PJM Grapples With Data Center Load and Supply Gaps

Game Insight on the innovators, disruptors, and tech-driven trends shaping today's complex economy. I'm Carol Masser. And I'm Tim Stenevek. Wrap up your workday with the Bloomberg Business Week Daily Podcast. We bring you deeper dives into the story shaping your world from the evolution of AI to the shifting priorities of global business. Plus, Silicon Valley power players and the latest tech trends. Catch up on the conversations you missed during the day. Subscribe to the Bloomberg Business Week Daily Podcast on Apple, Spotify, or anywhere you listen. This is Tom Rowland's Reese and you'll listen to Switched On the Podcast bought to you by Bloomberg NEF. After years of relatively steady demand, PGM, the largest power market in the US, is entering a period of rapid load growth driven largely by data sensors. This surge is reshaping the system, putting pressure on supply, pushing capacity prices to record highs, and forcing policymakers, utilities, and developers to rethink how the grid evolves. At the heart of the debate is a simple question. How do you meet that demand? In PGM, the options largely come down to two levers, building new supply faster, or finding ways to manage demand. But the balance between those approaches is far from straightforward, with competing interests across regulators, consumers, and large power users. So what happens if supply doesn't keep up? And what can other regions learn from the challenges PGM is facing today? On today's show, I'm joined by two members of Bloomberg NEF's US Power Team, Alice Lara Kammann and senior associate David Mohamedi, to discuss the report, PGM Power Market Outlook 2026, Race for Impact. The NEF plans can find this note and other power markets research by heading to be any if go on the Bloomberg terminal or be any if.com. If you'd like to learn more about how BNF approaches strategy research on the energy transition, including developments in commodity markets, trends across different sectors, and the cross-cutting technology shaping the future, you can find more information on be any if.com. And if you'd like to speak with a member of our team about becoming a client, email us at [email protected] But for now, let's take a look at the PGM Outlook. David, welcome to the podcast. Thanks for having me. Lara, welcome to the podcast. Thanks. So PGM, I was chatting to your guys, Boss, Helen, shout out to Helen, because I know she's put a lot of work into this. And we're talking about how PGM is kind of, if modeling US power markets was a video game, then PGM is definitely the final boss. I don't know if you guys agree with that statement. Why is it that PGM is the final boss? Why is it that it's just so hard to get your head around it? Yeah, and I'll take a stab at this and then maybe Lara, you can give your perspective as someone who's kind of fresh to this process. But sitting on the power team for roughly three and a half years now, I would say, I've been fortunate enough to watch our team model several other power regions. And each one has their own sort of circumstance that comes with the market. So with Urquit, it's notorious for its large load growth. And then with California, one tidbit is that it's part of a carbon market. And with Myso, you have to handle the gas to cool switching as well as the unique thermal mix that it has with both cool and gas. And then with New York ISO, there's behind the meter solar and then a new technology of offshore wind. And then you arrive at PJM, which is kind of like this final boss where you have to check mark all these other challenges with other markets. And then you have to model all of those in one region. And that's not to say that other markets are easy, but we had to take in a lot of the work that our team has done to make a successful market outlook. So Lara, David's talked about a lot of the challenges with PJM, not necessarily naming them. But what are some of the things that you had to deal with? I think one of the most obvious factors is large load growth from data centers. That was a really big challenge to model looking at that and then placing that in respective regions where we're seeing that growth. I think PJM south is where the majority of that growth sits. And so thinking about that and then thinking about how the region's going to handle that was what the biggest challenge I think in thinking through our scenarios. So yeah, I'd say the large load growth from data centers. I sit two desks away from you guys. So I witnessed you putting in a lot of late nights, a lot of stress. You definitely didn't choose the, I mean, we say it's the final boss, but you didn't play this game on easy by any means because I think you had four different scenarios as well. So we're trying to forecast the future of this power market and we instead of just saying, okay, here's what we think is our best guess at the future. We did do that, but we also throw in three other versions of the future. So why, why did we make life so much harder for ourselves to start with? It kind of stems from what Lara mentioned with the large load pipeline that's coming online. As we were researching, we noticed that there was constant debate from various stakeholders in the market of how to handle this large load growth and that includes PJM itself. It includes large tech companies. It includes consumer advocate groups and then even the federal administration and FERC. All of these parties contribute their solutions into well, how the system should approach this large demand. So there's all these different stakeholders that are throwing in their opinions as to how PJM can adapt to some of the challenges it faces. And we're going to get into what those challenges are in more depth. I mean, you've talked about large load growth, but I'm guessing all those different stakeholders are proposing different solutions and it probably means that their position is reflective of what suits them best. So what are some of the different positions and stakeholders that align with those positions? Yeah. So we typically found solutions in a couple categories. First, there is bring your own generation, which was advocated a lot by the internal market monitor, monitoring analytics. So could you just explain for listeners who don't know what an internal market monitor is? What do you mean by that? It's a FERC appointed party that is. FERC, by the way, is the federal regulator of energy in the US? Yeah. It's appointed by this federal regulation entity and their job is to review and suggest areas of improvement for the PJM grid. So how they're handling the challenges ahead of them? Yeah. So that was what the market monitor was saying, bring your own generation. Yes. Data center providers were saying something else, which was. Yeah. A lot of the data center providers were not in favor of curtailment because they want to keep their activity up and running. And so the data center coalition, a lot of their proposals hinged on voluntarily curtailing rather than being forced to do so during a stress event. And so that was one example. Can we just make sure we understand what we're saying? So there is this possibility that there were going to be rules that said data centers would have to turn down their loads at certain times to allow the grid to cope with it. And that was potentially going to be mandatory. But the data center providers are, I mean, unsurprisingly, I like, we would prefer that not to be mandatory. Exactly. Yes. No. And then as David mentioned, there were a lot of consumer advocates. And so their main priority is to keep prices down for rate pairs. And so they were kind of going into all in scenario of either data centers bringing online through in generation or having them curtailed during the stress events. And so got kind of got this array of different people putting forward their proposals. We've talked about data centers quite a bit. I realize we're doing this all in a funny order. And you know, maybe I in my questions, I put the cartbell before the horse. But let's just quickly for our listeners who are maybe not familiar with PGM, can we just talk about the importance of PGM when it comes to data centers and why this is such a central part of the question? Because I mean, the PGM is the biggest power market in the U.S. to start with. So it's already important. But why are data centers so central to everything we're talking about today? Yeah. I mean, one because, yeah, PGM hosts the largest data center fleet. It's I think I think I think I mean, my mental model on this is like half of the data centers in the world are in the U.S. And half of the data centers in the U.S. are in PGM. If that's wrong, then that's on me. You know, you guys, but like I think that's the that's the style I heard somewhere. Okay. I mean, as an analyst who focused on PGM, it sounds right to me. Sounds right to me. So there's a lot of data centers in PGM. Yeah. And how they handle the data center pipeline that they're seeing really could set precedent to the rest of the U.S. Right. And this is important just because of we're seeing such a rapid expansion of data center capacity with all of the demand for AI, right? Exactly. So when we're looking at PGM and we're looking at all of these things you just mentioned around bring your own generation and flexibility, what seems like maybe on the outside that we're being kind of power nerds and talking about the intricacies of that is actually super central to this broader question about how power grids in the world to everywhere could potentially be handling all of these demands from AI. Like PGM is the heartland of that question. Is that a fair statement? Yeah. I would say so. So okay. Cool. So you guys, you created a bunch of scenarios based on all of these different proposals that have been made as to how to how PGM can basically handle all of this data center load. Could you talk us through the scenarios that you created? So as we mentioned, it comes down to two main levers that we've noticed. How much additional supply can come online and how much our data centers willing to curtail. And so we run through, like you said, four scenarios. The first scenario is what we call the chicken little scenario. That's if none of these levers are applied and we just let the demand growth kind of run rampant and impact our power prices. What without building any extra generation capacity without any extra energy. So basically nothing nothing happens. No data center flex, nothing happens. And then we go through a scenario of well, just expedited supply. What are the impacts of bringing online? Something that we modeled through PGM's critical issue fast path process. It's one of their proposed solutions of an expedited interconnection track. And then we look into data center flexibility of different amounts. And then finally, we apply all the solutions into our base case scenario, which we publish in our report. I mean, you would think that it's really obvious that if there's going to be more demand. there's going to be more supply. Yeah, PGM has this, I remember, you called it this fast track process. Right. So why is that even necessary? I would say because the load growth at PGM is currently seeing, as we've mentioned, is unprecedented. And so this region hasn't really had to deal with this amount of load coming on this fast ever before. And so in order to, as you said, meet this demand, supply has to come on at a quicker rate in order to meet this demand. So like in a way, the system isn't set up to add supply as quickly as is needed, because it's never had to in the past. Got it. So the scenario where we have expedited supply is by no means assured, right? Not at all. Granted, it's modeled around a framework that PGM is currently proposing to FERC. So this framework is looking at getting 10 projects approved with a three year timeline to delivery. And it has some requirements on top of that, but we use that as our basis to figure out well, how much could we model if we needed to? So in short, these different scenarios you've created, they're reflective of the different things that have been proposed to deal with all of this additional load that we expect to come online in PGM. So I suppose like, cutting straight to the chase, what did you guys find? I think our main finding when looking specifically at PGM, South and West, where we expect to see the most growth from data centers is that in our chicken little scenario and our demand flex scenarios, both of which do not apply expedited supply, power prices go up much higher than in our base case, and our expedited supply scenarios in which again, we bring on the extra supply. And so this just signals that bringing online more supplied quicker rate is what is really going to be meaningful in suppressing power prices, specifically in those two regions in PGM. So I just want to make sure I've understood what you've said. Bringing on supply helped keep a lid on power prices without bringing on supply power prices. I mean, they increase by a lot. App prices go through the roof, but with more supply that keeps a lid on things. But also, I think you said the flexibility doesn't have the same impact on keeping a lid on prices. Is that right? Yes, that's correct. So you come with quite a clear conclusion then is that the expedited supply is going to be essential to protect the interests of consumers, I suppose. Yeah. So then why flexibility? I mean, why are we even talking about flexibility then? If it's not meaning for you keeping a lid on prices, well, while it doesn't impact average annual wholesale power prices or even really average monthly wholesale power prices, it can have an impact on other things such as grid reliability and the capacity market, which is another big component of the PGM power market. So it does play a role. And then we also even ran a kind of case study where we applied the 2025 January polar board text, weather conditions on the fleet in 2035, both data center and power supply. And we found that demand flakes does make a big impact during these high stress events, lowering prices by, I think, if 50% of data centers participate in lower prices by roughly 20%. So it does make an impact. It's just when you average it out, it doesn't make it gets a little bit lost. Yeah. Whereas adding supply keeps a lid on prices across the board. Exactly. Big Tech podcast from Bloomberg News keeps you on top of the biggest stories of the day. By fellow Americans, this is Liberation Day. Stories that move markets. Chair Powell opened the door to this first interest rate cut. Impact politics, change businesses. This is a really stunning development for the AI world and how you think about your bottom line. Listen to the big take from Bloomberg News every week day after news on the I Heart Radio app, Apple podcasts, or wherever you get your podcasts. Okay. So since we've established that adding supply, I mean, it sounds like flexibility plays a role in the future. If that can be brought online and so it's important, but it's almost what I'm hearing is it's it's almost irrelevant if there is an also additional supply. Yes. So we've mentioned the sort of how long it takes to get supply online and that sounds like it's like a regulatory issue, maybe to some extent. But what else is like holding supply back? If we cleared all the regulatory bottlenecks, would it just be a case of like, let's build some more gas turbines? Is it that simple? I mean, I think on the gas side of things, one of the major constraints that our gas team touches on a lot is the gas turbine shortage. And so that is something that can definitely hold up times for bringing a gas plant online too. And then looking at renewables, a major setback is the passing of the one big, beautiful bill act, which removed a lot of key tax credits. So I think those are two pretty large setbacks in both the thermal and the renewable spaces. I mean, you called your scenario where nothing happens chicken little, which I mean, I didn't actually get the reference. I mean, maybe you could quickly explain why you called it chicken little. Well, it's kind of based off of that movie, chicken little where there's a little chicken that has a little piece of the sky falling top of his head. And he runs around and tells everyone that the sky is falling and nobody really believes him. It turns out the sky actually is falling. And so we wanted to use this scenario as kind of a warning. Well, if you don't apply some sort of solution from what we found out that being some way to bring on new additional supply, then you'll see wholesale power prices increase. Got it. So I mean, I know I've kind of asked this twice now and you've given me answers, but like, it sounds like from what you've said and from what Laura said is this chicken little scenario is both disastrous, which is why you called it the chicken little scenario. And nothing is done to address the challenge faced in PJM and it is also distinctly possible. Yes. Do you see a situation where it's like chicken little nothing can be done to make data centers flexible? Nothing can be done to bring on more importantly to bring on new supply that some of those data centers just can't get built because of what it could do to the system. Is that a likely scenario? Is it more likely that just power gets really expensive in PJM and it's passed on to consumers? Well, with our demand growth pipeline that that we forecast, our data center pipeline, we do assume that a lot of the early staging committed projects don't make it through to operation. So we've already kind of given our forecast to have cut. Yes, exactly. It is still a substantial pipeline out to 2035 and not only are data centers adding load to the grid, we're seeing electric vehicles start to ramp up, especially in the 2030s and then even gas production itself draws electricity from the grid, which we expect to add on top of that. So there are multiple sources of load growth data centers just being the largest one and we kind of treat our demand growth pipeline in our base case as something that we stand by and we don't really look into alternatives to that. Right. So we're already looking at a constrained scenario. So yeah, there's going to be a lot playing out in PJM in the next decade. I guess it's one of my main takeaways here. The other thing that you guys look at and you might need to explain some of our listeners who aren't familiar what this even is is I mean PJM has a capacity market. It has a carbon market, but not universally applied. Can you kind of talk about what's going on with those and then also like how it comes back to this question of this sort of uncertain future we see in PJM? So in the market outlook, we also take a look at PJM's capacity market. And so in this capacity auction, they procure capacity for a set delivery year. And so the most recent one procured capacity for the 2027, 2028 delivery year and it had record clearing prices at about $333 per megawatt day. And with this, we saw a tightening reserve margins that BNF also forecasts stay quite low going on to the future. And so along with our forecasts, we've also seen PJM install a price cap for the next two years going forward. Just signaling that they expect prices to remain elevated and they need to put a cap on it. So it doesn't go into rate pairs. I just want to like kind of dive into what this all means. And firstly, like there may be some listeners who are not familiar with the capacity market. So I'm going to try and explain it and you guys tell me whether I got it right. Capacity payments are when certain generators get paid for being around not necessarily doing anything, but just being a friend in the room in case they're needed. Correct? They're also the payments can be reduced if they don't perform when they're needed to. Right. If you don't turn up, then you don't get paid the way you're supposed to. Right. And then that market is it's like PJM will say like, okay, we think we need X amount of capacity for next year and they will basically bid and say like, well, here's how much you can pay us to be your friend and stick around because maybe in wholesale markets, they wouldn't necessarily be recouping their costs, right? Is that correct? Yes. So then you mentioned that there's record prices for, was it 2027? For the 2027, 2028 delivery year. Got it. So basically that's kind of an early indicator of how much any kind of firm capacity is going to be hot property in PJM. Is that a fair analysis that never before we've seen this amount of need for capacity in PJM? Yeah. It's one of the market indicators to bring about new build as well. So that $333 while it's capped, I think if it was uncapped, it was going to be over $500 per megawatt day. That $500 per megawatt day is the market saying, well, we're willing to in a way fund new build of, say, gas capacity or energy storage. So one thing I want to understand, I mean, because I didn't realize, I know I'm supposed to have read this report. But I didn't actually realize that they'd put a cap on the capacity prices. And in all of these markets, obviously they are regulatory constructs. Capacity market is a regulatory construct. Energy market is a regulatory construct. But generally speaking, maybe the neoliberal view. is that they work best when they are sort of left without intervention. And I mean, my guess would be like, if they're putting a cap on capacity prices, that gives less incentive for new builds. It gives less incentive for some plants that would shut down to stay open. And so isn't that just going to mean them putting a cap there? Isn't that just going to mean higher wholesale prices? Given that you're modeling as established that capacity is the thing that's going to keep a lid on wholesale prices. I mean, do you have a view on whether this was a sensible move capping the prices? If the objective is to keep things cheaper for ratepayers, are they just pushing the problem from one market to another? It's the question that PJM themselves and the governors and legislators of the states within PJM are talking about. And it's, are we willing to take these high capacity prices? And those also go into the electricity bills every month that the rate-based pays. Are they willing to either pay for it in the capacity market or are they willing to pay it in the delayed, maybe new build, unless economic incentive to build new capacity in the wholesale market? So it kind of seems like they're putting a band-aid on the solution, which is a greater issue and just kind of letting the issue persist in the wholesale power market for the time being. Yeah, I mean, because I mean, in ERCOT, they don't have a capacity market and they just allow the energy market. So like basically capacity stays online in the hope of like extreme prices in the energy market because there isn't that much capacity. Is that like where PJM could potentially be heading with this kind of thinking or is it, or am I kind of maybe over dramatizing this decision? I don't know if I would lean towards that because on one end, you also have the Trump administration bringing in this 15 year sort of PPA-like auction for a new build. And that's almost like a capacity market, but just extended way out. Got it. So there's other incentives. There's a lot of different moving parts. It's hard to say. This is why I mean, it kind of comes back to what I was saying at the start. Maybe this is why PJM is the final boss because nothing is simple in PJM. Like you have all of these intersecting incentives going on. And then the other one that we didn't talk about yet, but we're going to talk about now the carbon market. How is that impacting PJM? Yeah. So the carbon market only exists in three states in PJM and they all sit in our defined PJM East and those states are Maryland, Delaware and New Jersey. And so within those states, they see a rising premium through the carbon market. And so we see fuel prices taking upwards slightly going forward. And so that's just going to play a role in power prices down the line as well. I mean, I just want to understand this because it is kind of you got one single market here PJM. And then you've got some operators in the market playing under a different set of rules to others. So certain generators in certain states have to pay a carbon price, but they are selling their electricity into a market where there are lots of similar generators not paying a carbon price. Does that just put them at a disadvantage? Because if it doesn't apply to everyone, I mean, does it not just mean that thermal like fossil fuel capacity just moves to a different state and generates there at a lower cost? And I mean, how does it all or is there some corrective mechanism to that? There's multiple factors. I think that's one of them. It doesn't help that the thermal assets in PJM East have to pay this carbon premium. That's one of the reasons why there's not a lot of coal capacity left in that area. But I think maybe a bigger player that determines the economics of these thermal plants is the underlying fuel costs that they pay. So our PJM West region, which encompasses Ohio, West Virginia and then Western Pennsylvania and Northern Illinois, that's home to the majority of Appalachian gas production, the region for Appalachia. And those thermal assets in that region benefit from really cheap gas prices, really cheap coal prices due to low transportation costs compared to PJM East or even South, which is mostly Dominion's utility region. Dominion and PJM East have to pay a premium due to pipeline constraints for gas. So yeah, it's kind of assets and PJM East might struggle a bit relative to the assets in PJM West. Okay, this is interesting because it also kind of maybe answers a question that I had in the back of my mind when I was reading your report because PJM East is playing by a different set of rules. And now you've just highlighted it has doesn't have the same access to cheap fossil fuels as other areas. And all of the data center load is being built not in PJM East. It's being built in PJM South PJM West. And is that why that is? Maybe one reason, but maybe not the biggest. So we found that a lot of the data center pipeline sits in Virginia because of the strong fiber network that it can take advantage of as well as the workforce that's already there. And then we noticed that campuses are starting to pop up in states like Ohio and Northern Illinois and Western Pennsylvania because of the amount of land that's available and the availability of power, which I think goes to your point. So yeah, that might play a part. I will say though PJM East isn't just suffering without any solution. It's a state that's seeing a lot of behind the meter solar bill. So as a result, that's kind of one of the factors that's driving prices down there is in the summertime. I mean, it doesn't get too bad. Because I mean, I think I'm kind of painting PJM East as a bit of a victim. But actually, I think in your modeling, PJM East is not subject to the same like price hikes as those other regions. It sits our price forecast and historical also power prices for PJM East sit below South and West. But we do see it growing the strongest just by a few percentage points. But yeah, overall, it sits below the other two. Right. So it, okay. So it grows. That's in the base scenario, which includes the expeditions supply. But PJM East is maybe more immune to chicken little. Yes. The sky does not fall on PJM East's head in the way that it does on South and West. Exactly. All right. Nice safe peat. I know where I'm going to live if I want just if I want cheap PJM power. There you go. The big take podcast from Bloomberg News keeps you on top of the biggest stories of the day. My fellow Americans, this is liberation day. Listen to the big take from Bloomberg News every week day after news on the iHeartRadio app, Apple podcasts or wherever you get your podcasts. What do you think needs to happen in PJM? And what do you think are some lessons that can be extracted for other regions? Because you know, although it's saying like 25% of the data centers in the world are in PJM in the future, they might not be. They might be elsewhere and other regions have data centers. So it's kind of like the canary and the coal mine in terms of this question of extreme load growth. So what have we learned? I think one of the main takeaways from our modeling is obviously that bringing online more supply can be a very useful lever in suppressing power prices. And then I also think even though in our wholesale power prices, as I mentioned before, demand flex does not reduce power prices on an annual scale. I think it's really key to look at those stress events and how demand flex can be used in those scenarios to reduce power prices, especially as we can possibly expect to see more cold snaps and more heat waves going forward. It'll be really valuable lever to be utilized in those situations. And so I think finding there's no one solution, finding a combination for what applies to not just PJM, but to whatever region you're looking at is going to be really key in addressing any large load growth. So it's like a combination of the old and the new. Yeah, exactly. David. So like Laura mentioned, it comes down to building new supply. And I think we made an example that successfully executing something like the expedited interconnection track that PJM is currently proposing offers a solution to that. And I like the idea of improving the process of bringing new supply rather than just bringing online a large generator every five years, for example, having some sort of robust streamlined way to do that repeatedly seems like a key solution. So other markets might consider getting that ready now to future proof their processes. Yes, exactly. I do know it's a challenge securing supply nowadays, especially gas supply, just the equipment that goes into it and the labor it's becoming more of a challenge. So we face uncertain times. We've identified the solution, but we've also identified that it's not necessarily easy to implement the solution. That's what I think we've kind of landed on. So I mean, perfect. I mean, be only if we bring clarity where there's chaos. And what we've learned is that there is chaos in PJM at the moment. And on that note, David, thank you for joining us today. Thank you so much for having. Laura, thank you for joining us today. Thank you for having me. Today's episode of Switched On was produced by Cam Gray with production assistance from Kamala Shelling. Bloomberg NEF is a service provided by Bloomberg Finance LP and its affiliates. This recording does not constitute nor should it be construed as investment advice, investment recommendations, or a recommendation as to an investment or other strategy. Bloomberg NEF should not be considered as information sufficient upon which to base an investment decision. Neither Bloomberg Finance LP nor any of its affiliates makes any representation or warranty as to the accuracy or completeness of the information contained in this recording, and any liability as a result of this recording is expressly declined. [MUSIC] The Big Take podcast from Bloomberg News keeps you on top of the biggest stories of the day. By fellow Americans this is Liberation Day. Listen to the Big Take from Bloomberg News every week day after news. the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.