Powering AI: another shale story? with Hill Vaden

0:00Welcome to the HC Comortis Podcast, a podcast dedicated to the Comortis sector and the people within it. I'm your host Paul Chapman. This podcast is produced by HC Group, a global search firm dedicated to the Comortis sector. Today we return to data centers. Data centers are the crucial rate limiting factor for the hyperscalers growth and in turn power supply is the rate limiting factor for those data centers. Who is financing that power infrastructure? How are they doing it? And indeed what happens if there's a correction in the market? All the flood of money leads to overcapacity. And what is the best analogy in historical capital cycles to examine this current infrastructure built out with? Our guest argues that it's the Shale Revolution.

1:01Joining me in my office in Houston is Hill Vaden, executive director of the Energy Capital Insights Group at S&P Global. Who themselves are hosting the financing US Power Conference on September 28th through the 30th here in Houston, covering many of the topics we discuss here. I'll put links to that event in the show notes. And as always you can really help support the show by leaving us a positive review in the platform you're listening on. And as always I hope you enjoyed the episode. Hill welcome to the show. Thank you. Call me with a drink of water in my mouth. Thank you. Well I'm welcome to our office as well. We're doing this in person. Story we're trying to tell here is energy capital markets and essentially how the energy infrastructure, the power demand for this data centers, the AI boom is getting financed. We all know the sort of the AI data center story, right? Incredible projections of revenue and productivity and that's feeding back into an incessant demand

2:04for compute and data centers. And that in turn is generating an extraordinary demand for power and power infrastructure. The question we're trying to ask is how that gets financed and is that financing is sane, is relevant, is stable, you know, pain. And what that might mean for the energy industry more broadly, right? Sure. Because I think it is a very consequential story. Before we talk about some of the risks inherent in that, can you just help set up for us? You know, how the financing works when one of the hyperscalers decides it would like another massive data center somewhere? So I mean I can talk about it more from the energy side than specifically the hyperscalers plans to build a data center, right? But the big thing, I think the big shift that we're seeing with the hyperscaler appetite for power, appetite for energy is we've got very much demand pull bringing new electrons and new innovation into the market, right?

3:05And so the hyperscalers will sign a long-term agreement for power saying, hey, if you build this, we will take the power off of it, right? And sometimes that's an offset in the case of, you know, a solar or a wind or something like that where that low carbon electricity will offset some of the more-mitting from grid connected power. Sometimes it's the direct, we are going to buy the power from this plant using this facility, right? But the important thing I think is that you've got this contract that gives you the ability to underwrite a project that says, hey, I've got a credit worthy partner on the other end of this contract. I need to borrow money or I need to do whatever I need to do to build this X power plant, X being the type of power plant, right? And if you don't have that long-term contract from a credit worthy customer, it's really hard to get the financing to build something new. And nobody's really building anything new without a contract or if you're utility, you

4:07can rate-base it, right? But nobody's taking merchant risk and just building a spec power plant right now. OK, so you have the hyperscalers who essentially create these power purchase agreements that are the basis of the financing of all of the power infrastructure that sits attendant to that data center or at least the source of power that gets that data center built. The last thing that sort of, I guess, where this sort of intersects with the broader economy in some senses is that you've got a new set of players or shift from private capital asset managers who've been building data centers, right? They have been looking at cloud for quite some time, are realizing that that bottleneck sits in the power slice and they themselves are now doing either package deals with IPPs or whomever it might be, or indeed creating vehicles in the case of KKR to actually set

5:11up and stand up these infrastructure as well. Can you just help us understand how that works? Yeah, I mean, I think that one of the big things is the, I think something like $250 billion was raised last year, 2025, by infrastructure funds, right? And so there's a ton of money going into infrastructure where everybody wants to own a toll road, right? Everybody wants the predictability of a heavy asset that has predictable cash flows every month and it's not going to give you the high 20 plus percent returns of, say, a natural resource or risk gear investment, but it gives you that long-life cash flows in. So that's toll roads, that's data centers, that's power plants, that's pipelines, that's all sorts of things like that. So there's a real overlap in that capital where in some respects that money is indifferent to whatever it is, right? All I'm looking for is that cash flow generating opportunity and where the data center and the power store you're kind of overlapping.

6:13Private equities have been involved in both private credit and both, right? But it's kind of overlapping because now one of the major constraints to the AI ambitions being realized is being realized is the access to power, right? And specifically dispatchable generation, dispatchable power. Yeah. And how exactly, you mentioned a couple of structures earlier. How exactly are they doing these contracts? Because I understand that there are some that are kind of behind the meter entirely. So, others are sort of with the utilities and they have the utility reserves to provide the power from different sources and so forth. I think that so from our understanding, every data center wants to be grid connected, right? Because if I'm building my data center and I'm grid connected, I've got optionality, I've got a less at risk of an outage, right? And the one thing that particularly the AI oriented data center, so one thing that they don't want is interruptible power, right? Because if they lose power for a minute or whatever, then you've got thousands or millions

7:16or whatever, you know, pick your number, you know, a dollar's on the line. So I want to be grid connected, but it's hard for me to be grid connected quickly, right? And the cliche that we hear often is speed to power, right? So there's a real aggressive movement to get these data centers built and powered for 2030 or thereabouts, right? And so I need a tenant on the data center, somebody to actually, you know, generate the demand for the use of the power, I need the power and I need the land. And so where I think kind of all of this is kind of setting up on the power side because I can't get the grid connection as quickly as I want, I'm having to bring my own power to the project. And so you're seeing behind the meter gas reciprocating engines, you're seeing behind the meter fuel cells, you're seeing all sorts of creative ways to address your power needs as you wait for the grid, assuming and expecting that the grid is eventually going to be there to back you up because for the most part, for anybody that we've talked to, you don't want

8:19to be fully reliant for the lifetime of your project on or behind the meter project only. No, it's a lot of risk for the power provider as well, right? They just have one, one customer basically who may, may or may not go through hot times. Yeah, and we understand on the contract basis, I mean, these are 100 pages, you know, multiple hundreds of pages of contracts and your risk as a power provider, if there is, if this data center, if that power is interrupted, you bear the burden of, or you can bear the burden of a lot of that, right? And so I want, again, the number here is 5.9 reliability, right? That's 99.99, 9.9% reliable. And so if something trips, if something goes down, I need to make sure my facility stays operating. So this is where, just before we sort of get into the risk, that is what's driving data centers to the likes of Texas where you've got, because previously you were probably, I don't really care where my data center is, it's close to, you mentioned before we

9:19started recording, right, to fiber or popular, whatever it might be, right? The skilled labor and the connectivity. Now your bottleneck is absolutely energy is power. And if you can get a speed to power, you know, I don't know, that's correct phrase, but whatever it was, getting it quickly set up in Texas, because you can just draw on the gas, there's the expertise there, you can create these, you know, generations behind the generators behind the meter. So that is what's driving the data centers to these regions. To an extent, yeah, I mean Texas is also big, right? So you've got, like you see a lot of data centers that, there's been a ton of activity in Texas, there's been a ton in PJM, Pennsylvania, Jersey, Maryland, Virginia, you know, which is close to the traditional, they call it data center, Ali, right, where you've got the laid fiber, you've got all of that from the early tech boom, there's been a lot of activity in MISO, there's a big project that Metta's doing in Louisiana.

10:19And then you're looking, so yes, you're looking for the availability of power, you're also looking for the availability of just permits and the ability to build, right? And it is easier to get things done in Texas and it is to get things done and say Massachusetts. Yeah. Somewhere pretty. So you've got the hyperscalers, the Metta's, the Google's, the OpenAI's or whatever it might be who come along and want to contract the data center. That part of that is a PPA or some long term power agreement. The infrastructure funds love that because they're turning kind of a chunk of money into this sort of more bond like return over which backed by a contract by a multi-billion dollar company, that looks good for their returns. Where do the banks step in in this at all? Are they, you know, in presumed this looks like lots of money, so they also want in. Yeah, the banks are involved from what we're seeing, there's no shortage of capital, right?

11:21And so whether that's private credit, whether that's banks, you know, whoever it is, people are able to find the money to do the projects that they need to do. And so there's M&A advisory that a lot of the banks are involved in, there's debt, there's on balance sheet debt, there's all balance sheet debt, that there's a lot of different, uh, call it creative financing happening, um, throughout this. And then you've got the building, like the activity where you're pouring concrete, you're putting in center blocks, you're, you're doing a lot of, um, you know, heavy activity, which is part of, I mean, you've seen some of these crazy numbers about how much data is in our construction is contributing to the US economy right now, that these are, people talk about bubbles and whatever else, but these are real projects, whether we're in a bubble is, you know, a separate question, but these are real projects that are getting built in real places, putting real people to work, those real people that are going to restaurants and buying real hamburgers and staying in real hotel rooms, and there's a, there's a whole level of economic activity around it. Yeah.

12:21This is the conveyance of that money from future expected earnings of open AI or whatever it might be, right? Which is ultimately tied to productivity and there are some significant, you know, expectations about what revenue might be in a couple of years. You mentioned that 20, 30 date, right? That sort of, all these forecasts of anthropic and open AI being worth a trillion dollars is kind of based on these quite astronomical revenue numbers that they're going to generate in the future to do that. They need to compute that compute needs power. And that's the conveyance of that money that's future those valuations into hard assets, which as you say is, is great for the construction industry at the moment. The question we're about to start asking is, is that bubble now linked to the energy industry and what might that have an impact on the energy industry if expectations don't bear out like they should? Just at the moment, just before we sort of talk about that risk, is this like everyone's happy at the moment? Everyone's, you know, this is sort of the best time it's ever been to, you know, own an

13:22IPP or, you know, if you're suddenly in the utility business in unregulated markets, I assume it's, you know, you're feeling quite happy with yourself. I don't, my initial response when you said that, I kind of laughed, but nobody's happy. It seems like everybody's upset about something, right? You're not able to build your project quickly enough. You're not able to get, there's a lot of contingent deals where I can build the data center if I can get a tenant, if I can get the power. And so there's a lot of our eight. All this is going to come together if I get this one thing right. And some of these deals, boom, you're getting all three of them, everything that you need and everything's going together. Some of the power permitting stuff isn't going as quickly as people would want it to, right? It's hard to build things. Now you've got this NIMBY backlash, not in my backyard backlash that is, which is essentially the midterms are now hinging on this, right? And you've had politicians who were in support of data centers going back into a longer being in support of data centers.

14:23So maybe the tone changes after midterms, you know, I don't know, but you've got all sorts of things getting in the way of big projects, right? And I think that's, that's been true. It's hard to build big things in this country, right? And whether that's a pipeline, whether that's a power plant, whether that's a data center, then it's easy to stop big projects or at least slow them down, right? My view on the overall kind of data center and NIMBY backlash, there's going to be some areas that stop or delay it, right? And there's going to be others that maybe get delayed, but it's not going to stop the overall movement. And if I think about kind of other big booms like this, you know, things get slowed down, eventually things get built, often things get overbuilt, but those things get used, right? And the example sitting here in Houston, if you and I were to build a highway to Katie, people are going to drive on it, right? Maybe you and I go bust in the process, but eventually that pre-way is going to get filled by cars. Yeah. Just before we get on to that, it is interesting, right?

15:27Because you've gone from an environment where four years ago, most counties, states were competing to win data centers and throwing all sorts of tax incentives, you know, whatever it might be to win those, to an environment where it's absolutely the obvious, at least in terms of the political zeitgeist, and that can have profound impacts on ongoing project economics. And let alone the ones that haven't started yet. Sure. And anything that slows it down is going to cost more, right? And I think some of the things that, so there's one project in Southeast North Carolina that I used to live in South Carolina, and this was a town that just got destroyed by outsourcing, right? Or accounting that got destroyed by outsourcing. And when I was there 25 years ago, 30 years ago, 25 years ago, they couldn't attract any industry, right? And it was all cotton textile stuff like that, right? And so then those who were stuck in the, who were living there, their taxes were going

16:30up because they were, you know, they still had schools that needed, you know, all of that stuff. And now there's a data center proposed out there, which could bring in very efficient tax revenue, right? Because you don't need to build a lot of schools for data center. You don't need to build a lot of roads for data center. And maybe, and that should spread out the cost of power. Power is a high fixed cost business. And if you've got more people using power, it spreads out the cost on others. But you've got kind of organized resistance saying, hey, community, you don't want the status center in here because it's going to use water. It's going to do all these other things and some kind of things, some of which are legitimate concerns, but some of which are not, right? And part of the water concerns from what I can tell concrete uses a lot of water, right? And so there's a lot of water use when you build anything made out of concrete. Once that's built, the water use isn't quite the problem. And a lot of these things are closed loop, right? And the power prices, people's power prices were going up with or without data centers. Yeah.

17:31We had Ademarika Prometheus Hyperscale on a year and a half ago talking about kind of exactly the engineering and the power consumption of these chips themselves and how it has been rapid. The market is solving for these problems, right? But yeah, you're right. And then we live in a world of social media and all the rest of it that sort of can quickly metastasize these different points of view, right, that cause back to that. Which is the great irony, right, that those who are organizing protests to stop data stores or entirely are hiring, like, it applies to the virus. But the others are protests. Yeah. Okay, but as it stands today, and this is kind of where I find this story becomes very fascinating, is that you have spent a career looking at capital cycles, financing of energy and energy infrastructure. Obviously, latterly, that's been a power story, but historically, certainly in this town, that the US more broadly has been shale, oil and gas. And these two worlds are somewhat interconnected in that you have one narrative out there, which is, look, all of this is predicated, essentially on those, the hyperscalers, their growth, right?

18:35Their forecast is revenue off the back of developing of these data centers. That is fueling all of this boom in infrastructure growth. The last time we had one of these stories in this town was shale, right? And there are some pretty analogous ongoing, I think, in terms of how the capital is flowing, the expectations, the eventual cycle that will naturally go through and consolidation and who wins and stuff. But I would love to kind of get your, if you could sort of do a, you know, you're the one that came up with the idea that said, you know, meeting, right, that this is a decent analogy to look at. You should be looking at the dot-com boom or bubble or the great financial crisis, the one to look at in some sense is how shale fed. Can you start, I guess, help build that picture for us? Sure. I mean, it's what's the line that history doesn't repeat itself, but it sometimes arrives. Yeah. That I think that there are, when we're thinking about the energy investment boom, shale

19:36was the last one that was really exciting to me where things were, there were kind of hidden stories in the little corners of what was going on and there was a real kind of euphoria on it. People like to compare what's happening now to the dot-com boom or the real estate boom, you know, 2006 period like that. And I think there's good relative comparisons in those as well. What I think is kind of interesting and this is true for the housing boom as well, but it's both shale, the housing, what we're seeing in data centers. You get into kind of location specifics, right? Where things, people are going to make investments in Texas because they can find the land, they can find the power, they can find the pipeline optionality for gas, they can find the, you know, that you're going to have surplus power in the grid from solar or things like that. People are going to invest in Virginia because of the fiber. And so there are things that work in certain places that don't work in other places in the

20:40same way that I have to drill for shale where there is shale, right? And then the overall kind of winter take-all mentality or winter take-most. Early on in shale, you saw almost kind of a carpet bombing effect with leasehold, right? People would get as much leasehold as possible. And then once they figured out what they needed, they would kind of high grade, right? And you're seeing that now where you're building very large data centers, trying to move quickly with this kind of winter take-all or winter take-most effect. And then I think one of the other things that is interesting is that demand pull, right? That what got shale off the ground originally was a pipeline contract that George Mitchell had with NGPL and he had to feed gas into Chicago, right? And he had to find a way to produce that gas from a resource base that was declining. And so he got creative and was experimenting with different recovery techniques, which led

21:40to the hydraulic fracturing that opened up the Barnett shale, which then fed the pipeline. And then everything kind of took off from there. And you're seeing some of that innovation on the behind the meter story where people are saying, right, well, I got to figure out ways to power this thing quickly. And I got to figure out a way to power it with dispatchable energy that is there 24 hours a day, whether the wind's blowing or the sun's blowing. So I think. Yeah. So I think the second part of that story, like I think is absolutely true at the speed of innovation. I guess the interesting part for this piece is going to be the very early shale days were kind of the collect a, the land man, collect a garage, prove there's some stuff on it and then flip it, right? And that was quite a successful strategy for some very early investors. And there's arguably quite a lot of capital destruction post that event up until really post COVID when you'd have such consolidation, you'd have the big players take over, they

22:41are now approaching the approach to the capital discipline, they consolidated the hot find out they're printing money, right? And I'm either, you know, so let's just, I guess, analyze that analogy, which stage do you think we're at? Are we still at, you know, are you seeing lots of decks flying around with proposed data centers that are missing one of those sort of three things? And if only we can get the third one, we're all going to be kind of millionaires overnight type, you know, where will we ask in that kind of, what was that place behind us? And we're now in the heading towards the big capital destruction event when demand is the thing. I don't know and I think it may depend whether you're on the tech side or the power side of the equation, right? That there is a linkage between the tech and the power stories right now. But I'm not exactly sure that they're moving hand in hand at the exact place at the exact time, right? You're seeing a lot of innovative power companies kind of move to market pre revenue, right? Or, you know, some, you know, some pre revenue, some P profit, but that, you know, some of

23:41the new gliers, some of the geothermal, some of those types of technologies, which you're not seeing that necessarily on the tech side. I don't follow tech, as closely as I follow energy. But if anthropic IPOs next month, they're making billions of dollars a quarter, right? In video is not pets.com. I think in video, I made 100, I think they were getting 100, was it a billion dollars a day last quarter in revenue? Is there something crazy like that? And I mean, you've got this, there's a familiarity too with the players involved, right? You've got people you, Jensen, Sam, Dario, like these are all first name celebrities in the conversations of normal people, right? In the same way that some of the, you know, Aubrey McClendon was always Aubrey when people were talking about it. So, so Elon would be another one in the tech space. That hasn't like, there's not a lot of power CEOs that one could say John, and you wouldn't

24:42know that you were talking about John Ketchum. Yeah. But, let's, I mean, if you do the thought experiment, right? So you've got, this all starts off with an enormous amount of money flowing into these asset managers, private credit, right? They are, this kind of yield seems really attractive. At the same time, you've got this huge, we do know these people by their first names, right? You know, huge expectations over AI. If, I'm just trying to, if the, if the expectations are not met, what happens to all these power contracts? To be built, it's, this power infrastructure all gets built and then sort of stranded. And there's over capacity for 10 years, or is this sort of the safer part of the equation because we need it anyway. I mean, I'm just sort of, you know, if you're going down the, if we're going down the shale route, there's a long period of over capacity and very low prices and consolidation and so forth. I'm just sort of trying to figure out what, there's, at the moment, if you've got a decent

25:46plan, how to be innovative about providing power to these data centers in a narrative where the same goes for the people building power grids and all the rest of it, right? Their valuations are through the roof. I'm just trying to understand, like, how, how much of that is a, is narrative and how much of that is sort of even at this rate is still pretty, pretty decent. If you can end up owning it through whatever is going to be the over capacity period, it's still a pretty good asset. I'm of the mind that there's, you know, reasonable chance of overbilled or whatever, right? But I think that in time, you eventually use it, right? And I go back to the railroad analogy or the highway analogy that sometimes things get built, sometimes they get overbilled, but eventually it gets used, right? And if we go back to the late 90s, when you had the merchant gas build, we built too much gas generation, right, with Enron and all of that. And then, and that's what's interesting about what's happening in power right now, right? It is after that merchant gas build, there was a shake out, there was too much gas fire

26:48generation, too much power. And then we started outsourcing demand because of the, we started outsourcing industrial process in whatever else. And so for 10 years, we didn't have power growth in this country, right? And now all of a sudden, we retired a bunch of gas, we retired a bunch of coal, we retired a bunch of nuclear, we retired a bunch of this base load generation, built a bunch of intermittent renewables. And data centers are showing up saying, hey, I need more power and I need it on an, on an as needed basis. And that is kind of leading to this build, that this boom now. I think as you get into those project specifics, that's where I think the power sector is carrying more risk than the tech sector. But in both cases, it's going to get into a name specific, you know, rather than the whole sector, right? Like, if I'm building most of the gas fire generation right now, it was being built by utilities who can rate base it, right? Much of the build out how to being outside of the utilities, we're talking about, there's

27:49a corporate contract on the other end of it. Nobody's building merchant. The question gets into, you know, how those contracts hold up and if there is overbilled, you know, what happens and how long does it take for us to consume it, right? Yeah. So, and as you say, importantly there is those, yeah, as long as those contracts are with those, some of those big entities and those big entities are still around, right? In other words, sort of shoving my question, what were the best types of contracts to hold when if you were the pipeline company and there was these busts? And secondly, when you come back to it, it might make a lot of sense at the moment to be building a data center, to be able to power to a data center, way out in the middle of nowhere in West Texas. And maybe this isn't how power could work and as an engineer screaming at me. But if the chip started falling, I'd quite like to have my power sent station next to a big population center and all the rest of it, right? And be quick connected, say that, okay, well that data center is not there anymore, but

28:51at least I can sell it, sell my power. That's a lot of the behind the meter projects that's what you're doing right. And so a, a, a, so opassal, for example, opassal and meta are building a opassal electric or working on a data center out there that is using gas reciprocating engines. And after the data center and the gas reciprocating engines are completed, opassal electric will take ownership of that fleet of gas reciprocating engines, which then act as kind of a peaking instrument that if opassal needs the extra power, it can call that service, right? And you're seeing some of those things where as we were talking about each of these data centers once the grid to come to it, right? And then you've got this kind of peaking instrument because it's generally gas, you know, every time that whether it's fuel cells or reciprocating engines or turbines or whatever else that can be used by the power company if it's got those relationships in place.

29:53I don't know about, you know, some of these, some of these projects further away from demand, but I think those are some of the, I'd have some questions around those, I guess, would be a way to put it. And then I think some of the others, if you're getting into offtake with companies that we've never heard of, you know, some of them we will maybe have heard of in years to come. But that's when that, not all offtake is created equal, right? And you and I can go and sign an offtake agreement for somebody, but there's a pretty good chance that you and I can't cover whatever we say unless things go really well. Maybe you can. Yeah, they say, is that a, is that a pressure that's necessarily keeping the scale quite large? Because there's a pressure for data centers to be bigger and greater, right? That's a function of not, not having lots of small ones with all the headaches and the rest of it and the, the railing meeting factor is that power into connection. So that's a, that's a pressure to scale. You're at the gigawatt multi gigawatt scale and suddenly that means actually unlike the

30:54shale analogy, there's just no room for you and I to go by, cobbled together some a courage and try and flip it, right? The scale here is just so much greater. Yeah, and I think you're, I mean, that's like, there's tons of data centers all over the country, right? Yeah. And these are smaller kind of cloud oriented data centers that haven't kind of the term that I've been using is AI oriented, right? That the AI oriented data center just requires that much more power, that much more compute whatever to do the things that AI is asking you to do. And you need land and that's where, if you're thinking about kind of that land flipper, there's a ton of deals, you know, where you see, you know, a data center project that when you look at the detail, it's dirt and West Texas, right? And so there's some speculator activity there. There's a bunch of others that are real energized land products that are projects that can actually kind of, that have the tenant that have the funding that have all of that. If you don't have all that lined up, it's just the speculative is, you know, you would be signing a lease to drill a well in West Texas on nothing but go faster.

31:56Yeah. Yeah. How, so taking it a step further. So how much of the moment, if you think about, because it seems to me that if I'm in, you know, the further, I guess, upstream I go of all these data centers, right? The AI oriented at once. The more it becomes my solver for any NPV, any economic, any pitch I want to make, right? Like, hey, I want to go build offshore gas, you know, in West, in off Louisiana because of gas and data centers, right? Like, are you seeing that money flow all the way through to the hydrocarbon chain if you'd like? Are you going to, you know, is there now, because lots of people would be Jeff Karing who has been promoted, right? Has been bemoaning the lack of financing going into upstream oil and gas. Are we starting to see that happen now with power demand as kind of that driver? I don't think directly you've seen it on kind of the public equity side where we started

32:59seeing some of the public equity holders move into gas exposed names. A lot of the ENP companies, like we were talking about earlier, they're met in money. They don't need access to capital partnerships right now. And so, but you haven't seen, I'm trying to think, you haven't seen the deal that goes all the way up where say Google is buying molecules of gas underground that it plans to eventually convert into intelligence, right? It's doing, and I think that's kind of an important point, right, that the hyperscalers want to be in the hyperscaler business. They don't want to be in the energy business, right? And if one of the more interesting kind of landmark guilds signed over the past couple years was Microsoft's deal to bring back three mile island, grain energy complex, and Pennsylvania for nuclear, right? If Microsoft wanted to buy a constellation, if they wanted to be in the power business, they would have bought a constellation.

33:59But they didn't, right? They signed a PPA. So all the risks is with constellation. Constellation has to bring back that power plant, that nuclear power plant. Microsoft said, hey, if you bring this back, we'll buy the power from you. The only example that I can think of a tech company of a hyperscaler going into power was when Google bought Intersect, which I think was a one off for a couple of different reasons. But the power companies, I'm sorry, that the hyperscalers have been very careful not to get into the power business. And if that's something to tell me, they said, he'll, all I want to do is buy power just like you buy power at your house. But I can't because the power sector isn't prepared for the demands that I have. How, let's fast say, so a couple of questions there is. One is the other thing is, and it is slightly outside of the podium, but the, there's always this expectation that the Amazon, the Googles, the Teslas, etc. They're going to get into power trading game. So be great to understand that.

35:00Have they thought about that optionality and all the rest of it? And then I'll come onto my second question. I have heard that they're getting in, that they've got power, you know, teams of power experts, just because they're having to sign the PBAs. Yeah, they've got to understand power in a way that is different than they did, you know, 20 years ago. I don't know the level of their commitment to power trading and things like that. Yeah. Now, at the moment, I guess I'm kind of coming up this slightly jaded point of view, which is sort of my continued expectation of a big private credit bubble that blows up. And somehow now there's this linkage and essentially energy being the ability to convert these speculative loans into sort of more 15 year bond like returns that can then get sold to the public markets because they've converted, you know, hope in AI into long-term power assets, so to speak. That's probably my slightly pessimistic midlife view at the moment right about this.

36:02When you talk to these hyperscalers, how worried are they that they won't get the power that they need and that the power that they need is going to cause such societal issues that, you know, it shuts down. I mean, do they tell me about it? What are they saying about it? I mean, I haven't, I guess to be fair, I haven't spoken directly with hyper-scale servers and I speak more often with power companies that I do hyperscalers. The conversations that I've had, I haven't heard fears of we're not going to get the power. I've heard, you know, I guess stress or concern about getting the power in the timeline that they wanted or the price that they wanted, those types of things. There is the kind of the existential threat that we continue to see in the news and hear about that, hey, if we don't, if we the US don't get to superintelligence before China does, then we're putting ourselves at a marked disadvantage. I assume that some of those say, if our computers didn't get to, they might kill us all in

37:04the process, but the US computers win versus the Chinese computers. Right. So you've heard those kind of sovereign fears of, you know, US tech sector needs to win this race. I assume that some of the tech sector, you know, is aware of, I'm sure that they're aware of those and contributing to some of that, I don't want to call it hysteria, but some of that hype. But I, maybe this is my naivete, but I feel like there's an assumption that we're going to get there. It may be lumpier and maybe we don't get there in certain places, which I think again comes back to the shale analogy, right? Some of the best shale gas rock in the countries in New York. And you can't drill there, right? Because New York made decisions that it wasn't going to do it. And so you're trying to figure out where can I deliver these projects most quickly to meet the demands that I've got. And the consumer is pulling this, right? I don't, I'm, you and I are both using AI, probably less than the coders and other very

38:06sophisticated tech people, but we're using it every day. And as long as people like us keep using it, people will keep finding ways to deliver it to us. Yeah. And the better of course is that that use translates into productivity rather than just sort of designing, deciding which drapes to put up in the end. Yeah, but I think it, I mean, the way that I see the productivity thing, that we are going to be less productive in the immediate term, right? Because we're getting so much information and some of it's wrong, right? And so we're needing to learn how to use these agents to give us correct information. And it's just like Excel, right? I can, if you give me a data sheet, I can prove whatever I want with it, right? If you give me agents, I can get the agent to give you the answers that I need to defend the point that I want, right? But then you start getting more sophisticated and that's when the productivity, and I think, accelerates, which I think is one of the big interesting things on when all of these data centers come out, right? Over the next 10 years, 2035, through 2035, we've got data center demand accounting for

39:1453% of our U.S. power demand growth. And then another 25% comes from general economic activity, right? New housing starts, air conditions, all things like that. One of my questions about that is if we're getting that much power demand growth from new data centers, surely those data centers have to be leading to increased productivity, which has to be pulling that 25% closer to 26% or 30% or whatever, but the economic activity should accelerate if the data centers are using that much power because the whole promise of AI is that it makes us more productive. Well, all the flip side of that, which is sort of the paradox, is if AI works as advertised and grows as advertised, right, that replaces all the white collar workers who then have no money to pay for their subscriptions for AI and therefore the revenues off. And you can't have, you can't have both, right?

40:16Yeah, I mean, this is where I get into my, my optimism, right? Maybe you replace, maybe there's not as many coding jobs, right? But then there's people who know how to work with agents as a specialty, right? And so maybe rather than have a team of 100 coders, I have a team of 100 agent developers, right? The jobs change, but in theory, we should all be able to accept more work as we show people tasks. Yeah, yeah. And I agree with that. It's just, I think we're all reeling from Mr. Musk yesterday, saying that his whole plan is to remove like human work and the idea of our way, way jowers and stuff. In, let's say that 2020, you're a span of your career. Are we firmly back in a period of the capex cycle in energy, right? In other words, what we've been seeing, you've alluded to it, right? It's a huge capex cycle into assets, whether that's LNG, liquefaction plants, the trillions

41:20of dollars going in. You know, where are we at in that cycle? Does this look like, is this the same scale as back in 2010, 2012, in Shells Heights? Is it orders of magnitude more? I mean, it'd be great to kind of, you know, get that lens. I mean, I think it's new and different that it is that same boom that's same euphoria, but it's a different type of money, right? And in the Shell days, your poke and holes in the ground and extracting commodity from it, right, that there's a different level of risk and a different return expectation. Now we're building things, right? And I think, you know, Jeff Curry, when he was on your podcast a couple weeks ago, talked about the heavy assets, right? And this is more heavy assets that the whole Shell game took off when money was really cheap. You know, it's 1% to borrow whatever. And so you had private equity go in. It was just a whole different set up. Now money is more expensive. It's not necessarily people are expecting a higher for longer rate environment. And people want access to heavy assets that get used in almost any situation, right?

42:25And so the money is going into building things. And you're seeing, I mean, one of the other consequences, COVID's come up in this conversation, right? The backside of COVID, everybody globally was like, oh my gosh, I don't want dependence on a single source for things that I need on a daily basis, right? And so you're seeing investments in redundancy, which I think is also going to be true on the data center side, right? You've got sovereign concerns of countries in Europe or countries in Asia or countries in wherever don't necessarily want their data to be managed by a neighboring country, right? And so you're having to build extra, you're having to build extra on power. And that's one of the things on the power story that's interesting. So, the solar is a fantastic resource, right? But it's only available during certain hours of the day. So every solar plant that I build, I have to build another solar, I have to build another plant to firm that, right? And so there's all sorts of redundancies in the system. Just in time economy to the sort of the just in case. Just in case? Yeah, exactly.

43:25And you had this series of shocks, right? So you had COVID and the world's reliance on kind of Chinese, you know, everything from socks to COVID masks to whatever else, right? Then you had Ukraine with Europe's reliance on Russia for gas. And then you've got right now the war moves with the world's reliance on the straight of war moves for things. And so people are looking for optionality in kind of everything. And money is not as expensive as it used to be. So the whole, well, there's a lot of similarities. The whole system, like all of the rules are different, right? Yeah. And so I guess part of the argument would be there is a private credit bubble crash, you know, because of exuberant expectations and all the rest of it, which is kind of natural, right? You'd probably still be, and this is my friend Eugene's argument as he came on the podcast a while back, you'd probably be pretty good buying a data center if it had a decent supplier power to hatched, you know, and only that for the next 10 years, right? I think it would be the sort of the rough argument at the moment.

44:27Yeah, I mean, whether we're in a bubble or not, like in all these things, I mean, there's always names that survive bubbles, right? That there are, there's a lot of pain for individuals and there's a lot of pain for sectors as a whole. But I mean, if we go back to the dot com bubble, Amazon's still with us, right? Microsoft is still with us. We still use a lot of email. We send a lot more emails than we used to. So the thing, the system kind of re-orges the work. The capital, the capital will get washed out at various places, but the technology is, right? The same shale is the analogy there, right? Is that if you stayed in and you ended up being Conoco Phillips, only what they have, right? Right. It's a great place to be. Great. We're still using oil. We're still using gas. Yeah. We're still using the internet where maybe we're not buying whatever pets.com was doing. But pets.com was also a zero revenue company that was trying to, it was a whole different concept. Yeah, just had a website. Yeah. Right. You know, if that is exactly. So it's been fascinating.

45:28You have an upcoming conference on the 28th of the 30th here in Houston. We do. Which ties up to the 70th. Yeah. Yeah. Yeah. Thank you. So S&P, we've got a series of conferences every year, right? But this is called a financing US power. And it is a conference that we're holding here in Houston. And we've got participants from the Department of Energy. Who will be speaking? We've got several investors. We've got several utilities, several renewable developers. But over the two days there, we'll be talking about all of these things. So we've got a training on Monday and then Tuesday, the 29th and Wednesday, the 30th. We've got a series of panels, fireside chats and networking sessions that we've been doing this conference now for several years. And I hope many people listening already planning to be there or a Googling to register now. Well, I'll put links in the, I'll put links in the share notes as well. So people can find it. I just, you know, obviously a very relevant discussion here. And then thank you for coming down three floors to my office.

46:30And then actually, Hill and I are neighbors and live three houses apart. So it's a very small or opposite thing. Exactly. There's a nice symmetry. But thanks for the discussion and look forward to having you back on in the future and see where we are. Thanks a lot, Paul. Thank you for listening to find out more about HC group, our global offices and our expertise in search within the commodities sector. Please visit www. HCgroup.com