going to—” with an asterisk here— “we’re going to tell you that you have to shut down. Because the grid can’t handle everyone, and you’re the ones who we’re going to do connect and manage.”

Stone: That’s the curtailment, right?

Rutigliano: That’s what curtailment is. Right. Curtailment is cutting off power supply. Now the little asterisk is just this jurisdictional quirk that, because of the way we manage electricity in the United States, and the state versus federal power balance, PJM can’t actually tell a data center to shut off. But they can tell the data center’s utility, “Hey, you’ve got to curtail the amount of power you consume by some amount.” And then it’s up to state regulators to decide, do we turn off the data center? Do we turn off the public? But ultimately, yeah. Curtailing is just saying, “We’re going to reduce demand for electricity by command when there’s not enough to go around.”

Stone: And that sounds great from a grid management standpoint, if the data center requires more electricity than the grid can supply to it in a period of peak demand, when the grid is stressed. Having the data center back off on the amount of electricity it needs could maintain the stability and the reliability of the grid. But from the data center standpoint, that’s potentially a bit tricky, because depending on what kind of data center it is, they might need to be running all the time. They might not be to run their business if it’s turned off at certain times. Is that right?

Rutigliano: Yeah, that’s true. I mean, reliability is probably the most important metric for a data center. And in the PJM CIFP process, we saw that the data center coalition was unwilling to negotiate on getting firm power. So in practice, what this means is that when PJM curtails a data center, the data center is going to fire up its diesel engines, or hopefully its batteries. But realistically, these are going to be diesels, which brings us into a lot of local environmental concerns.

Stone: So they’re not necessarily going to stop operating, but they’re going to switch to different, potentially dirtier types of generation. That’s another problem.

Rutigliano: Yeah, exactly. I don’t think anyone builds a data center that doesn’t have on site backup power.

Silverman: I think this question that you and Tom are going back and forth on is really important, about flexibility. So the fact is, if we could convince the data centers to be flexible— to turn off during peak hours. Even a very small number of hours, 10, 20 hours a year. The amount of infrastructure we need to serve them goes down by a lot. Because, you know, we build out the grid to meet that super-peak period where, you know, everybody has their air conditioning on, and the data centers are operating full tilt. So if you tell the data centers and get them to agree to be flexible— or to connect and manage, really. It’s another term for flexibility. We don’t need to build as many transmission lines. We don’t need to build as many new power plants. Prices don’t spike the way we’ve been seeing. I’m not gonna lie, they’re still going up, and it still is a huge problem for affordability. But everything gets easier. Except, of course, the utilities make less money, the generators make less money, and the data centers are really unhappy. Because, as you’ve noted, they want to be operating 24/7, and the value to them of staying online is extremely high.

So we have this really kind of interesting question, where, from a policy perspective, can you establish market price signals that reward the data centers for being flexible? Or can you give them other goodies, like faster connection times to the grid, in exchange for being flexible? And I think one of the things that both Tom and I, when we were at the CIFP process, really tried to make it clear. Data centers, if they want to, can just build their own power plant. They can agree to be flexible. They can agree to buy a power plant from somebody else, so long as it’s new investment. And, you know, we are going to give them a whole bunch of commercial options, and they get to pick. This is like capitalism, 101, right? They will manage their own scarcity better than we can manage it for them. So that’s, I think, the big challenge right now. Is, are the data centers, frankly, going to put their money where their mouth is, and say, “Yes, we are willing to own this problem.”

Rutigliano: You know, I don’t know if it’s quite as simple as, “Everybody gets to build their own generation.” You’ve got one market that serves general customers, one generation market that serves the serves the data centers. Because there are also supply chain constraints impacting the rate at which new generators of all types can be built at this point, that point to other problems here that I don’t know if we have time to fully get into. But if there’s a demand for certain number of new generators and grid infrastructure, only so much can be built in a certain amount of time. For example, for gas fired power plants, there’s a four or five, six-year waiting list to get turbines.

Silverman: This is sort of the elephant in the room, right? Which is, everybody recognizes that no matter how high we send the capacity market price signals in the next couple of years, we are looking at the 2030s, at the earliest, before new power plants are online. And that’s for a whole bunch of reasons we can get into. Everything from slow processing times at utilities to slow interconnection queue at PJM. But also because everybody wants a turbine right now. And if you go to GE, and you put down your $2 billion to buy a new turbine, they’re going to tell you to come back in 2029, 2030 at the earliest. And then you’ve actually got to build the thing.

So here’s the problem. I mean, you know, this is really the challenge. Prices are going up, and the supply response can’t get online fast enough. So this gets us into a lot of the questions around, what do we do in the meantime? You know, if you start thinking, “Okay, by 2035 we’ve solved the supply chain constraints. We’re building on new SMRs or new natural gas turbines or whatever the answer is. Okay, great. We can deploy infrastructure fast enough to meet the new incoming load. But that’s certainly not the place we are today.

And so I think a big piece of what the PJM board, and FERC, and frankly, all of us are trying to grapple with, is this reality that the load is coming way faster than the supply can respond. In a lot of cases, there isn’t even supply to respond because of these supply chain shortages and everything else. And so how do we allocate the financial and reliability risk between where we are today and where we hope to be in the mid 2030s? And that is, at its heart, what this whole CIFP business is all about.

Rutigliano: The problems Abe laid out are really why we think that enhanced interconnection track is not going to solve the problem. Because even once you get through the queue, you have to get a turbine. The problems Abe mentioned, plus a bunch of other things, really point towards large scale energy storage being a solution here.

Stone: Okay. So I want to jump to this issue here of investment incentives. Abe, you just brought that up. And incentives are very important in these competitive markets like PJM and other markets around the country that are like it. I wonder if you could first tell us, why are investment incentives and grid reliability so intertwined in the case of these large electricity markets? And then Abe, I wonder if you could introduce us to the concerns that the board has raised around the current incentives in the market, and what it might want to see changed?

Silverman: Yeah. So the platonic ideal, and the way these markets were originally envisioned, is we had this very orderly process where PJM would look out into the future a few years, and say, “Okay, we need some more generation.” And prices would start increasing. Investors would respond. They would either increase the capability of their existing power plants, they would spend money to maintain their existing power plants, or if the prices got high enough, they would even invest in brand new infrastructure. And there was this lovely concept that this three-year forward market would send prices that investors could then use to go finance and actually build the new power plant.

And, you know, that’s a very good theory. However, PJM has really struggled to make it work. And you could have an entire podcast. In fact, I think maybe we have had an entire podcast on all the reasons why the PJM market has failed to send these appropriate price signals. They haven’t even been to hold the auctions on time. So we’re stuck in this much closer to prompt auction market structure.

And so we basically are trying— we’re being asked by the data centers to bring more new generation, more power to market than the system is capable of handling. And the system, even if it were in peak form, couldn’t respond to that quickly. But we’re not in peak form. We’re in rehab right now in a pretty big way, because we’re still stuck in a broken interconnection queue and market signals that aren’t there. So investors are looking at the PJM market and saying, “Huh. Boy. The price of power plants has doubled in the past couple of years. So we’re talking about financing much bigger chunks of money than we were in the past. And the price signals aren’t out far enough into the future for us to really make an investment. And there’s just a ton of political risk, because honestly, we don’t know what the PJM market structure is even going to look like in a couple of years.”

And so as a result, we are not seeing— and the folks on Wall Street have been very clear about this. The independent power producers, who are the main investors in power plants, have been very clear about this. They do not see the current PJM market structure as giving them the investment certainty they need to make a multi-billion-dollar investment that they’re planning to recover over the next 20 to 30 years. And so as a result, this lovely system that we have that’s designed to reduce risk and to drive investment isn’t having that impact.

And so a lot of what we’re talking about here— and I think we’re about to talk about some of the some of the capacity backstop reforms that are under consideration right now— are because Wall Street simply isn’t showing up with the capital to drive the investment that the data centers are asking for. And so we’re really getting into this question of risk assignment. How much of the risk should be borne by utilities and regular retail customers like you and me and Tom, and how much of it should be put off on the data centers. And that’s really the heart of, who is going to pay? Who’s going to pay for this next generation, this trillion dollars of investment we’re expecting over the next decade? Is that on consumers? Is that on data centers? And how do we get investors aligned to actually deploy the capital, the enormous amount of capital that we need them to deploy.

Stone: Okay, so so far, we’ve gone over— and I’m just looking at the PJM— the 14 page letter. And it’s got this nice little table that talks about the six key elements that are in its guidance to PJM staff. And we’ve talked about the load forecasting issues. We’ve talked about the “bring your own generation” and the expedited interconnection track. We’ve talked about this connect and manage. And now we’ve talked about market investment.

So we’re going to go on to the last couple here, and I think if everybody can hang with us, we’re going to tame this big monster. The next one we want to talk about is something you just mentioned. I’m going to pass this over to Tom. This concept of a backstop auction. So in PJM board’s letter, they address the backstop auction. And that backstop auction conceivably exists in the market today under a very defined set of circumstances. But a backstop auction has never been run. What is a backstop auction, Tom? Under what circumstances, under current rules in the market, can it be used? And what is the board looking to do with that backstop auction?

Rutigliano: The current rules basically say if PJM is short on reliability for three years in a row, they can have a second auction that is oriented just towards new supply. So I think we’re largely just going to take the name of that and build a new auction structure. No one’s going to wait two more years, right? 2027 is the first year PJM is actually short on reliability. And the White House also demanded, and the governors, that PJM this year hold some kind of auction. And I think what the board is going to do is use those backstop auction rules to move forward with something like that.

The basic idea is going to be to address many of the problems Abe mentioned, in that this will be an auction that’s just for new supply, and will give them some sort of longer-term firm price. So it’s kind of an investor’s dream auction, right? It lets them get a price that’s higher than people might be willing to pay for existing power plants, and gives them some kind of guarantee for revenue over many years. But the problem here is that if you have an auction and nobody comes, it’s not going to help. There are so many barriers to new supply in PJM right now, that it’s unclear simply holding an auction and writing some very big checks is going to solve the problem.

Stone: There has also been the concept that’s been raised, or the concern around, what’s called— I guess people are terming it “cannibalizing” the market. So if you have this backstop auction, which would be an auction to purchase capacity for the AI data centers, as you mentioned, the prices in those auctions would likely be higher. Because the data center companies are willing to pay a lot of money, because they really desperately need the power. Conceivably, more money than regular people in their homes like you and I would want to pay. And does that create a problem, then, where all these proposed new generators or companies that want to build new generation would participate in that backstop auction, and then kind of overlook the main auction, to the detriment of everyone else?

Rutigliano: That definitely is a risk. I think any new power plant now is almost certainly looking to the tech companies to sell their capacity. So yes, you’re absolutely right. I think the only place we’re likely to see new generation that just sells into the general PJM market will be generation built by utilities under some kind of state order. Investors and developers are going to go where the money is. And right now, the money is serving data centers.

Stone: And you mentioned also— and this is kind of curious. A few hours before PJM issued its letter on January the 16th, the White House and a group of PJM governors issued another letter, as you just mentioned, requesting that PJM look into running an emergency backstop auction, which I think would be in August or September of this year, which is an extremely short timeframe under which to put that auction together. And any structure of the auction would actually have to be first approved by the FERC.

Rutigliano: That’s right. And, you know, PJM is under no obligation to do what the White House says. FERC is independent. And you know, we certainly saw during the last administration, PJM obviously listens to the White House. But they don’t necessarily follow marching orders. So we hope that PJM maintains their independence now. That’s critical to a well-functioning grid. But yeah, they are likely to file something. I don’t know if it will be by September. And I would expect FERC is going to move pretty quickly. So, you know, it might not be on the timetable the White House wants. But we are going to see this move forward at a pretty rapid pace for PJM.

Stone: Okay. And there’s one final issue here, and I think we’re through. And this is what they call the capacity auction price collar or price cap, that was put in place to kind of tame these capacity prices. Abe or Tom, I don’t know who wants to just tell people what that’s about, and what the debate is at this point.

Silverman: You know, I’ll just say, we probably should have started with this issue. Because the single most important thing we can do today as a society, to maintain affordability, is to re-up the price collar, they call it, because it has both a floor and a ceiling, that was enacted an auction and a half ago, because of recognition of how broken these markets are, and what a period of turmoil we’re in. The market monitor—and I just want to be really clear about this— came in and said that without the price collar that was result of a complaint from Governor Shapiro of Pennsylvania to FERC— if that hadn’t happened, consumers would have been paying $26 billion in the last auction. Which is 10 billion higher than the 16 billion it cleared at.

And you know, I know I’m throwing around a lot of numbers here. But this is a really important affordability issue for consumers. And we often talk about— consumers, frankly, don’t really care about what’s happening in the PJM capacity market. But you know what they care a lot about? They care a lot about their bill that they receive in the mail. And a number of customers had this really unpleasant surprise of seeing their bills go up 20%. Of course, it depends on exactly where you are. But a lot of customers saw their bills increasing by 20% between the 2024 and 2025 energy years. Without the cap extension, we are likely to see another 20% bill increase for retail consumers, and that is simply politically unpalatable.

So I think, you know— and it’s no mistake that the first thing that the governors talked about, and that the White House talked about, was extending the existing price cap to at least stop the enormous increases that consumers could face if we don’t take action. And there’s a lot of reasons why I think it’s also sound economics, just given how uncertain the current market structure is. How uncertain the load growth is. You know, talking about, are we going to enact some sort of mandatory flexibility requirement? How many data centers sign up for FERC’s new flexible transmission service. I mean, there’s just a lot of reasons why it makes sense to extend the cap. But it is literally a $10 billion a year problem, facing consumers who are already having trouble paying their electric bills.

Stone: And some might argue that that cap shouldn’t be there, because you’re stifling or muffling the investment incentive right? But on the flip side, if all of this future load never materializes, and we’re paying for this capacity we don’t need, that’s the other side of the argument as well.

Silverman: Yeah. And I would say, Andy, that it’s also about whether anybody can respond. I mean, listen. Anybody who has a turbine or has an existing power plant that they can improve, is already doing it. You know, people are desperate to get new capacity deployed on the system. So why are we going to send an even higher price signal that people simply can’t respond to?

Stone: Okay. I do want to ask you another question. Both of you. A question that is also equally difficult. What is your view? Start with you, Tom. Does the board’s plan strike the right balance between maintaining reliability, protecting rate payers, and enabling rapid data center growth? What’s your overall take on this one, wrapping it up?

Rutigliano: Overall, I think they’re moving in the right direction. You know, we agree with Abe pretty strongly that they should maintain the cap. But it’s still not a full solution. PJM needs to really dramatically reform how they connect power plants to the system, and let this happen in months rather than decades. So the stuff they’re doing is going in the right direction. But until they fix that fundamental supply block, this is always going to be band aids. The best they can do is buy time.

Stone: Hey, I want to ask you one other thing. You’re very concerned with the environmental outcomes here as well. What’s your view on the board’s direction and its potential environmental impacts or not?

Rutigliano: You know, that’s relatively neutral. I think we’re actually really in favor of the bring your own generation approach, because we think, if that’s coupled with the right interconnection reforms, that lets the states that host the data centers have more to say about what kind of supply. And you know, many of the data centers are in Virginia, Illinois, states that are heading towards zero carbon. So if PJM’s proposal pushes that into more locally-sited supply— and we’ve been saying generators, but we have to realize, energy storage sits side by side with power plants in all of this. That is finally going to create the structure that would give the tech companies the incentives to invest in clean supply, which in this case is almost always going to be storage.

Stone: And Abe, last question for you here as well. So you know, going on what Tom was just talking about, the balance here. A year or two from now, if we’re looking back at this process, what will tell you whether PJM got this one right or wrong?

Silverman: Yeah. I think it’s a question of whether Grandma is seeing her electricity prices continue to rise, or whether we are successful in directing the costs associated with serving the data centers to the data centers. And I think the PJM— from what we’ve seen thus far, both the governors, the White House and PJM seemed aligned in this concept that data centers need to pay their own way. But, you know, the devil is absolutely in the details. Pick the cliche you want. We’re going to have months of seeing just how serious the PJM board is about protecting existing customers.

And, you know, it’s really hard. It’s really hard to go against some of the biggest businesses in the country. But at this point, you know, I think— I hope— that we’re reaching sort of this national consensus that existing retail customers need to be protected and that the data centers should be paying all the incremental costs on the transmission grid, in the energy markets, and in the capacity markets associated with the new infrastructure needed to serve them.

Stone: Abe and Tom, thank you both very much for talking.

Rutigliano:  It’s been a pleasure, Andy. Thank you.

Silverman: Likewise. Thank you very much.