Ep. 026 - PJM's $12B Modeling Mistake Is Hitting Ratepayers Again (Datacenter, Energy)

This is my mother's background. We have crystals that help you sleep, a Buddha, and a kiwi—or what looks like a kiwi.

Yeah, that's a kiwi from New Zealand, right? They lay the little green eggs that turn into fruit.

Yeah. I love those. They're really tasty.

We're going to talk about U.S. ratepayers wasting $12 billion they don't have to waste because PJM is making a modeling mistake, and they want to do it again. Rob, welcome to the show. I appreciate that you're there in your man cave with the crystals and the kiwi. It's looking great.

Yeah. I've got to balance things out and bring out my feminine side, I guess. I love the crystals. Who doesn't love crystals?

Yeah, we're going to have a nice feminine podcast where we talk about the electricity grid and Pennsylvania, Maryland, and Virginia today.

Let's start with the basics. The headline claim you're making here is that—well, first of all, maybe you can explain, at a really basic level, what PJM is, what's happening in the electricity grid in the U.S. right now, and why Americans are going to see their power bills rise.

1. What Is PJM

Well, that's the thing: They don't necessarily have to. PJM stands for Pennsylvania, New Jersey, and Maryland, but it's actually 13 states: Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, and West Virginia. On top of that, it includes the District of Columbia. It's not all of those states; some of them are only partially included. It's a bit of a mouthful for the acronym to include all of the states.

But this is US East 1. This is data center alley.

This is 66 million people. This is data center alley. It's the largest grid by electricity delivered in America, or the largest sort of operational area in America. It's also what's driving a lot of the narrative around data centers driving up electricity prices.

There are hundreds, maybe thousands, of articles, and I always check this whenever I see that data centers are leading to increased electricity rates. I look at what the source is, and it's almost always specific to PJM. There are some very small instances of this happening outside of PJM, but PJM is a particularly large driver of both the narrative and the area that has seen, by far, the largest rate increases due to increased demand.

Our take—we've written about this before, and we're writing about it again from a slightly different perspective—is that this is due to market design, or not really even market design. It's more like bureaucratic auction design.

2. Capacity Auctions

Okay, let's dig into that specifically. PJM is unique compared with other grids, even in the U.S., in the way that it does these auctions, right?

Yeah. It's normal in electricity markets to have real-time markets and then more-ahead-of-time markets. You want to build in a buffer to how the electricity system runs, so you run these auctions years in advance to make sure that you have enough spare capacity on the system and that, in the worst-case scenario, it operates well.

Normally, you only run those markets for new capacity, because new power plants need long-term contracts and a degree of certainty to get built. Or you're running them for refurbishments, life extensions, and things like that. You're basically just trying to keep some give in the system.

PJM is unusual because it runs this for the entire system. Every power plant can bid into it, including fully existing power plants that are completely paid off. What that means is that any price increase has a massive multiplier effect across the system. If you think about an X and Y axis, you have a small price increase on the Y-axis that's multiplied across the entire X-axis.

Because this is a very large grid—the largest grid in America—that can lead to some quite large price spikes, which is what we've seen.

Yeah. This is definitely why people talk about how, as AI data centers—or just energy demand in general in the U.S.—increase over time, we need to solve this with baseload, but also with variable capacity because there are fluctuations in demand throughout the year.

This isn't implying that there's actually variable capacity for the entire grid in PJM. It's just saying that they pay for everything and do their planning up front with this market-auction-style approach, and that's going to lead to—well, what is that leading to in this case?

What's happening is that, for various reasons, PJM has a constrained supply. In the article, we map out the supply curves, and you can see the supply curve literally shifting left as fewer power plants are able to join the auction, even as demand is going up.

This is because PJM delayed its auctions for a while, and it's now running its auctions with very little lead time. Once you sign one of these contracts, you need to be delivering that power within a year, or now 2 years, which is simply not long enough for a new power plant to get built.

Supply is very constrained. The more constrained the supply is, the more it shifts to the left, but also the more vertical it is. If demand then moves at all, the price jumps. That's very normal in markets: If you constrain supply and demand increases, you get scarcity pricing.

That's basically what's happening here. Two things are happening simultaneously. PJM is constraining supply through bad auction design, not giving enough warning, and not signing long enough contracts for new supply to enter the market. At the same time, it is inflating demand, which is the main focus of this newsletter and the model in general.

3. The Emergency Auction

Because it underestimates how much power it already has available, PJM is trying to buy more than it needs. It has bought more than it needs in 2 previous auctions, which is where the $12 billion comes from. It also has an emergency auction coming up, where there's a risk that it will buy more than it needs and put ratepayers at risk of further price increases.

Basically, they're expecting data centers to take on this liability, and they might not be able to, or they might not turn up. There's a kind of moral hazard there, but I can get into the demand and supply modeling as well.

Yeah. Maybe first put the $12 billion excess payment in context. This is actually still a large percentage of the total consumer spend on power in PJM.

4. The $12 Billion

Over the last 4 auctions, they spent $63 billion, and we estimate that $12 billion of it was avoidable. All $12 billion came from the first 2 auctions.

Because of pressure primarily from Governor Shapiro, but also from other political pressure, price caps were put in place. Those price caps meant that demand and supply were so out of balance, and supply was so constrained, that even this modeling adjustment wouldn't have had enough of an effect to change things.

But in the first 2 auctions—the auction for 2025–26 and the one for 2026–27, which run from June to July and are split into those years—modeling them differently would have saved $7 billion and $5 billion, respectively: $12 billion in total across those 2 auctions.

There's only such a huge effect because supply is so constrained. You're working right at the edge of this supply curve, and it's going nearly vertical. Ideally, they shouldn't do that. That's why you have these huge price increases.

Yeah. Can you dig into both of those in a little more detail?

I've talked about the supply going vertical a bit, which comes down to not having enough warning. The interconnection queue is the other thing. PJM is currently very slow to connect a new power plant to the grid.

Because of various things PJM did, it had a pause and made a whole bunch of changes to how it runs the interconnection process. Those changes introduced a lot of delays, and PJM is running these auctions with far too little warning.

These auctions should be run with roughly 3 years of warning, and even that's a bit tight in my opinion. But they've been run with 1 to 2 years of warning, so you simply can't get a power plant up and running. That really constrains supply.

The thing we're actually getting into in this modeling effort is that power plants are more efficient in the winter. They're cooled more easily if they're coal or nuclear power plants, but most importantly, for a gas power plant, the air inlet is denser.

The power plant is easier to cool, and it's also running more mass. It's basically just a more efficient system. There are also certain ways in which cooler temperatures help electricity systems in general, but the main one is the dense air going into gas turbines.

This has been proposed in the past, but currently PJM just runs the system as if it's summer all year round. It basically acts as if it has summer ratings and efficiencies. As much as an endless summer would be a lot of fun, that's simply not the case.

Winter will come—winter is coming. It inaccurately forecasts that, or inaccurately models that, when it’s deciding how much to buy.

We model 2 things that change how you understand current power plants operating. One is that, and the other is imperviousness to winter: winterization or weatherization.

During Winter Storm Elliott in 2022, and also during the polar vortex in 2014, power plants came offline because components froze or pipelines cracked. A lot of effort has been put in since to prevent this from happening again, to make sure that these components don’t fail. A lot of these components actually failed within their operational margins, so they should have been fine. It wasn’t the case that the weather was so bad that it was physically impossible to keep these things online. If they had been properly winterized, this shouldn’t have happened.

There have been federal orders and various local mandates to winterize these units. However, the graph you’ve got up on screen explains this as well: the way that PJM models it is purely backward-looking. They only look at when power plants have failed in the past and then extrapolate that out into the future. As any good financial professional knows, past performance is not a predictor of future performance.

5. How Plants Get Paid

Rather than that, what I think would be a better way of doing it would be to reward power plants that winterize by recognizing that they are more reliable, so they would get more money out of this auction system. Rather than the system currently being mainly just a stick, you would also introduce a carrot. You would prevent yourself from overbuying and prevent the system from spending more than it needs to.

Okay. Can you explain the auction concept there again, maybe in a bit more detail? When you’re saying they’re going to get more money, explain how the auction works and how power plants actually get paid by PJM for the power they’re producing.

The key thing is that it’s megawatt-days. I’m going to lose all of our listeners at this point. The unit is the megawatt. So it’s capacity: it’s not generation, not how many electrons are actually made. It’s the ability to make electrons.

It’s like paying a farmer for having a field, as opposed to paying them for the wheat that comes from the field. It’s capacity to produce. That’s so you have this reliability margin and so that, on the worst possible day of the year, you make sure there’s enough electricity.

Not all markets have this. There’s a lot of debate over whether or not these things are necessary, but they’re seen as the more reliable approach, or the more sensible, well-insured approach.

What literally happens is PJM runs an auction and says, “We need whatever it is, 160 GW of capacity on our worst possible day. We need that much, so we’re going out to market and buying 160 GW of capacity.” They draw a price curve. They say, “If it’s really cheap, we’re willing to buy 170 GW,” because that extra margin is worth it to them. If it’s really expensive, this is the price they’re going to top out at.

They literally draw, straight out of an economics textbook, a demand curve. Then they run an auction, and people bid. They say, “I’m willing to pay whatever, and here’s the amount of capacity that I’m bringing to the table.” That draws the supply curve. Each one of those bids is a dot on this supply curve. Wherever the 2 curves cross, that’s your price and your quantity.

The key thing is that everyone gets paid this price. If you bid in at $5 per megawatt-day—which a lot of people do, and that’s not excessively small—you’ll still get paid $300 per megawatt-day if that’s what the last unit needs. That’s why you can have very big total-cost changes to an auction due to a very small number of megawatts right at the end of the supply curve.

6. The Winterization Gap

Can you explain again the concept of getting credit for a power plant that has been winterized?

Every power plant gets rated for how reliable it is, and that determines its volume. For example, nuclear power plants are rated higher than gas power plants. They’re considered more reliable and more robust.

If you’re a 1,000-megawatt nuclear power plant, you get rated at about 90%, say 93%. So you get more money, basically, per power plant than a gas-fired power station, which might be rated at, say, 80%.

If you say, “We actually think this is a really good gas-fired power station. This is a really well-winterized plant. It’s in a really good location, it’s got a very short pipeline, and various things are really well-built with good design,” then you could say, “Actually, we think that should be worth 85% or 88%.” It should get more money by bidding. It should be able to bid into that auction with more volume, basically, and then it gets more money.

This money literally covers a year, and you get paid for it. I don’t know how to say it more simply than that: the money just comes in. There are ways to violate the contract, but short of violation, it’s just revenue. It comes straight in the door.

I think, just to go back to what you said about winterization, people have gone out and done this pretty good thing after they saw Winter Storm Elliott or the polar vortex and upgraded their power plants to be better winterized, and then they don’t get paid more money.

It’s hard to say to what extent this is out of the goodness of their own hearts, because the National Energy Regulatory Commission told them to do this. They said, “It is now a legal mandate for you to go and winterize your plants.” There were a number of other mandates to do this as well. They were also fined; the ones that failed during Winter Storm Elliott were fined. So they have financial reasons not to do that again.

There are 2 things going on here. One is, you’re right: they’re not being incentivized to do the right thing, or to do the thing that the system wants them to do. But also, the system is just overbuying. If the system isn’t recognizing the value of what it already has—that is, how reliable its existing power plants are—it then goes to market and asks for too much.

What you put on the screen is trying to talk through this. It’s quite a large range here, and to be clear, this is very much a range. We’re not saying this should be the number; we’re just saying there’s a lot of margin here where PJM could be accessing value. It could be saving this money for its ratepayers if it were more accurate in terms of how it rates technology.

The big-picture question here is that the old PJM is kind of the poster child, for good or for ill, of the old-fashioned way of running an electricity system: very centralized control, pretty bureaucratic, pretty democratic in an interesting, complicated way. It very much takes it upon itself to decide how much capacity is needed and all of this sort of central planning, versus other grids. The famous one at the other end of the spectrum is Texas, which is much less centrally planned and much more decentralized.

Currently, PJM, due to odd reasons of governance and system modeling, is failing to keep up with the times. It hasn’t seen demand growth in a very long time, arguably in about 2 decades, and it is really fumbling the bag now that demand growth has turned up.

This is a shame because data centers should be a way to make electricity cheaper, not more expensive, because electricity systems have very high fixed costs. You’ve got the pylons, the power plants, and all of this physical infrastructure. As a commodity, electricity has incredibly high fixed costs, so there are great economies of scale. The more demand you have, the more you can spread that fixed cost across everyone; the higher utilization rate you can hit, the more efficiently you can run.

Also, with data centers, because they’re value-creating or money-printing machines, this has got to be by far the largest differential between the cost of electricity in and the value out that has ever existed in an electricity system. That’s an amazing opportunity for electricity to get really cheap, because they can invest in these systems and grow the system, drive efficiency, and they’ve got loads of vested interest in making this system work.

I don't want to stand too hard for the data center industry, but if you look at the transcripts and the meeting minutes, the data centers are really coming into this with a lot of solutions, and they’re more than happy to put their money where their mouth is on most things, with some exceptions. I’m sure there are exceptions.

There are always exceptions, but they are meaningfully committed to this region. This is a region where loads of their workforce lives, and where many of the people in these meetings call home. They are currently going, “Right, well, either we figure this out or I’m moving to Texas.” Those are my options as an individual, but also the options for these companies and organizations. So they want to make PJM work, and it’s a real shame that PJM is not figuring this out.

7. Data Center Demand

Yeah. Maybe we could go back to that demand side. I think what likely happens here is that people see a high-level headline, or people who live in the region see their electricity bill and see the price go up, and then they immediately blame the demand side, as you’re describing. There are so many more data centers being brought online; therefore, there’s more demand for electricity. There’s some limit in terms of how much electricity can be brought into the grid, how much the grid can support, or how much power generation there can be. Therefore, more demand chasing fixed supply means increases in prices.

I think while that dynamic exists to an extent, we have made 2 cases in sequential articles. One is that PJM’s forecast for actual data center demand was being criticized because it was less accurate than our model, and now we’re also talking about the supply side being an issue and being criticized. Is there something beyond better modeling that they could be doing to actually upgrade the grid itself? Maybe you can just talk generally about what’s going on with electricity grids right now. What’s the technical stuff that everybody’s struggling with?

The key thing I’ll pick up on there is that you said “fixed supply.” If demand goes up with a fixed supply, then the price goes up, but there’s no reason why supply needs to be fixed. You can have increased supply, and theoretically, you should be bringing on new supply that’s more efficient and more modernized than old supply. We can get into the nuances there, but on the electricity system, I think the opportunity that’s being wasted here is basically the deployment of batteries. That’s probably a very big one. There’s a huge amount of opportunity for bringing down the price of electricity through better deployment of batteries.

Is there something in the transmission of electricity, in the substations, or in the transformers? Is there something to do with the actual power lines in the grid—the physical infrastructure—that prevents people from putting a power plant on one side of the grid, hundreds of miles away, and then the data center on the other side of the grid?

This is a slightly separate point, so I’m going to give you my high-level take on this. It might be something that I need to do further research on before we write about something similar.

PJM has been described as too big to function. One of the ways in which it is arguably too big to function is that it does these deliverability studies that work across a very large area. It’s basically saying, “Right, if we deliver a power plant here in Virginia, can it reach Illinois? Can it reach the other end of PJM? Is there enough space in those wires?” You arguably just shouldn’t do that. You should arguably try to have much more localized work. That way, you’ll be able to connect power plants and new demand to the grid much faster.

A lot of grids do this, where they basically just say, “Right, okay, we’re actually just doing operability areas. We’re saying that you only need to be able to supply your power to a large enough area that we think overall this is going to work out.” PJM is a lot of why its interconnection studies take a very long time, why connecting to the grid takes a long time, and why—this is what I’m not certain about, but I’ve seen some data that says—PJM is building more grid than it needs. People call it “copper plating,” where you basically just turn the entire area into wires, such that it’s copper-plated.

That is a potential criticism of PJM, which I’ve read elsewhere and am susceptible to because I know that it’s happening. You might be able to tell from my accent that I’m British. It’s something that we’ve done in Britain. In Britain, we’ve arguably significantly overbuilt our electricity system, for different reasons than PJM.

Building out these wires is expensive. Transformers have a very long lead time, and this kind of infrastructure is hard to build. There is some degree of that going on, but really, this is about better pairing up demand with supply. That can be done locationally, which is relevant to these wires, and it can also be done over time, where new demand needs to bring on new supply, or at least be linked to new supply. It doesn’t necessarily need to bring it to itself.

That’s what these capacity auctions are supposed to do. They’re supposed to help bring new supply online to meet new demand, but they’re just very bad at doing it.

Makes sense. Where should we go from here? Is there anything that we haven’t covered?

The emergency auction is the next thing. PJM models that it had, or will have, a 6.8-gigawatt shortfall in 2028–2029. It will be short of what it forecasts demand to be, relative to the supply it will have, by 6.8 gigawatts. This is partially because of the price caps, meaning that these auctions didn’t clear.

Frankly, even if there hadn’t been the price caps, I think the price caps were probably the right thing to do, given the other constraints. You have the 6.8-gigawatt shortfall, and we think that if you better took into account the cold-weather uplift and winterization, you would actually only need about 3 gigawatts. You would save more than half.

This is important, first, because it would save that money—it would save the roughly 3.8 GW required to procure it. Second, the way that PJM’s board, specifically the board, has chosen to design this emergency auction is actually against the wishes of its members.

PJM has these big, complicated votes. It has more than 1,000 members, split into 5 groups. Each group votes, and you need the equivalent of 66% to get anything through. So, surprise, surprise, a lot of the time things don’t get through. But they actually did agree on a different auction design that didn’t have this problem.

PJM’s board has decided that it is going to sign these contracts. These are long-term contracts going out to 2042, which is a good thing. Long-term contracts are good for new capacity, but they don’t necessarily have a counterparty.

PJM doesn’t have its own money. It just has ratepayer money; it passes these costs through to ratepayers. The theory is that PJM is going to pass this through to new demand load in the form of data centers, but it’s not a given that this new load appears.

It could contract 6.8 gigawatts of new load out to 2042–2043—roughly 11 to 15 years, depending on the start date—and then not have anyone to pay for it. It would have to go on existing ratepayers. If you’re a data center looking at this—and we go into this in depth in the article and in the newsletter—you really don’t know what you’re signing up for.

There’s a very high price cap of $555 per megawatt-day, which over the full 15 years could be as much as $21 billion for the amount they’re trying to contract. If you’re the data center, you don’t really know what you’re agreeing to. This is all still being decided. The auction is going to happen pretty soon, in September or October this year—basically, at the end of next month.

I think there’s a very real possibility that data centers don’t turn up and don’t take on these liabilities. Then PJM is stuck holding the bag, and PJM can’t hold the bag because it doesn’t have its own money. What’s actually going to happen is that the existing ratepayers are going to hold the bag.

8. Board Overrules Members

Okay. You were basically saying that there was a coalition making the case against how this was going to work, and they got overruled.

Yeah. They basically said that PJM just shouldn’t take on the liability and then pass it on. The data centers should just take it on directly, which I think is a much better system. Data centers should show up, sign on the dotted line, and say, “I need this amount.” Then they could either bring their own power through their own contracting, or you could even run an auction where the data centers have precommitted.

They might not be willing to precommit; that’s debatable. But the key thing here is that PJM’s board overrode this because it thought that the alternative supported by the membership wouldn’t procure enough power.

And so, if our modeling is correct and they don't need that much power, then they didn't need to overrule the membership. That's the thing it all comes back to. It's very important if you're centralizing.

Sorry, hold on. They have this vote, they get more than the required two-thirds, and then the board just overrules it and says, “No, you're wrong.” Why have the vote?

Yeah. Why have the vote?

I understand. I mean, it's quite something.

Not everything is good to be run as a democracy, right? Democracy doesn't work in all cases. But, yeah, it does.

Democracy until you do something we disagree with.

Something like that. Yeah. I mean, but then again, the board—there are plenty of times that I wish the board had overruled it. The real depths of this are in why this has been voted down in the past, why this was voted through this time but then shot down. You need a politics thesis and three psychiatrists to actually get to the bottom of the back-and-forth of what's gone on in PJM, which is not something that I go into in any great depth in the newsletter.

Plenty of people are more than welcome to look into that in their own time. Maybe we can loop around. You can grab your crystals from behind you, and then we can make some forecasts here. We can ask the oracle what the zodiac signs of the different board members are and—

Really, you get to understand it.

Is Mercury in retrograde while this vote was going on, or what? Yeah, honestly, it's the complexities of it: the motivated voting, the ability to veto, and the vested interests. A lot of the people voting are existing generators, right? They're currently getting windfalls off the back of this constrained supply and increased price.

I think a lot of them also want to build new power plants. That's not to say that they're all just sitting on their hands; it's a highly complicated system that PJM has built. My interpretation of “too big to function” was actually something said by the Federal Energy Regulatory Commission's chair. My interpretation of that was that it's too big to function in terms of its bureaucracy and its governance. It literally can't function, and it can't navigate its own systems of governance.

It's also just physically very large. But being a large grid shouldn't be a problem, right? Electricity grids benefit from scale. Being big is good if you're an electricity grid, so long as you're able to make the design choices and the governance choices that mean that works. It means that you actually benefit from those efficiencies of scale.

But, yeah, PJM is really struggling, and it's got a lot of political engagement now, with the governors' races coming up at the end of this year, in November. We've been digging right into data center moratoriums, and there's loads of chatter in Ohio, all over the place, about the political problems around this.

It is very frustrating that, again and again, PJM is the evidence for data centers leading to price increases. Surprise, surprise, if you suddenly start trying to explain that it's actually the nuances of market design, it's not as catchy.

Yeah, I mean, maybe not on the technical side, but I'm getting the sense that there are some potential investigative-journalism opportunities here, or some soap-opera-style drama that could be coming up if you're sitting in on some of these board votes and stuff. Maybe we can hook in some new viewers to the PJM board votes.

Yeah, I have a whole bunch of transcripts of some of these meetings, and they really are something. It does make me wonder what's going on behind the scenes. I'm sure—I hope—that there are some journalists slogging it out, trying to walk the corridors of these meetings, but sadly I doubt that there are. I don't think it sells copy.

It's a hard thing, right? Lots of electricity grids, or lots of electricity systems, have this problem where, for understandable reasons, they try to bring industry into the decision-making process. Then there's regulatory capture, right? It's classic regulatory capture, where you bring industry in. Industry is just much more motivated to actually understand and control the system than anyone else is.

The generation, transmission, and distribution companies—this is their entire job. This is their entire existence.

Yeah. This is right out—

And they immediately take over. In Britain, we have all these complicated systems called industry codes. That's how the system is run; they're the rules of the system. They're written by the players, by the people who are in the system.

Ostensibly, there's this overarching regulatory body that's supposed to mediate this, but they get run rings around. Of course they do, because these guys are just much more financially motivated to understand and guide the system than the regulators are. That's what's happening here as well.

The bit I really don't have a good grasp on, and I'm still trying to get my head around, is the deep politics of this. It really confuses me who is on whose side, who the players are, and when the board can overrule or when it can't overrule.

PJM really is just a stack of contracts. That's all it really is in terms of its actual existence. There are 3 contracts that all of the utilities and generators have signed, and these are unbelievably complex contracts. That's all that these things really are: agreements to operate under the terms of these contracts.

Because it's such a coalition, such a conglomeration of things, it's got utilities—it's got deregulated utilities, or utilities that have been vertically disintegrated as you separated out all the layers—but then it's also got vertically integrated utilities. It's got co-ops, transmission players, and independent power producers.

About 10% of the system doesn't even participate in this auction, right? About 10% of the system—vertically integrated utilities—just say, “Nah, we're good. We've covered our own demand. We've got our own supply. We're going to do our own thing.” Okay, off you go.

This thing is just infinitely complex, and I basically agree with the FERC chair that it's too big to function. You either need to significantly simplify its governance, which would probably fall foul of the U.S. Constitution, or you've got to somehow operate it in a more localized manner.

Yeah, I think there's a House of Cards- or Billions-style TV show here. If anybody from Netflix or HBO is listening and wants to bring in some actors to play the FERC chair and cover the politics—

Maybe it's more like Justified, where they're

wearing cowboy hats and threatening to shoot each other or something.

But it's such a funny thing—the fates are really laughing at us—that, to some meaningful degree, the future of Western civilization is riding on this governance cluster, right?

I wonder if there are news aggregators out there that we can pay that currently scrape everything ever written in America, so we can see what proportion of the electricity-price debate stems from PJM. If I'm willing to bet, I honestly think it's something like 40%. I think something like 40% of the first-tier debate is from this kind of stuff, and then arguably a lot of that 60% is just downstream of the 40%—a lot of the Facebook posts or tweets, et cetera.

It's unbelievably frustrating. You look at all the other grids, and this just isn't happening because this is a completely resolvable problem. Demand goes up, supply goes up; you have forecasts for that reason, and life goes on. Ideally, the price comes down because you find efficiencies, you have economies of scale, and you introduce a load of batteries and load flexibility.

9. Turning It Around

You make the data centers overpay. You literally just say, “Look, you have to pay more than your fair share because you're making absolute bank.” The data centers would be like, “Yeah, totally cool with that, because this way we actually get to get built rather than not get built.”

Yeah. Okay. Is there a way to turn this around and end this on a happy note? How about we say this: It is generally good that they're costing people money and making sure that there is no shortfall, because shortfalls are worse than having too much capacity and paying more money than necessary.

Yeah. I mean, the counterargument is you can't be too careful, right? It's fine to overpay and to have more capacity than you need because you can't be too careful.

Electricity is incredibly important. You should be able to bring on more load now. You should be able to fund these guys to build out new power plants that they want to build by giving them more money, all at the expense of the American taxpayer. I can give the negative case against the positive case, but I feel that positive case.

Try to end positive here, man. We got 40 minutes of negative already. All right. Unfortunately, we need to come up with more positive pieces on the utilities that we like. We obviously like RTO and PJM. They can still turn it around, right? They really can. These auctions are only every year; they only cover the year that they cover. So, the moment they time out, you stop paying those high prices, right? These are not necessarily baked in for very long periods of time.

The auctions currently last until partway through 2029. So, only 3 years—3 more years of higher rates—and then they can come back down again. Not too bad in the grand scheme of things.

If data centers can make their case, electricity load growth is by far the best way to bring down electricity prices. You can bring your own power in terms of behind-the-meter or backup power. You can use batteries for increased utilization. You can even do demand flexibility, either at the data center or with other people. I’m gently skeptical of demand flexibility at data centers, but there are definitely very impressive people working on that, so I hope they’re right.

Awesome. I really appreciate you coming on today. I enjoyed this time, learned a lot, and good job.

Thanks, man. You too.