For a long time, fossil gas was considered a bridge fuel. Even the gas industry called it a bridge, working hand in hand with environmental groups to push coal off the grid.
Then came the pushback over methane leaks, air quality in homes, and residential gas connections. The industry got so rattled it started hiring influencers to win back public opinion.
Well, all that has changed radically. Who needs influencers when you have the tech companies who run the platforms?
This month, Meta announced it would fund 10 gas power plants for a single AI campus in Louisiana totaling 7.5 gigawatts. Microsoft, Google, and Crusoe are all investing in gigawatts of new gas capacity. Utilities and independent power producers have tens of gigawatts more in their development pipelines.
Suddenly, this once-called bridge fuel is suddenly looking like a four-lane highway.
This week, we dig into what’s behind all these gas deals, what they mean for the power mix and emissions targets, and what an off-ramp could look like. We’ll look at the internal logic of the hyperscalers, the possible impacts on rates, and how the turbine crunch may impact development.
Credits: Co-hosted by Stephen Lacey, Jigar Shah, and Caroline Golin. Produced and edited by Stephen Lacey, Sean Marquand, and Anne Bailey.
Want to watch this episode? Subscribe to our YouTube channel.
Open Circuit is brought to you by FlexGen, a leader in integrated battery energy storage solutions and energy management software. FlexGen helps owners and operators gain greater visibility and control across complex energy systems to maximize performance. Learn more at www.flexgen.com.
Join Latitude Media on April 13-14, in San Francisco for Transition-AI 2026, a two-day, in-person conference on the digital and energy infrastructure buildout needed to support AI load growth. Our podcast listeners get a 10% discount on this year’s conference using the code PODS10. Register today here!
Transcript
Stephen Lacey: You guys, it’s a good morning.
Caroline Golin: Good morning.
Stephen Lacey: I finished my taxes and the president didn’t commit war crimes.
Jigar Shah: Well, the latter part requires some investigation.
Caroline Golin: I need to do my taxes.
Jigar Shah: But I appreciate the fact that now I can confidently pay $4 a gallon for the next nine months because Iran is taking a dollar a barrel of taxes for the rest of my life.
Stephen Lacey: Should we go take a trip to the Strait of Hormuz now?
Caroline Golin: You too can go. I’ll stay. I’ll stay.
Jigar Shah: I’m good. I heard they’re letting the cruise ships in now.
Caroline Golin: I’m good.
Stephen Lacey: From Latitude Media, this is Open Circuit. For a long time, natural gas has been considered a bridge fuel. Even the gas industry called it a bridge fuel. At one point, working with environmental groups to unseat coal. But then came a wave of pushback, concerns about methane leaks, air quality and homes, cities even passing laws to phase out gas connections, and suddenly the industry was on the defensive, even hiring influencers to improve its reputation.
Well, all that has changed radically. Who needs influencers when you have the tech companies who run the platforms? This month, Meta announced it would fund 10 natural gas plants for a single AI campus in Louisiana, enough capacity to power the entire state of South Dakota. Microsoft, Google, and Amazon are all investing in gigawatts of new gas capacity. Utilities and IPPs have tens of gigawatts more in their development pipeline, and this once called bridge fuel is suddenly looking like a four lane highway.
This week, we’re going to talk about what’s behind all these gas deals, what it means for the power mix and emissions targets, and what an off ramp could look like. The dash to gas is coming right up.
Hey everybody, welcome to the show. I am Stephen Lacey, the executive editor at Latitude Media. Caroline Golin is back with us this week from Charleston, South Carolina. How are you?
Caroline Golin: I’m wonderful. It’s beautiful here. And the beach is right around the corner, so I can’t be in a better mood.
Stephen Lacey: Still recovery from CeraWeek?
Caroline Golin: Never recovered. I think CERAWeek is like having children because you just don’t sleep and then so you forget how strenuous it is so you decide to do it again and again and again. Yeah, that’s how I feel.
Jigar Shah: I do not have any firsthand knowledge of this analogy, but I will take your word for it.
Stephen Lacey: Caroline is the chief growth officer for NRG and Jigar Shah is the co-managing partner at Multiplier and host of Energy Empire, along with co-host of Open Circuit. How are you, sir?
Jigar Shah: I’m great. I am back in the home office, thank goodness.
Stephen Lacey: You ready for Transition AI? You ready to throw down?
Jigar Shah: I need to work my way up to it. I think I’m going to need to psych myself up.
Stephen Lacey: Well, that’s coming up right next week. So this is your last chance to get a ticket. Go to last weekmedia.com/events.
Caroline Golin: And if that overwhelming advertisement from Jigar didn’t do it for you.
Stephen Lacey: Psych yourself up with Jigar and you can use the code PODS10 for a discount. So come join us. We’re going to have a great show and we’ll be doing a live Open Circuit there.
So let’s get into the show. We’re going to talk about gas this week. We’ve seen a slew of gas plants planned by the hyperscalers in the last few weeks. And what’s striking about this, the reason why we’re talking about this now is because it’s not just like one company making a big bet. It’s all of them almost simultaneously. Meta tripled its gas commitment for Louisiana in under a year going from 2.3 gigawatts to 7.5 gigawatts across 10 plants at a cost of nearly $11 billion. Microsoft entered exclusive talks with Chevron and Engine No. 1 to build a natural gas plant in the heart of the Texas Permian Basin. That’ll start at 2,500 megawatts at a projected cost of $7 billion with the potential to expand to 5,000 megawatts.
And fun fact about that, Engine No. 1 was the investment group that ran the activist campaign against Exxon years back. And at the Goodnight campus in Texas, a project Google has invested in alongside Crusoe Energy. There’s a permit filed for almost a gigawatt of gas that would operate behind the meter. And speaking of Crusoe, this is a company that once declared plans to build a climate aligned cloud, but is now primarily investing in gas. And I think I saw that it secured about 4.5 gigawatts of gas equipment for gas generation.
So let’s walk through some of these examples first and then I want to just step back and talk about what it means collectively. Caroline, first to the Louisiana deal for Meta, this isn’t structured like a normal PPA. It’s financing generation, transmission and storage for a single campus. Gas is by far the dominant resource. What do you see in this project?
Caroline Golin: I mean, to be honest, I see this project as very similar to the way most of the hyperscalers went to market for a very long time. I mean, I think what is different here is that we didn’t eat up or Meta didn’t eat up existing capacity on the system. But historically, for the most part, hyperscalers, we go to market, we take the basic C&I tariff, right? So you would fly in, you’re going to build 500, maybe 600 megawatts max, and you would just fall under a traditional wholesale agreement with the utility, depending on where you were operating. Or you would go and just take the C&I tariff. And then over the course of the first couple of years, because it takes five, six years to actually build a data center and ramp into it, that was the old way. You try to work on what your climate strategy was with that.
And maybe that would be to build a C&I solar program. Maybe that would be to purchase solar and win someplace else if you didn’t have a locational matching program like Google and Microsoft did. Or you would do nothing and you would put it in the hopper of, when we can find cheap offsets, we’ll pursue cheap offsets. It’s not that different. What’s different is it required new construction, new power plants being built in order to service load. And so I think what Meta did was they said, “Whatever the utility thinks is best is what will get built and that’s what we’ll take.” And that’s a pretty standard response in most of these systems. It’s just what’s shocking is how much new had to be built because you couldn’t eat up existing capacity on the grid. But in terms of the actual mechanics, it doesn’t really look that different to me than what many did before.
Stephen Lacey: Jigar, how do you read seven and a half gigawatts of gas in Louisiana?
Jigar Shah: Well, I mean, it’s not shocking, right? I mean, Entergy has always loved gas. And so I think that that is how you would expect to get powered if you were in Louisiana. I think the bigger challenge for me is trying to make sense of the number. I don’t understand how energy has magically figured out how to get really high quality gas turbines at half the price of what everyone else is paying for gas turbines. And so that price is not $2,800 a KW. That price is lower than $2800 a KW. Now in Southern company’s case, they had bought the turbines in 2021, right? And so it was fine. They got it at a cheaper price when gas turbines were $800 a KW and there were no orders in the books. But in Entergy’s case, there’s no chance that they bought these gas turbines in 2021 because they were not expecting seven and a half gigawatts of load growth, right?
And so I’m trying to figure out whether there’s a repricing coming later when they actually figure out what the actual costs are that they’re going to have to talk to the commission about.
Caroline Golin: Yeah. I mean, I haven’t looked at the exact filing, but I’m sure there’s a band on whatever their CWIP is going to be on this and it’s pretty high.
Jigar Shah: It’s a heavy metal band in this case.
Stephen Lacey: Well, so Caroline, you described this as not that different from what data center developers, hyperscalers did in the past. Just in this case, you had to build a lot new generation. I guess someone who isn’t in the power market world or utility planning world might say like, “Why do they need that much gas? Why don’t they build other things? Why do we have to rely on this much gas?”
Caroline Golin: Yeah. I mean, so I’ll step back and say, I think the biggest difference with this Louisiana deal was actually that it was done through a third party developer, right? This wasn’t a Meta built. I think Meta had procured the land and then hired a third party to raise the capital and build the shell, but the energy deal stayed with Meta. And so the difference there is that the team had to decide what was their agenda. And I think their agenda was probably to build as quickly as possibly and to build with the least amount of friction and the most amount of security. And from Entergy’s point of view, Entergy is going to say building gas is going to be the way that we get you on the system fast enough. And so you can point the finger at Meta and say, “Why did Meta have to build all this?”
Or you can point the finger at Entergy and say, “Why did Entergy have to build all this?” Or you can point the finger at the regulators and say, “Why did the regulators approve to build all this?” I think all of them essentially came to the collective conclusion that it was the easiest path with least resistance to getting electrons on the system. And it’s what Energy knew how to do and I don’t think that Meta was going to push to do something new. And that’s just, I mean, that’s the boring story of it, but that’s the story of it.
Jigar Shah: Well, there’s more to the story though, right? Which is that once these things start happening, remember Louisiana has been historically against all renewable energy, right? So they’ve built maybe 1500 megawatts of solar in the last 10 years. They’re planning though of building 10,000 megawatts of new solar over the next 10 years, right? So they’re building a lot more renewable energy. And part of that is because it’s cheap and then they’ve been behind. And so there’s actually a lot of great sites to put solar down. And I think one of the big challenges I see is that this is leading to the least optimized grid you could possibly build. And so one of the big challenges I see is I don’t think that this ends with a one-time deal with Meta. I think this is going to be, we’re going to put in a one-time deal.
Then there’s going to be a whole bunch of other people who build solar and build other things. And then they’re going to like, “Well, we had to back the gas off. And so now the gas plants are costing us a lot more money on the capacity side just to keep them available. And we’re only running them 35% of the time instead of 57% of the time like we thought.” And so now we have to raise rates another 10% to like cover those costs. And then something else is going to happen. And along the way, you and I both know Meta’s not going to cover all of those costs, right? They are going to be peanut buttered across all of the different rate classes and they’re going to say, “Well, this wasn’t Meta’s fault. Meta did the first rate case and now all of these other costs are going to be spread across all the poor people in Louisiana.” And so part of my challenge with this is that I’m not taking an anti-gas approach.
My thing is, is that one, the cost of the turbines right now everyone knows are in the 2000s per kilowatt. That is not the number that they reported and I don’t know how Entergy got a good deal. Two, I don’t think Louisiana is being honest about what the five-year trajectory is for the costs of the full integration of all these things into their grid, right? And so I think that there’s a lot more shoes to drop here. And at that point they’re going to be like, “Well, the turbines have already been ordered, so we can’t change our mind now.” And like all this stuff is already in like … So they’re not giving people enough information to actually have a robust conversation at the regulator.
Caroline Golin: Well, I think if you have extra turbines, you’ll find a buyer.
Jigar Shah: How many extra turbines do you think Entergy had?
Caroline Golin: I don’t know how many extra turbines are in the entire country right now. I think they’re mostly accounted for. But I think what Jigar’s getting at is that in order to do this sort of responsibly, you have to integrate this with a more robust understanding of where your grid is going and where load is going. And if load doesn’t materialize in any circumstance, if load doesn’t materialize and grow, then everyone’s at risk of that particular situation. And that’s why it’s important that the right tariff structures be put in place and the right guardrails be put in place and then energy be held accountable to that.
Stephen Lacey: This is a 15 year agreement, isn’t it?
Caroline Golin: Yeah. Yeah, it is. I mean, I think the bigger concern that Jigar’s articulating really for me only falls down as if Meta decides not to continue growing there, but I don’t see why they wouldn’t do that. I mean, most entities, once they have enough land and they have enough space, they’re going to continue to grow out.
Jigar Shah: No, but I’m not saying that, Caroline. I’m saying that like right now Entergy is in a battle with the city of New Orleans because they didn’t want to do virtual power plants. Now the city of New Orleans is making them do virtual power plants, right? You’ve got companies like Carrier that are now deploying air conditioning systems with four kilowatt batteries associated with them, right? You’ve got all of this innovation happening around them that is most certainly going to lead to them not running the natural gas plants as much as they plan to in the base case. When that happens, that by definition raises the cost per megawatt hour from each megawatt hour that does get produced by that natural gas plant and entergy is going to have to file a new rate case when they figure that out. This is stuff that they could know now, but they refuse to be honest about now.
And so like because they’ve got their head in the sand and they think that they can just keep VPPs and edge technologies and other things like off the radar screen and not into Louisiana. And I was like, “That’s not possible. This stuff has come into all 50 states, electric vehicles, everything is. ” And so when that happens, these gas plants are going to run at a lower capacity factor.
Caroline Golin: They may run at a lower capacity factor, or if Meta continues to grow, they’ll just continue to ring fence those costs and sleeve it directly on their tariff. I mean, it could go either way, right?
Jigar Shah: I mean you’ve been to Louisiana. You know that what they do is hurt poor people on a regular basis. They don’t actually care about putting the costs onto Exxon or putting the costs onto the fertilizer company or whatever. I mean, come on.
Caroline Golin: They signed that pledge, right? They’re responsible for it now.
Jigar Shah: Oh did they? Did they sign a pledge. Oh my God. There’s a reason why environmental justice was basically invented in Louisiana.
Caroline Golin: That’s true. That’s true. No, but I think what is missing in this conversation, which I agree with Jigar is that we have talked about natural gas as the firming technology, as the bridge to renewables, right? But really the conversation in my mind needs to pivot a little bit or expand a little bit to say, what is the role of natural gas vis-a-vis virtual power and a flexibility conversation? And that really transcends all the market structures we’re talking about. We’re so focused on PJM and natural gas on a capacity product, I mean, for a capacity signal and capacity costs in that market. But what we really need to look at is over the next 10 years, everything’s going to get more micro, right? I mean, this is what Jigar’s talking about, like behind the meter solutions for customers and load management, smart thermostats, storage, electric vehicles, everything.
Everything’s going to get more micro. And so in those micro markets, the signal you actually need is flexibility. And this is sort of what I think the UK went through at some point, and then we come out on the other side. But if you have a flexibility signal, certain type of gas and the way gas is built and the way you expect it to run can be in the money or out of the money. If you expect it to run at a 90% capacity factor the entire time, you could risk it being out of the money, right? If we’re moving into what’s going to be more micro market, more flexible space. And so I think the conversation needs to be less about how does gas work to firm the duct curve essentially of solar, as opposed to saying, how does gas work in a highly micro, highly flexible grid?
And then how do you build it according to that and match it with a better portfolio around load management and demand response? I don’t think that’s what happened in Louisiana, but that’s what I hope is happening moving forward.
Jigar Shah: Well the reason it bothers me so much is it did happen in Georgia Power’s territory. Georgia Power did put in a ton of batteries. They went from 600 megawatts to 6.6 gigawatts of batteries in the last IRP. Duke is doing the same, although on like a much longer timeframe, although I think we’ll see it shorten. And so the fact that this isn’t a dominant battery story, and instead it’s a dominant gas story, like speaks volumes around how backward energy is.
Caroline Golin: Well, I think the battery story actually is more about transmission. And I can’t be positive because I haven’t looked at and compared the two, but what I’ve seen trending is where utilities are very comfortable in investing in storage is really more around transmission as a transmission asset. And there’s sort of two things that happen. Either you have enough generation, but you don’t have the transmission capacity on your system because you haven’t built out your system to be robust for massive chunks of load growth. That’s what happened all throughout AEP system or you’re in a situation where you don’t have enough generation, but you have the transmission capacity, right? And if you don’t have enough generation, the low hanging fruit, the easiest thing to do is like, okay, we’re just going to build a gas plant. But if you don’t have both, which I know transmission has been a struggle in Georgia, just the way it’s been built in terms of where demand is and where supply is constructed, then storage becomes a very easy component of it.
But I don’t think storage is still being seen as a generation or even … And just to add one more thing on that, all of these data centers, if they’re AI training, are going to have to have storage behind the meter.
And I think this is the one thing we keep missing, which is that even in all these co-located scenarios, the natural sort of load shape of a training center is going to burn through any CC in two seconds, because it’s just so erratic, right? So you’re going to sort of see that micro need for storage anyways. So that’s what’s a little more shocking to me, Jigar, is that that wasn’t part of the initial plan because it’s a massive data center, which I assume is going to have training and I assume is going to need a significant amount of storage. Maybe Meta’s doing that on their own. I don’t know.
Stephen Lacey: And so is this Meta just being lazy?
Jigar Shah: Never. Bite your tongue, Stephen. Meta is known as the most progressive, best company in the entire hyperscaler universe, second only to Oracle.
Caroline Golin: Well, here’s the truth. Here’s what’s interesting is that this deal was kind of revolutionary on the shell build outside. This deal and the way that they raised the capital, the way they structured the capital was sort of first of its kind when it came to how you use the existing markets and to rapidly build out a shell. So what’s interesting to me is that if you take this single case study on one side, it’s being praised all throughout the data center development space, but then on the flip side, it’s being attacked for how sort of seemingly run of the mill it is on the energy side.
Stephen Lacey: Wait, so what did they do differently on the shell side? How did they build it?
Jigar Shah: They used Blue Owl or whatever.
Caroline Golin: The used Blue Owl and they raised, I think, probably $30 billion in less than a month because they took a historical approach, which is that they went and looked, which is a really smart way to do it. They just went and looked at the way ports had been financed, the way smelters had been financed and said, okay, well, if we make a data center look like that in terms of deliverability payments and milestones and the capital requirements to fund it, we could probably raise capital quicker, right? And they were right. And so they sort of standardized the way they were doing that. And standardized is probably generous. They created a model that could be replicated if picked up that way.
Stephen Lacey: What do you guys think about Microsoft’s push in the Permian Basin using stranded gas for generation?
Jigar Shah: Well, I felt like this deal was just waiting to be announced. I mean, Chevron had announced that they had jumped the queue with GE Vernova last year for seven turbines, right? And so we were all waiting to figure out who’s going to get those seven turbines. And so, I mean, I don’t know if you know anything different, Caroline, but I think everyone was just competing to figure out who was going to get the seven turbines.
Caroline Golin: Yeah. I don’t know anything different. I think that’s probably right. I mean, I think there’s an opportunity here within the Permian to sort of the tech companies to come together and sort of say, “Hey, let’s do this more responsibly.” And actually, I think Chevron’s probably a major proponent of that. And if they’ve sort of pioneered thinking about responsible natural gas or renewable natural gas from the supply chain, which I think has legitimate impacts.
Stephen Lacey: Wait, what? Chevron has been responsible in renewable natural gas?
Caroline Golin: Chevron’s been … Yeah, they’ve been one of the early leaders and just trying to get a market for it. I mean, in the same way they were testing, “Well, if we clean this up, is there monetary value in cleaning this up?”
Jigar Shah: So two different areas, right? Renewable natural gas on one side and then the low leak methane pledge on the other side. On the other side. Yeah. Chevron’s certainly done a lot in that area. And we did that loan when I was at the Loan Programs Office to long path technologies of which Chevron was one of their bigger customers and saved like 1 BCF of gas leaks or something in just a short amount of time.
Caroline Golin: Yeah. So I mean, the silver lining on a lot of this is that everyone … I know there’s all this like, oh, tech companies are now running on gas. We were always running on gas. We were always running on gas. What we were trying to do was create, to use the word bridge again, a bridge to where there was more than just gas as our opportunity to power our data centers. And in Google in particular, wanted to completely decarbonize, but gas was … We were always generating gas to serve these data centers or generating electrons on gas to serve the data centers. And because we didn’t want to talk about it, we actually didn’t put our weight behind cleaning up anything in the gas industry. So now my question is, okay, if we’re being more open and honest about this, are we going to be held accountable to put our weight behind cleaning up?
Things that are kind of low hanging fruit have significant environmental impacts on cleaning up on the gas side.
Jigar Shah: Yeah. All the gas deals we did at the Loan Programs Office, we required them to sign low leak methane procurement. I think … I mean, the Permian is weird to me because the Permian is 80% wind and solar, right? And so, and the Waha pricing is negative and has been for months, right? So because there’s not enough pipeline capacity to actually get the gas out of the Permian. And so gas is trading at a negative price. You have to pay people to take the gas in the Permian because there’s not a pipeline capacity. So I’m assuming Chevron’s going to build more pipeline capacity too, to be able to like get Permian gas to…
Caroline Golin: I think so. I mean, we can talk about all of this, but then if you actually look at permits and how quickly you can build, I think that shrinks, it shrinks what’s available. Yeah.
Stephen Lacey: To your point about the tech companies always relying on gas, a lot of people were surprised when Michael Thomas at Cleanview issued this report showing that Google was investing or partnering alongside Crusoe for this 933 megawatt natural gas plant behind the meter. And Google’s obviously the poster child of 24/7 carbon-free energy and doing things the right way. Was it surprising to you that there was gas in this data center or just Google has always had more gas than people realize?
Caroline Golin: Well, I mean, anyone who understands the way that a grid works, unless we had our own pun and were completely off the system, our stability was based on whatever was creating stability on the grid, and in large part that was gas. However, I think Google was the most aggressive and I think still remains the most aggressive in saying in like, let’s take this Louisiana scenario, pushing back and saying, “No, there needs to be a more balanced portfolio here.” Okay, we understand if we have to build seven gigawatts if maybe two, three, four need to be natural gas, but I think Google would have been the ones to push and say, “It’s not all going to be natural gas.” And I think that’s really what the point was the entire time. And I take more responsibility for this than anyone because I took that mantle and I ran with it at Google, which was to say it is possible, therefore we shouldn’t accept anything less, but I don’t think anyone at Google truly expected that others would be able to replicate everything that we did.
And so that’s where the gap widened. And the truth is, Google is not going to be able, the energy team is not going to be able to go to leadership and say, “Well, Amazon is going to kick our butts because we decided we couldn’t run on natural gas.” That conversation is never going to be allowed to happen internally. So as hard as they are trying, and I think they are trying more than anyone, I actually think that they also recognize that well, responsibly built natural gas plants that provide that stability, allow for more renewables to come on or integrated flexibly with virtual power is a good future for the system. What’s not a good future is when it’s sort of built behind the meter irresponsibly, and we’ve talked about this before, a ton of recipes, no load planning, no entity that has the desk to be able to balance it.
That’s an irresponsible way to do it, and that’s going to end up in stranded assets, for sure, in my mind, right? So there’s a responsible way, an irresponsible way to do this is all I’m saying.
Jigar Shah: The thing I don’t understand about this, so the energy piece I get, and I’ve dealt with Entergy my whole life, they are who they are. I think on the Texas side though, what I don’t understand is that Texas has all of these resources getting added to the grid on a regular basis, and the natural gas gets dispatched under the merit order, and so it only gets dispatched when wholesale market prices get to a certain And price, right? And so now you may have to dispatch it in the Permian just because there might be certain stability issues or the things that you need to compensate for. But my sense is that in Texas, a lot of this gas is actually going in as capacity that’s going to run whatever it is, 300, 400, 500, 600 hours a year, that it’s not going to be base load per se because there’s no need for it. There’s actually plenty of power in Texas 90% of the time.
Caroline Golin: Well, the reality is the ERCOT queue is 10 times the existing system.
Jigar Shah: I mean, that’s all fake. Let’s not kid ourselves. I love an ERCOT cube like the next person, but the notion that they’re going to increase peak demand above 120 gigawatts, let’s say, by 2030 or 2031 is sort of fanciful thinking. And so they’re at like 70 some or 80 gigawatts now. And so that’s a lot, by the way. That’s 50% increase from 80 to 120, but they’re not going to be 250 gigawatts of additional loan.
Caroline Golin: Even if 25% shows up, they’re going to need to build.
Jigar Shah: 25% is not going to show up. I will bet you right now that it will not exceed 40 gigawatts of peak load of growth from today.
Caroline Golin: What do I win if you lose that bet?
Jigar Shah: I mean, you’re always a winner. So I’m not worried about that. But I think people are just mad right now. When you think about the 2026 numbers and how of all the projects that hit final investment decision in 2026, 50% of them are still not under construction. And so when you look at the data, they are not actually building in any way, shape or form based on all of the ridiculousness that I have to read on Twitter and these stupid reports and whatever else. And I only have to believe that they are purposely obfuscating the data. It does not seem impossible to speak clearly about which projects have received funding, which projects are moving forward. And to Caroline’s point on four different podcasts, there’s not a single person building a merchant natural gas plant, not one person. And so then if you’re not building a merchant power plant, then there is a buyer, there is a seller, there’s an equipment provider, there’s an EPC contractor that got a contract. This should be something that you can be precise about.
Stephen Lacey: Yes. And so why does that matter? Why is that so important that we know exactly?
Jigar Shah: So nobody has ever managed the load queue. So the load queue is basically a made up number. It’s just whatever. A bunch of people who’ve said, “I plan to build something, I’m going to put it in.” Right now recently folks have said you have to put up a bond. You can’t just put random stuff into a load queue. And then magically like 70% of the stuff goes away because some guy said, Oh, I got stuck with this land that my daddy gave me. I thought I could make a quick buck on it with data centers now I’m not going to spend $20 million to be in a load queue. Fine. But then on top of that, you still have all of the physical constraints of the grid. And so when you go into a public service commission, you and I both know it’s hard enough to educate those public service commissioners on a good day.
And then when there’s all of this just ridiculous data just like being thrown up into their faces and saying, “Look at this data, look at that data, look at this data without any sort of screen around like, these are real projects, these are not real projects. Here’s data that you should look at. Here’s data you should not look at because it’s purposely meant to just like make some sort of Twitter meme.” I think that like when you look at Michael Thomas’s report, like Carolyn’s been saying Six Ways to Sunday, I talked to Jeff Bladen from Verrus yesterday, they’re like, “There’s no chance any of those are being built.” There’s like 56 or 67 or whatever gigawatts of projects. And so there should be, when you’re spending a trillion dollars on infrastructure or whatever the number is that people make up in the Wall Street Journal today, like someone should spend money to figure out what’s truth and what’s fiction.
Caroline Golin: Okay. I’m going to break this down. I feel like I’ve broken this down before on this podcast, but- Break it down. Tech companies, hyperscalers, whatever you want to call them.
Jigar Shah: What did you call them before? You had some weird fancy name you tried to sell.
Caroline Golin: No, we just always called ourselves the tech companies.
Jigar Shah: No, no. There was something else in a previous episode. What was it called?
Caroline Golin: Divine beings. I don’t know. So the tech companies have at most a four year planning cycle at most. When I left Google, we had two and I think we were trying to graduate to a five year planning cycle. So even if you were to go to each of these tech companies and say,” What is your actual load growth? “They may know for the next two years and then have, they could be like, ” And then the P90 on this, who knows? “Right? And I think what ends up happening in a lot of these circumstances is this is just sort of extrapolated based on some statistical number around where we’ve seen the stairstep in demand. And the stair step in demand is not reflective of anything other than a catchall approach, right? Everyone’s just going out and trying to get as much as they can right now because they functionally do not know are they going to get another hockey stick in load growth in two years or is it going to flatten or is it going to come … It’s probably not going to drop off.
So worst case scenario is that they’re left with something they have to sit on for three, four years as they grow into that training load or they grow into that inference load. And that three or four years, it’s not a rounding error, but it’s not dramatically significant into the cost of not having enough to train. So on the flip side though, power plants are built over what, a five year, six year scheduled timeline, and those two don’t connect at all.
Jigar Shah: Right. But then on the other side, we’re saying nobody should blame data centers for rate increases because it’s not data center’s fault. And in fact, all this load growth is actually reducing costs for everybody and the people who build data centers have lower bills, right? But that’s clearly untrue. If you look at PSE&G’s rate filings in New Jersey, they’ve gone from $18 billion over the next five years in 2023, $22 billion in 24, like $26 billion in 25. Now they’re going to come out with $30 billion over the next five years, right? Duke, same thing. And then when you read their transcript from their thing, they’re like, well, we’re building all these new natural gas plants because of data center load growth. And so then on this side, they’re saying, I’m raising your rates 9% a year because of data center load growth and also the load growth that we’re facing and the natural gas plants that we have to pay $2,400 a KW for, right?
Because we’re talking about gas now, right? And in that case, the utility doesn’t care about an off-take agreement because the utility is saying,” I can rate-based stuff because I like to rate-based stuff and I have old gas plants I built in 2004 that I’d like to replace. And this is a convenient way for me to shove this right in. And so the data center companies by deliberately being terrible at planning are leading to higher rates for everybody and like all of this speculation on gas by these utility companies who expect to have data centers in five or six or seven year planning cycles, which the data center companies themselves have not committed to, but I am having to deal with all this stuff in public service commission filings.
Stephen Lacey: Caroline, do you think that there could be a fundamental change in the way that tech companies plan? I mean, there’s such a frenzied element to this race. Do you see there being another way?
Caroline Golin: I mean, to some degree, I think it will amend itself within the next couple of years, right? In what way? Because training, everyone who has met their training goals will either meet them or they’ll be consolidated. And then you shift to sort of what is a long-term projection on training for product iteration and what moves back to sort of an inference. An inference is much more customer based, right?
So in some ways I think it will amend itself. So help me understand what that — Jigar, what you were saying is actually interesting, and I think this is something we haven’t talked about, but it’s something we need to point to, which is that in a vertically integrated market, the IOU can absorb the shock with its balance sheet and with the way it can raise capital because it doesn’t have the same investor scrutiny than compared to a competitive supplier. And that ability to absorb it is one of the reasons why hyperscalers and data center developers will go within IOU because the sticker price, they’re not paying for the capital costs of building out these plants. They’re just paying the energy costs for the generation. Everything else is absorbed into the rate base. Now-
Jigar Shah: Are you calling President Trump a liar?
Caroline Golin: No, and stop putting words in my mouth. Okay. But the second piece is, if you do this bilaterally in a competitive market, there is more risk on the hyperscaler. There is more money and skin on the table, and so they will have to evolve into the muscles of what it takes to build a power plant and what it actually costs to build a power plant. And so what’ll either happen is they’ll develop those muscles and in those situations, in those bilateral situations, absolutely I think you can control easily for the rate impacts being shifted onto the residential customer or the commercial customer. And I will say in the vertically integrated space, you can as well, but it requires some articulate math, right?
Which the utility may choose to do or not do. And so I think that there’s a learning curve here, and if we go back to the Meta example, the tech companies, none of us are … We weren’t hired because we knew how to build power plants. Most of the commercial teams and these tech companies were built because they knew how to do renewable finance development. They knew how to run large RFPs. They knew how to do these bilateral contracts. They knew how to negotiate. They knew how to understand a hedge. That is a very different skillset than saying, we’re going to build a new CC and deal with the interconnection of that, the pipeline risks, and then balance that against the overall wholesale market. Totally different skillsets. And so there’s an additional learning curve in here and what I don’t want to see is build out, get pushed completely into the IOU territories where it does get sort of lazily absolbed and then you have a residential cost system or a cost increase problem because the tech companies just, they don’t know how to do it yet.
So maybe they’ll learn. They don’t know how to do it yet. And it’s the easier way out to be like, Well, let’s just let the utility rate base cover what really should be our cost responsibility.
Stephen Lacey: To pull this back a bit and to maybe touch again on what Jigar’s point was about the lack of transparency, I mean, when we look across all these new developments from Meta, Microsoft, Google, Crusoe, et cetera, all making major gas commitments simultaneously, what is the aggregate signal that this is sending?
Caroline Golin: I don’t know the answer to this question straight out. Maybe one of our listeners could do the math and send it back to you, Stephen, but if you look, if we could actually take a look or someone, maybe Jigar, I don’t know if someone has done this, how much additional capacity was in the system that was actually gas five years ago?
Jigar Shah: It was very small.
Caroline Golin: Very small.
Jigar Shah: We have about 500 gigawatts of operating gas in the United States. That number was basically designed to be flat, right? So the vast majority of those gas plants were built pre 2004, right? And so gas plants generally last 20 years, right? So when you think about the queue, right? So like, for instance, if Chevron were to take one of these turbines that they’re building in the Permian and they actually ran that data center off grid and then they blew a shaft on that turbine, it would take five years to get a shaft.
They basically would be in the same queue as people who want to buy a new CCGT, right? So basically that was what Southern Company was buying gas plants for in 2021. All of these folks were basically buying stuff to replace stuff that was retiring in 2004, right? And so now, if you have growth from that number, then that growth number comes from lower heat rate turbines, probably like recipes or like recycled aircraft engines or whatever like startup company that just like decided to take their airplane business and turn it into a power plant business, whatever goddamn thing that like I’m having to force to read on Twitter.
Caroline Golin: You know, you can just not read Twitter.
Jigar Shah: The problem is that Stephen Lacey asks me questions from people who are clearly not credible.
Stephen Lacey: Wait, on Twitter or in this podcast?
Jigar Shah: On this podcast. I just think that anything that doesn’t have a high heat rate is not going to run. And I think everyone who knows anything about this stuff knows that, but we act like it’s going to run. When we talk about stuff, we say, well, there’s a hundred gigawatts worth of gas plants that have been purchased, right? How much of those 100 gigawatts are high heat rate gas plants? Like that is not provided in the question, right? It’s just an aggregate number and then Caterpillar stock is doubled. It has real world consequences, even though no one’s going to use those damn solar turbines.
Stephen Lacey: I mean, and that’s why I’m asking the question, right? A lot of people are looking at this from the outside in, just seeing these announcements and trying to figure it out and they’re asking the same question, which is like, what is the aggregate signal here? And so I don’t know who you’re referring to about what…
Caroline Golin: I think the argument signal is just that there’s more load growth than we plan for and we’re playing catch up and we’re using a technology that we know how to deploy. I mean, I don’t want to oversimplify this, but like we, I’ll just use Google because that’s the one I know. We set our 24/7 goal at the end of 2020. By the end of 2023, we realized we were an order of several magnitudes off on what we thought our load projections were going to be. So we had two and a half years to scale technologies to be commercially viable to meet our load. That’s not enough time when you think about the hockey stick of what happened. And so what’s happening is maybe if we’d had five more years, there would be less hypothetical gas in the system because you would have scaled other technologies like geothermal or maybe even some degree SMRs, or more likely we would have figured out how to do load management better and commoditize that, which I think is still the silver bullet, but that’s two and a half years that it’s not a ton of time to see a sea change in the way the vast majority of this industry feels confident and willing to take deliverability risk and willing to take reliability risk in a market. And I think that’s what it’s saying.
Jigar Shah: Let me say it slightly differently. So if I were to say to you that Generac or Kohler or any one of these companies had a banner year and they sold a gargantuan number of backup diesel generators or PSI piston natural gas generators, right? Like behind the meter at a hospital or a nursing home, whatever, would you include that in the 100 gigawatt number? Of course you wouldn’t, but that is what you’re doing. Every one of these turbines are basically backup generators, right? And they’re designed to run 50 hours, 60 hours, 80 hours a year, right? But like they’re included in the number with CCGTs and other things, which is not how we defined the space in the past. People have been buying backup diesel generators for a long time. That doesn’t mean I’m predicting that 8% of all US electricity is coming from diesel, right?
But now we’re saying like 43% of all electricity from the United States comes from natural gas. Are you saying that that number’s going to go to 60? If you are, I’m telling you that I know which 40 utility CEOs will lose their job by 2030, because that is dumb. Nobody in their right mind would have an energy mix that’s 60% natural gas.
Stephen Lacey: Just to bring in some numbers here, I think you’re referring to the global energy monitor numbers showing that the pipeline for gas now stands at nearly a hundred gigawatts. And we had just talked about that before we hit record, but I don’t think we had brought it into the conversation, but are those the numbers you’re referring to?
Jigar Shah: Yeah, but I mean, unfortunately the numbers are random. Michael Thomas’s numbers are like 57 gigawatts of like behind the meter, natural gas and then this guy’s numbers are this. All I’m saying is that what matters from a climate standpoint is that we go from, if the original question that we posed during this podcast is whether natural gas is a bridge to somewhere, right? Then that means that you go from, you go up to 43% because natural gas wasn’t like 25% of our mix, right? And then they basically destroyed a lot of the coal production from the 2004, 2005 timeframe, right? Now, the question is, where’s it going to go? And is it going to go up a little bit? Probably, right? Is it going to go to 60? I doubt it. Is it going to go back down in the future when? That’s sort of what a bridge conversation looks like, right?
Stephen Lacey: Well, let’s look at like equipment shortages and costs. The equipment to build all this is in short supply as we’ve been talking about. Wood Mackenzie came out with a recent report showing that global turbine orders exceed manufacturing capacity by a huge margin. Lead times are now six years, order books are sold through 2027, prices are up 195% since 2019. It doesn’t seem like any of that has slowed gas development. I mean, are people finding their way around these challenges?
Caroline Golin: I think that’s more of a timing issue. I think what you’re using applications and projections for demand as one guidepost, but those are all ambitious. And then you actually have to go build a power plant, and that’s where things get hard. And that’s where a lot of players in this industry … I mean, let’s be honest, most utilities in this industry, in this country, have not built a power plant in a decade. They don’t have the workforce, they don’t have the desk. They probably still have a lot of the contracts in place, but then they’re going to come up with how are we actually going to do this? And that’s where rubber is going to meet the road. But no, the workforce available workforce does not match the eyes or the ambition for growth and neither does the supply chain. And so what’s going to happen is this is going to consolidated to a handful of players who actually know how to build these things in the same way that you can say that’s what’s going to happen on the data center side.
It’s going to consolidate into a handful of players that actually know how to build these things. And you see that happening all over Texas right now.
Jigar Shah: Yeah. And the prices have gone way up, right? If you hire Kiwit or Bechtel or somebody to build a gas plant, they’re charging you three times more than they did 10 years ago.
Caroline Golin: Yeah. If you don’t have an existing relationship or you don’t have the existing workforce, right? Yeah. Yeah. So if you’re a new entrant who hasn’t done this in a long time and are trying to put your name on it, it’s going to be real expensive and not as competitive. Yeah.
Stephen Lacey: Caroline, how is NRG thinking about gas development? I know that you just acquired almost 13 gigawatts of capacity from LS Power. I know you’re still getting a handle on that portfolio, but like generally, what kind of projects do you think will be in the portfolio? What’s the view on gas generally from NRG’s perspective?
Caroline Golin: Yeah. Let me take off my non-NRG podcaster hat and put on my NRG hat for a second. We’re absolutely committed to the bring your own power and a bilateral approach to new build. The thing that’s interesting and I think phenomenal and one of the reasons why I joined NRG is that we are responsible for a lot of residential customers throughout the US, right? And so we’re just as sensitive to the affordability issue as anyone, because we’re in competitive markets and we don’t want to lose our customers and we care about our customers and we don’t want to see their bills go up because this power development isn’t done responsibly. So we’re very committed to the bring your own power and that needs to be done through a bilateral negotiation. And that looks different in ERCOT than it looks in PJM. But I also think that there’s a lot of opportunities for NRG to partner with cooperatives or even smaller IOUs that don’t want to take the development or balance sheet risk of building some of this because we’re one of the few entities in the country that actually knows how to build power plants.
And we’ve kept our entire gas team and gas desk and we actually know how to integrate new gas with virtual power and with renewables and how to wrap everything from the C&I down to the residential space. So I think for us, it’s an opportunity for growth, but it’s also an opportunity where we feel like we know how to do it better than anyone. And so we’d like to set the bar for how you do it well. And so none of these residual issues that we’ve been talking about for months occur and that’s, at least that’s what I’m focused on. Yeah.
Stephen Lacey: I want to bring this back around to the tech companies to close out and try to understand like how to view a lot of these new projects. So on one side, you might have someone saying like, “Well, if you know anything about power markets, you’d know that they were going to be relying on gas to build this quickly.” And then you have maybe folks who are looking at this from a climate perspective who are saying like, “Look, these companies are walking back their climate commitments. They’ve been promising the world to us and now they’re doing everything they can to walk away from what they said they were going to do and this is a climate disaster.” How should we think about what these companies are doing considering those two ends of the spectrum?
Caroline Golin: Okay, I’m going to take off my NRG hat and I’ll go put on my corporate tech company hat. I think it’s culturally and behaviorally interesting that customers were allowed to, and no one did this more shamelessly than me, and I’ve said that before, but a handful of customers out of the west coast of the United States who happen to have a very large influence on our overall GDP, of course, but we were able to set the narrative of who’s responsible and who’s not responsible. And I saw this a little bit in states three or four years ago where I was like, “Why are the tech companies being held to a higher bar in legislation in terms of their renewable deployment than the utilities that have been there for 50 years?” Like that was interesting to me, but I sort of said, “You know what, whatever, we’ll meet the bar, let’s move on.”
And it was somewhat a social license to operate question for me as well. And I didn’t mind having the bar be high for us because it meant that I would get to do more cool stuff, right? But what the risk of that was, is that we put all of our faith and all of our energy, so to speak, behind a handful of companies that weren’t energy companies. We weren’t energy companies.
We don’t know how to do a lot of this. And if we don’t know how to do a lot of it, and there’s sort of a gun to their head around meeting a fill need in order to stay competitive, which in a rat race that we’ve deemed a national security risk, they’re just doing whatever is right in front of them because they weren’t given the run rate and they don’t have the workforce to do it. And they also, in my opinion, shouldn’t be the ones that are held first accountable. We should be holding the utilities and the suppliers accountable for doing this well.
Stephen Lacey: Jigar, how much responsibility do you lay at the tech company’s feet?
Jigar Shah: Oh, I agree with Caroline. I mean, and you and I have talked about this since, what was it, 2013 when you and Shayle Kann talked about Google buying a utility company. I was so mean to you mercilessly. Yeah, remember there was a whole thing about Google should buy a utility company and I think you did a podcast.
Stephen Lacey: You’ve been so mean to me over the years. I forget which example.
Jigar Shah: I just think that in general, the reason why I think there’s so much whiplash right now is that when we were at the US Department of Energy, we were providing all the technical assistance to the tech companies. So when they were thinking about doing these things, we were helping them with modeling and this and that and whatever.
And because we had a relationship with the utility companies, they sort of went along and they looked at those models and worked with us on that, right? Whether it’s Georgia Power or others, right? And now that we’ve left office and the new folks are in the US Department of Energy, they have been vibe coding energy markets, right? So they’ve been saying, “We can actually just run the entire energy market with 202 letters to keep coal plants open, even though those coal plants aren’t needed in those markets, or we can just power the whole thing with natural gas because I woke up at a Holiday Inn Express last night and that’s what they told me.” Fine. And so the tech companies are doing exactly what DOE is telling them to do, which is great, and they’re getting permission.
The problem with that is that there is no modeling on the planet that actually says that that’s the most affordable way to run the grid, right? And so we are in this weird spot where we are deliberately choosing to do things based on a Secretary of Energy who basically is like natural gas or bust because all of these other things are unreliables, right? And I’m like, okay, but there is a whole body of work that shows that that is not the cheapest way to provide generation or transmission or distribution. And so we will just have to wait for sanity to reemerge.
Stephen Lacey: Well, to be continued, Caroline, I’m going to let you get back to your Charleston vacation there and then we’ll see you next week in San Francisco.
Caroline Golin: Sounds great.
Stephen Lacey: Enjoy. Bye, Jigar. Have a good rest of your week and we will see you in SF.
Jigar Shah: I will be pumped up for transition AI.
Stephen Lacey: I don’t know, maybe we’ll have a bet on off-grid data centers or something.
Jigar Shah: I don’t know. We’ll have to see. Well, I need some more data.
Stephen Lacey: All right. Transition AI, if you want to learn more, get your ticket. This is the final week. I think we’re almost at sellout latitudemedia.com/events is where you can get your ticket. We just have a handful of tickets left. Really killer line up there so you can see the whole agenda at latitudemedia.com/events.
And Open Circuit is produced by Latitude Media. The show is edited by me, by Sean Marquand and Anne Bailey. Find all of our episodes of Open Circuit on YouTube. Just subscribe to Latitude Media. You can find the audio version, of course, anywhere you get your podcasts and head on over to latitudemedia.com to read our newsletters and get more stories on the stuff that we talk about here. Thanks so much for being here. We will catch you next week and we will see some of you at Transition AI.
