Who Gets America’s Power: AI, Factories or Homes? | Scott Ebeling

The U.S. electricity grid is now at the center of the AI discussion. At the same time, manufacturers are also needing power with robotics and onshoring. Will we have enough power? And what will happen to the billions of dollars invested in AI infrastructure if we cannot meet the power needs on time? Today we sit down with Scott Ebelin, who will give us an insider perspective into the grid, how it works, and what is the realistic timeline for the grid to be able to absorb the demand coming on. What happens, though, if they build that out? Hyperscaler comes in there, they're there two or three years, and they're like, yeah, thanks for building out that infrastructure for this two gigawatt facility we built. We don't really need this anymore, or we only need like half of this, or we're bankrupt. Now, credit risk on their end, right? I mean, these utilities can't just go and eat $300 million of cost and infrastructure build out. It's like, oh, oh, it's lost $300 million. Oh, well, let's just raise the rates by everybody by 30%. I don't think that's going to play out too well. So timing and accuracy, it's really important. I'm Siamak. Welcome to Market Insider. Scott, it's great to have you on. Welcome. Thank you. Thank you. Very nice to be on here today. We want to talk to you about the power. And there's a lot of thought and discussion about power needs going forward. And we talked to you before, and there's different elements to the power needs. One is manufacturing, which a lot of people maybe not thinking or talking about it, because there's going to be a lot of automation and a lot of manufacturing coming back. Another side is the AI and the data centers. Can you tell us what you're seeing right now with power needs that we have? Power needs have totally exploded, Siamak. From 2000 to 2023, we saw about a half percent growth rate, which is normal. That's relatively normal. A half percent growth rate we had. A lot of efficiencies, energy efficiencies happen. And so the grid didn't grow much, right? So the whole setup is supply chain is completely set up for this half percent growth rate. And then, boom, this thing called chat GPT came along. And there was this cool little tool you could play with. And, wow, this is great. But with that comes a lot of power needs. And it's growing and growing and growing. And the forecast is for that to explode. And so with that, now the next 10 years are expected to be a 2% growth rate. Now, a lot of people will think like, wow, OK, 2%, big deal. That's a big change in the supply chain setup to be able to build that all out. It's not as easy as just saying, hey, let's go build a generator. Let's see if a data center and boom, let's go at it. There's a lot of equipment, a lot of needs that take place. We talk about that. And then, of course, you have the data centers are looking to grow. And these are gigawatt centers. But there's also a lot of businesses that are like in big businesses, which are a lot of our clients that we have. There are these 5, 10 megawatts that are having to struggle and get interconnections as well. And it's really because of this. And it's almost like a battle going on right now. And people are saying, our costs are going up. Our costs are going up. And of course, we all hear it from the residentials. We see this in the political spectrum right now on the cycle we have coming in the midterms. But we're also seeing a manufacturing side saying their rates are going up too. And the issue with that is with their rates going up is that that impacts them enormously. AI companies are coming in, the hyperscalers, their role is price agnostic. They're like, OK, they just want the power. They don't care if it's $70 a megawatt, $100 a megawatt, $120, $130. They just want the power. It's actually a small cost to them for the power from what they can see in their profit margins from the future on the cash flow. They're hopeful projections of the cash flow going forward. In a sense, what you're saying is that the data centers, the AI players, they're willing to pay whatever it takes. The rate doesn't really matter to them. But then when it comes to the manufacturers and the residential customers, there is pain there. They're not getting power. Can you tell us more about what you're seeing out there? Yeah, they're getting power. If they're building a new plant that's like, let's say a manufacturer wants to build a new plant that's 20 megawatts, 30 megawatts, right? And a megawatt, just for users can understand, is one megawatt is equivalent to roughly about 1,000 homes of usage, right? So now a one megawatt, a gigawatt, which most of these plants are, and it's the same size like a nuclear plant as well, a generation size. You look inside in a hyperscale about a gigawatt. That's 1,000 megawatts, right? And so what used to be a big load going onto the grid was like, let's just say a big auto manufacturing plant or a hospital that's 10 megawatts, right? An auto manufacturing plant, that's 15, 20 megawatts or a very large 80-story office building that's going to be 5 to 10 megawatts. Those are very large. Those are the largest loads that usually kind of hit the system up, right? And so what happened is now they're needing interconnection. And it's the same kind of equipment that's needed, these transformers and all this and this stuff. And so they're kind of needing that equipment as well if they come on. So what used to be like, hey, yeah, we're going to come on. Here's our date. We're coming on. We're going to build this. We're going to be completed by middle of 2027, let's say right now. Like, well, hold on. We're not sure we can get you hooked up yet. You got to put you in a queue. So there's that issue going on there. But the other side of it, too, is a facility that's been around for a long time. What's going on there is that their costs are going up. And you're seeing these costs go up enormously. They're like, whoa, our costs are going up. These are companies that maybe have a 5%, 10% profit margin. And electricity was, say, 10% of their cost basis. And if those costs double, that eats heavily into that profit margin. So they can either raise costs or they just have to eat it. They got to figure that out. And so we're seeing a lot of issues with that. And there's different strategies you can use to get around that. They can do. But in the end, they're facing these larger costs. When a data center comes in, that's equal to, you mentioned, it's like 1 gigabyte, right? Is that like 1,000 megawatts? So it's like almost a million households for a data center. Yes, exactly. It's almost like a million households, right? 1,000 times 1,000. Yeah, it's massive. What about robotics and all that? Because there's a lot of discussion about humanoids and robots and all that. Do you think, how does that work in this mix? Does it take a lot of power to have a robot running around or working in a factory, actually doing robotics? Yeah, it's crazy how things are just going so fast in this technology, right? I remember back in 10 years ago, it was all about like, oh, battery cars and how much it's going to take on the grid. If we go to like a 50% saturation rate of electric cars coming out, and what's that impact on the grid? And then all of a sudden now, this AI thing comes in and it blows that apart. And now you're talking about robotics, right? And generally, they need electricity, right? If it's human workers, all you need to do is just feed them, right? Like, I need some carbohydrates, you know? It's like, go throw them a couple of Domino's pizzas and they're fine. But this actually need energy to wire up. This is something, I mean, to be honest with you, the market, everyone's still trying to understand that and understand also, as well, how quickly that really comes on, right? It's still generally like, okay, this kind of works, the concept. But, you know, I mean, technology happens fast, it happens quick. But, you know, and also to get the factories going to build these as well and get them all set up. I think that's kind of the next level. I think what we're going to have here is this build out of AI. I don't think it's going to happen as quick as projections are just because of the limitations we have to build out everything that, you know, like, you know, transformers used to be, you know, 14 weeks to get back in 2020. Now it's two years, right? So just to do all these things, it's just going to take a while. Yes, just everyone demand and they can build a data center, but to get it hooked up to the grid, get all the equipment there, do all this, that's going to take longer than people think. But I think all that's going to be is that maybe out through like 2033 to 2035, that's going to happen. And the build out hopefully with robotics, as far as like maybe some of those energy needs, because of course they're going to have to be charged up and all this and the electricity they need. And if that really comes into fruition, where, yeah, this is all going to happen quicker, that's going to be, I think, more of a mid 2030s kind of story. But it could happen quicker. Now, Scott, you mentioned 2% growth. Can you tell us what it means in this field of power? And you also brought up transformers. You know, we don't have enough of them. How does that all work together? Transformers basically are what kind of like, you know, bring the voltage levels into, transform the voltage levels into the facility. Also from a generation, a thing to bring it down into the grid, all that, right? So it's just kind of like a way to transform it, right? These are, you know, relatively expensive things. And so the supply chain, you got to think about this, they were set up to build these things at a certain number a year, right? And now what's going on is that there's a, really is 2% growth rate. It's now like you need it to revitalize the grid, the brand for generation side, also for the side of the user. So you need all this going on at once. And like, it's not like we can just say, hey, yeah, let's just, it's not like, you know, you're a farmer and you're like, hey, we need more corn. Well, I'm going to plant more corn on more acreage. You know, it's not that simple as to just start it up. It's a lot of materials going into this, a lot of rare earths going into all these materials. And it's also, it's a side of, too, of natural gas generation as well, which everyone's is really kind of a fan on here, except for my state, Illinois, of like having natural gas generation and not increasing. We got a plethora of gas here in our country. It's relatively cheap. It's probably the cheapest energy base of anything in the world right now is U.S. natural gas. And like, hey, let's just build these plants and do this. And like, it's not too hard to hook up to a natural gas pipeline and just set up a generator, as some of your guests have talked about before, and put it and then have it be a, behind a meter or a kind of a hybrid approach, right? But the problem is turbines are, you know, they weren't set up to build out, you know, you know, Vernova, which is GE's company, and Siemens weren't set up to build, to have 3X the production that's wanted versus 10 years ago. And it takes a sea change to do all that and to source the parts and get it all. It's, and that's the complex part where there's a lot of people who don't really appreciate and understand. Now, Scott, how does the grid operate? You know, we have gained more efficiency out of the grid. How does it operate? And is there ways to use what we have, optimize it more? I'll start out with this. The load isn't flat across it, right? So it's not like 24-7. Now, a lot of places operate that way. We're 24-7, the power's kind of relatively even. One in the morning, you're using, you know, two megawatts at, you know, five in the afternoon, you're using two megawatts, right? So it's not set up that way. So the way the grid works then is, there's higher peak hours, right? So really in the middle of the summer, you know, in most regions on a highly hot summer day, the grid peaks, right? And so that's what all these RTOs, they're called Regional Transmission Organizations, so ERCOT, PJM, MISO, even if it's in a regulated state, they have this, and their goal is to keep the grid at the 60 hertz, right? So they got to have the same generation coming in, the same power, so nothing gets fried. Whether you go above or below it too much, it goes into, I'm not a physics expert. So this is at the peak, right? That's about my knowledge of it. This is at the peak, or it's designed for the peak? Their goal is to maintain this 60 hertz level, right? To keep it at this level, right? And so you got to have the generation and demand kind of matching, right? So, but the point though is that they have, you know, for instance, ERCOT peaks out at about 91 gigawatts right now, you know? And so they need to have 91 gigawatts in the summertime to have the available load. PJM peaks out at around 160 gigawatts. They need to have, and PJM is RTO for, you know, most of the, like Maryland, Pennsylvania, Ohio, Illinois, all that, you know, go into it more. But the, so you have to have this grit, you have to have enough generation for that hottest summer day, and then a little reserve margin in case everything kind of breaks down. So for that, you know, PJM does this thing. Now PJM just had an auction about a month and a half ago. And typically they try to target 20%, and so they have an auction for capacity to make sure they have these, they've subscribed right levels. And there was a 15%. They tried to target 20% called reserve margin. It was 15%, which I don't think you're going to try to get much lower than that in the next elections. So anyways, what they're trying to do there, so they have those levels, right? So the power peaks out here, but what you can do with batteries is basically with a battery, you're basically moving electricity through time, right? So you're taking it at the hours in the middle of the night at three in the morning, for instance, when the power is way low, the power needs are way low, boom, and it's cheap. And you load up the battery and you dish it off at that peak level. So what you're doing now is just smoothing out these peaks. And what's great with this is that we had about 10, about three, four years ago, about 10 gigawatts of battery storage in the U.S. We're now at 52, about, I would say, probably in the past month or so, we're at 52. This is the forecast. And the forecast is there because it's all subscribed, ready to go. And it's utility scale, meaning it's connected to the grid, is expected to grow to 110 in the next three years. So almost doubling again in the next three years. That's an incredible amount. And it's really going to help out. And this is where the grid needs to understand having solar, wind, batteries, and Texas nailed it right, having it all, the true all, bringing it all in, is the best approach. So battery technology is getting better and better. And I think that's going to be kind of really, continue to be a secret sauce to the mix here. What about the sources of power? You mentioned Texas has done a great job with mixing all sorts of power. What about nuclear? Is there any thoughts on nuclear? Yeah, so good old nuclear. Yeah, let's do nuclear energy. Of course, we all know, or maybe some of us, those of us here are younger and much younger than me, but Three Mile Island pretty much blew up nuclear plants from getting built, right? And then we've had one nuclear plant since then. And that took place in the late 70s, early 80s. One nuclear plant that got built since then. And that was out in South Carolina. And it took years to build, big cost overrun. So the 10 billion project turned like 25, 30 billion. So these are very expensive to build. They're not cost effective. They take five to seven years at best to build it. You're typically longer. We don't have the expertise in building these things out anymore because it wasn't being built. Who's going to learn how to build these things? And so I don't think nuclear, I think nuclear is a great idea, but gosh, it's hard not to build a natural gas plant right now, let alone a nuclear plant. And I don't think that's necessarily the thing. Now, what's really funny is that society right now is a poll that was recently taken. And 44% of people would rather have a nuclear plant built in their neighborhood than a data center. 44% would take a nuclear plant in their neighborhood. 15% would approve of a data center in their neighborhood. So, I mean, that was unheard of 10 years ago, a couple of years ago. And so I just don't think it's happening, all these like small nuclear, small modular reactors and all this. These things are still relatively theory-based stuff and it's not cost effective to build these out. Those are all smaller scale. I just think that's a story for more of the late 2030s, 2040s, but there's just not this consensus getting around, like let's build these, you know, and they're huge cost allocations. They take years to build and it's not gonna be this, two years ago, everybody was talking, hey, let's build, we gotta build nuclear. I mean, look, I think there's a huge consensus, probably 90% of the public and politicians agree, we need to build nuclear. And it's just not, and two years later after that, there's still nothing been broken ground yet on a nuclear plant. I bet you by 2035, we'll have two built, by 2040, we might have another dozen built. Good, it's good to have it. Now, do you think the power bills will go up significantly from here? I think, and this is our thesis and it has been, I think power is relatively, we made a call on ERCOT at the, about three or four months ago, and it's come true that power prices have come down. There's just different levels to it, right? There's the energy costs, and then there's these other RTO costs, right? And RTO costs are, in some states, there's capacity, transmission, you pay for those. Those are demand levels. You're kind of like set for like, what's my peak demand when the system's at their demand? What's my demand when the system's at its peak? So here's my share of when the system's at its peak, what I need, right? So a church is gonna have lower because it's gonna probably peak on a midsummer day, on a Wednesday, not on a Sunday, right? So they're gonna have lower capacity, transmission costs, things like that are part of that. There's things called ancillary costs, which are really things like I talked before about keeping the grid stable. You know, it's grid stability at keeping at 60 Hertz and those costs, all those costs are rising. Your transmission costs are like things for like these big transmission wires and all this and upgrade. These costs are all increasing. They can be controlled by handling those demand costs, but those are all increasing. Also the energy commodity costs increase and there's behind a meter. I mean, then there's the next level, which is the actual delivery costs, which is your local utility and delivery costs actually delivered over those distribution lines. We all see the smaller lines. Those are all set up. So those are all the costs. There's different ways to do it. Now, most of the charges in most states is energy costs and then there's other costs, RTO and utility costs. Those are all basically set up as almost demand-based charges. And if you can control that, like we talked about, you can really control your energy costs a little bit better as well. We still think, we think the raw energy costs are about as high, are pretty fairly priced, if not overly priced a little bit right now. There's been a lot of hedge fund activity in buying up some of the power because of the story behind it. But if natural gas, because they're all set up, a lot of them are set up to be at the level where natural gas, because then the marginal fuel for generation is natural gas, and natural gas prices rise, those will rise as well. But we think they're fairly valued for where natural gas prices are right now. But we think, so those will rise. We think those will rise a little bit, but I think those non-energy costs, the non-commodity, those will rise even more percentage-wise. And that's where, if businesses can get control of controlling their demand when the grid needs it most and controlling their share of that, they can really save that 10% to 15% a year. Do you think that some of these data centers will not get the power? And some of these factories, as we are bringing, on-shoring more, do you think there's going to be scenarios where places are sitting, they're waiting for the power? Well, it's already happening, right? I mean, it's a, we have, it's interesting because I think the kind of pro-America agenda here is to bring on as much, let's reshore all these jobs in manufacturing. Let's dominate the AI space, right? And also let's ship out a bunch of gas, natural gas as much as we can as well. And it's all these kind of things. And like, hey, let's have really low prices, right? And there's gotta be some give and take. But in the end of the day, they're not gonna allow, none of the grid operators are gonna allow demand to come on if there's not generation to match it up, right? Because it's just gonna blow the system apart, right? So, and we're at those levels right now where every bit of, there's a cushion here for the last three or four years. We're like, yeah, we have these bigger reserve margins than we targeted. And like now it's kind of coming into where we're not there. Any new stuff has come on has to be that. So there is, it is happening already. I mean, data centers are getting delayed and they're getting it from both ends, right? It's also, it's coming to be a political issue as well, but a local political issue, but you're also getting it from like, it's just not ready. It's just, the system's not ready for it. And the manufacturing as well. I mean, it's a, it just has to be a longer delay. It will happen. I think that everything will, the supply chain will come into, to get to the point where it can learn out. Will we be back to 2020 levels where I can order a gas turbine and have it here in a couple of months or a transformer and have it here in a couple of months? No, but I think we'll see these supply chain issues get cut in half. Things will be in the queue. We'll have a better idea. Like these talks of ERCOT going to, it's at 81 gigawatts now going to 300 gigawatts. That's not happening. I see in 2030 being 120. It's not, there's a lot of things that were in this system that are just, we're in there and everybody's gotten smarter about it. So they're getting a better vision on what's really real and what's really not. And what's really just people putting it in there because it's really cheap to just get in the queue and let's get it in there and like hope that we can like, we get called that we have a slot. We can now go build, right? And so they're trying to get a better idea of that and I think they're starting to. And with that, then they can like bring in these manufacturers in the queue and everybody knows those are the real jobs. And I think all the regulatory agencies and all this and their RTOs, I think they'll figure it out and they'll kind of like start understanding that like there's jobs here available. This is really more the lifeblood of our economy is really these manufacturing jobs and reshoring manufacturing and having that be a little bit more of the true thing that we really want to have here in our country. There's a lot of money going to the hyperscalers and they're borrowing a lot. And is there a mismatch and is it a few years? Is it a couple of years? What are you seeing out there in terms of them getting their power needs? Yeah, we see through 2028 right now is a 40 gigawatt shortfall of what the data centers want and what's going to be available, right? So that's from our analysis we have here, right? So with that, that's going to have to be kind of be brought into the picture as well as it goes on. We see kind of this, and that's why we're the reason why we think prices are a little overcooked, right? In PGM markets, right? And I think it's energy costs and we get in that strategy when we talk to our customers. But these 40 gigawatt kind of thing, what's going to happen is that's going to have to come into the system eventually. I think what really is going to happen here is that there is a, there's going to be some efficiency gains on the chips. Of course, we're seeing that a ton with NVIDIA chips and other companies are coming out with these chips and it's because there's incentive to do it, right? Instead of the knockoff NVIDIA, NVIDIA sees huge fat profit margins, right? So with that, there's incentive to build these and make them more efficient. And I think these data centers become more efficient. And with that, I think there's going to be kind of a 20 to 2028, yes, there's 40 gigawatt shortfall. Those will start coming to the system. Meantime, some of these needs will kind of come down a little bit as well. And I think we'll smooth into where we'll probably be even with like the demand supply and really just kind of getting things built when they need to be by like 2031, 2032. And that's the problem, right? I mean, everybody wants it now. I want everything now. Like we're America, we need it now. And it's just not set up that way. And I think people need to just kind of realize that. But hey, we adjust. We do adjust, systems adjust. It's a capitalistic economy here where there's incentives to adjust and figure out solutions and get the things that people need. But I think it's just going to take time. And I think it's going to happen by 2031, 2032. In terms of the manufacturing side of things, how do you think this will impact on ensuring? Yeah, it's very interesting. If you're under 10 megawatts, I think it's pretty easy to get in the system. I don't think it's, I think you'll be able to get in there, right? I mean, it's not that hard. And that's gonna be typically most manufacturers, right? There's not, you go down the road and you go down an industrial center, industrial corridor. Most of those loads are in that five megawatt area, right? So, but these very large factories, these require energy, right? But also, it's not as hard and as big of a deal an internet connection side. And also they can be, they don't have to be clustered around large population centers either. For instance, these data centers, we're talking about 2% growth nationwide. They're clustered mostly in like Virginia and Chicago area. And they're like, so that's really these more bigger population that is growing. But a manufacturer, you don't have to be necessarily in those population centers. So you can go find it and put it, source it anywhere. And there's, we're finding spots where like, yeah, we can quickly grow this up. We can quickly make it grow up. We just have to get the equipment and the transformers, all that. We can do it. There's slot for you. We have excess generation or there's a lot of still, for from the Rust Belt area. I know there's a ton of like, I've driven through Ohio and looked at sites before for looking for like five to 15 megawatt sites. And there's a ton of like mothballed old factories sitting there with the hookup already, right? And the transformers are there. It just needs to be like, oh, we're now back on. So they're there. I don't think it's gonna delay it heavily. I think this most of this delay and this kind of thirst for power we have is gonna be more in the hyperscaler side than the manufacturing side. Do you think the timing could impact some of these companies if they were expecting to get the power next year but they didn't get the power on time? Do you think there's gonna be cases like that where a data center will be sitting, waiting? I think they could be waiting, but they already have data. It's just a matter of they're just, they're not able to like go full tilt. They want to go. They're getting their power. It's just the needs that they have. It's more like, here's what I want, but I'm willing to take a little bit of that, right? Like, and to slow down. It's gonna slow them down a little bit, but I don't think it's gonna like stop them. They just have to adjust a little bit to kind of like be like, okay, you know, maybe they start charging all of us a little bit more to go check out and see what's the best itinerary when I wanna go on a trip to Japan for two weeks, right? Like those kinds of things. Now that's not a huge amount of it, but maybe they gotta start charging users a little bit more because I think the real money for these companies is of course on the corporate side, right? Of using it and agents and all this. And again, it's, there's, I see that in token usage, right? I see that in, you all see that in a lot of companies and the waste of it and how many people are trying it. And it was like, when we started the year out, we're like, we're like, you know, all these companies were, and I talked to my friends and their colleagues like, go out there and use AI and learn everything and just go at it, right? And all these companies, everybody was doing it. And then like, what, about three or four months ago, I was like, oh my God, this is a lot of money. These tokens actually cost money. Maybe we should have a better plan and just have a few people working on this stuff. So I think we're getting in those efficiencies. It's also a question of really what, do we need like eight great, like eight great AI companies and they're all fighting, right now they're all trying to figure out who's the best. And I think the end is gonna be like maybe two or three that are really good. And maybe Google just says like, wow, that was a fun 400 billion we just spent. And now we're not in the top here anymore. I'm not, again, I'm not saying Google has a bad, a bad thing to have, it's great. But like, it's like, that's kind of the thing or Meta or somebody, Meta's done a lot of experiments before like, ah, okay, oh, well, what's the next thing we can do? Robotics? Okay, you know, who knows? I mean, it's, and so they don't need, Meta maybe don't need, they don't need a five gigawatt facility down in Louisiana, right? Like who knows? I mean, so it's, these are questions that I think a lot of people don't know yet. And so that's the other thing that could, this 40 gigawatt shortfall, we can eat into that is like, oh, maybe we don't need all this. We don't really need all this. We just kind of adjust and we move through, move through the system. And so, yeah, I think that's kind of the, really where I see this potentially going. Again, I don't, I'm not necessarily making that prediction. I just see like, as a former trader, I always look at all sides of it, right? What all can happen? And I judge it by probabilities at that point. What about us versus China and Europe? Great question. Europe, way distant behind us and China. They have much different regulatory levels. They're very green over there. And I'm not, I'm not an anti-renewable person. I'm all types of energy, but they're much more renewable based over there. They have natural gases. They need to export it. They're having really trouble with this potential winter. They can have some huge troubles because they got most, a lot of their gas came from Qatar and it really increased LNG costs. And they got to import the LNG to get there to take a lot of it from us, for instance. And they're now paying, you know, four to five times what we pay for natural gas here. So they're way behind. China, they've built out and talked about doing it all. I mean, they are the largest, they've increased the most in coal and also in solar in the past five years. So they're bringing on the largest increase in coal, largest increase in solar. So they're really just saying, bring it all in. And since they started more, part of our issue is in our country is like, we're so built up, right? And we built up this grid that's over the last 130 some years, right? Built out this kind of grid thing and we have it all out. And so, but they're mostly starting this, a lot of this from scratch. So they can like, and they're also, where do we've been importing most of our parts we need for this build out of our grid? It's from China. So they have the needs, they have what they have, they need right there. And there's also just not the regulatory issues over there as well. If the Chinese government wants to build a transmission line and knock out a whole neighborhood, because that's in its path, they just go ahead and do it. It's not a, and so they're ahead of us in as far as like the rate. I mean, their costs of their hyperscalers or their electricity costs are half of what they are here in the United States. I mean, so there's a huge cost advantage to be out there and to, and if you're a heavy electrical user to manufacture over there, then it is here. It's still that huge cost advantage with that. A few people have told us like the power is the safest part of the AI play because you have all these different layers of the AI play. And at the end of the day, the power companies will benefit from anybody succeeding. For utilities, so that's on the generation side, right? So your constellations, NRGs, they own a lot of, it's called merchant generation and deregulated states. They own these generation plants and they monetize them, they make money. And it's a, they're well-priced for this current thing that can happen. Now, if prices come off, a lot, and there's just kind of a crash somewhat and the prices come way off, those stocks are going to take a hit. I mean, there's no doubt about it. Then you have utilities, your actual utilities that are out there that, you know, Exelon's, your, you know, Detroit Edison's, you know, some companies are hybrids, they have both, right? I mean, these companies are public companies, they're utilities. They're set up on a rate case base. It's a monopoly. So, if they, some big data center comes in, I want to build this and they build out the infrastructure for it, those costs typically are covered in, they put in the rate case, cool, we got to build this infrastructure and now we're going to collect those costs over a certain amount of time from our rate payers and they'll collect those costs back. I mean, I think one of your previous guests discussed this, let me see how they're really in the infrastructure game, right? They just want to build infrastructure, the more infrastructure they have, it's a, you know, they just make more money and usually that's 10 to 12%, you know, return equity sort of level that they just get guaranteed. It's a great business, right? Like, yeah. Monopoly and then you got, you get paid. You want to open up one of those businesses with me? Let's do it. Yeah, right. So, but if the market crashes, they're still guaranteed really to get that. Now, I think what's going to be interesting here, and I think some people aren't really talking about, is that now you've got where these companies are, the backlash going on right now. We want the hyperscalers to pay for their own grid and we don't want to have the residents pay us. There's special tariffs that get arranged and with that then, like, okay, we're going to build out this infrastructure, have the special tariff. You're going to pay this off over this 10, 15 year period. What happens though, if they build that out, hyperscaler comes in there, they're there two, three years, and they're like, yeah, thanks for building out that infrastructure for this two gigawatt facility we built. We don't really need this anymore. We only need like half of this or we're bankrupt. That's now credit risk on their end, right? What's going to happen? Are they going to like, be able to come back and say like, okay, now we just lost a bunch of money here. We're supposed to be guaranteed this return on equity. Can we go shift this back to the other customers that aren't hyperscalers? I don't think that's going to work. So that's, again, as a former trader, I always look at each side of that, right? So there's risks, there's risks. Worst case scenarios. Yeah, and I think that's a risk that people are in this counterparty risk that could potentially happen if you go to these special tariffs, which is everything's moving to that. And it's really the only way these hyperscalers are going to get, most states are going to get done is by these special kind of tariffs because it's not, even in a deregulated state where you're responsible for buying your own energy and stuff like that. These are the states we work in heavily. You still need to connect to the grid and connecting the grid is the local utility who owns the wires and all that. And they're the ones that have to build up the infrastructure and you have to pay, right? So it's like when a manufacturer comes in, like you can typically decide like, okay, I'm going to have this build out from the wires to my points, transformers, all this. You can pay for that yourself. You can have that be charged over time and the rate. These are choices you can make in most areas, right? So it's similar sort of thing that could happen here. And most of these hyperscalers will probably say, yeah, we'll just pay that rate over time. And that's, they go out. That brings another risk. Yeah, who's paying for that? This isn't like, this isn't like a, you know, a small little widget maker that's got two megawatts of load that goes out of business. And it's like, oh, okay, yeah, we just lost, you know, 150,000 of the build out of the infrastructure for that. Oh, well, eat that, right? No, I mean, these utilities can't just go and eat, you know, 300 million dollars of costs in infrastructure build out for some of these AI things, right? And it's like, oh, oh, we just lost 300 million. Oh, well, let's just raise the rate by everybody by 30 percent and we'll collect it back. You know, that's, I don't think that's going to play out too well. So timing and accuracy, it's really important. If somebody wants to, if people want to invest in the power side, they have to pay attention to potential big losses if it doesn't work out. Exactly. I mean, if you're an investor, make sure that you are paying attention to all the risks that are there. And like everything else, not just live this, you know, kind of this dream of like, yeah, here's everything, follow the herd of like, yeah, all this stuff's going to get built. Let's go buy this, buy this stock, buy this stock, buy this stock, buy this stock. And granted, like I am bullish that on the fact of there's going to be increase in power usage, right? I am bullish in the fact that AI is going to be, and it already has been a transformational technology, right? I'm just not so sure how quickly, if it's going to grow like this or more like this, you know? We don't know that just yet. I think, you know, everybody's got that case scenario where it's going to be this huge, huge level of growth. And I think it could be less. I think these forecasted 2% growth nationwide are kind of over the next 10 years could be more like this one and a half to 1.6% kind of percentage growth, which is still substantial, right? And I think then everything just kind of smooths out from there. Scott Ebeling, it was great to have you on Market Insider. Thanks for sharing your insights with us. Oh, it was great to be on here. Really enjoyed it. I really appreciate what you do, the material you put out there. It's really enjoyable to listen to. Tell us what you think about the data center build outs and their power needs. Have you been impacted? Have your rates gone up? And make sure to always comment. We read all of them. I'm Siamak Khorami. This is Market Insider. We'll see you next time.