With a round Segal end from London and Gerard read from Berlin. This is redefining.
Energy Today at Rated Finding Energy, we're going to talk about hyperscanners and the utility landscape.
Yeah, because what it's crazy is we see the growth of demand is geometric, and the utilities the supply it's arithmetic, so there's really stress on the system. But first of all, from our partner.
A b Loco Energy is Europe's premier leaser of ten foot container mobile batteries built in Europe with COTL best LFP cells. A Bloco Energy serves fourteen European countries, including France, Germany and the UK. Our Blocko's batteries can be leased for any duration between six weeks and six years and they are monitored by the Dutch award winning platform school a block O Energy. Make your life easier, make your
business more flexible. Back to the show, Lauren, But let's be correct you that you're talking about us, You're not talking about Europe.
Yeah, in the US, absolutely, because.
In Europe are not saying that's called demand growth. That was saying in the States or anything like that. Yat, you know it might happen, but right now or not.
In order to do so, we had invited, but unfortunately you could not come because of your knee operation. How it's getting better, Sliman. Our guest was.
Chris Seipel, who is the vice chairman of Whatmark Power and Renewables Group.
And really he has delivered a master's class describing the relationship between ut ties and I postcan.
Well, let's bring them on the show.
Chris, Welcome to the show.
Great to be here.
Well, Chris, we both have been working three years in LG and I've just a first question for you. Have you ever seen anything that what we're saying right now?
Not in my thirty years of working in power business. I got into this business because in nineteen ninety two I was in graduate school and I had two professors from Washington, DC who would come up who were actually negotiating the Energy Policy Act of nineteen ninety two that
deregulated the electric power business. And I got into this business because I thought that was such an exciting time because moving from monopolies to deregulation was going to change the business so much, and it was a very exciting decade as we went through that. But what's coming now is having a much more fundamental impact on transformation of the electric sector, which makes it just a super exciting time to be in this business.
Yes, because for the past twenty years there's been a lot of technological changes, but by and large, the demand has been flat. So as we're going to try to analyze what's going on, they are really two different aspects, the demand and the supply. So first, probably let's jump into the demand. And you are very kind to send me this document called AI to War twenty seven, So probably first explain to our listeners what is that crazy document?
It's a website. I spend some time in San Francisco lifes Fall meeting with people deep in a research and they all pointed me to this document and it lays out month by month path to how AI could evolve
between twenty twenty five and twenty twenty seven. And a core thesis that they have is that while you and I are probably interacting with chat GBT and experiencing hallucinations and things that it gets wrong and seeing some of the limitations of it, all of the research that's happening behind the scenes right now is oriented around actually dealing with one of the biggest constraints, which is the availability of AI researchers to improve models, and the focus is
on building the capabilities within the models themselves to become the AI researchers. And it lays out this thesis around how the pace of innovation in the AI model itself is just going to start to go faster and faster as the models are essentially able to work twenty four hours, and instead of having a limited pool of researchers, you're just limited by how much compute you can have in
having researchers work on the model. And so it lays out this picture of just massive improvements in models achieving artificial general intelligence, which is models that are as smart or smarter than humans, and ultimately all of that resulting in massive increases in electricity demand as more compute is needed to transform industries and economies.
I know nothing about AI except what I read, but I don't suffer from the dun Incog effect where liberally reads five articles and think is becoming a world expert. More compute I can buy, but with marked meat is the infamous Mark Mets. Twenty five years ago, we predicted that the Internet would don the planet and consume half of the world power by two oh ten, and of course nothing happened. So it's also realized a lot on how fast the chips and the softwares can get the efficiency.
What do you hear?
I love that reference to the Mark Mills study. It's a great case study. And for those who aren't familiar with it, there were US power companies that were telling Wall Street analysts that they were building so many new power plants because of this explosive demand in digital economy and the need for power to supply all of it.
And if you look back on that study, one of the things that got most wrong was the extent to which the industry that was managing all of that information became so much more efficient than what it was in two thousand. I do think today is fundamentally different than when Mark Mills produced that study back in nineteen ninety nine, in the sense that here's just an interesting statistic for you. The hyperscalers, the large tech companies, all of them except
for Amazon, actually publish their historic electricity consumption. And when you look at that data, even before AI was becoming the growth engine, like the large language models and data center compute, just the migration to the cloud and the need for cloud compute was resulting in essentially a growth rate where their need for data centers was doubling about every four years, and the data center industry has already gotten to a scale we're just like that doubling every
four years requires a large amount of data center capacity. But so what's different is that there's this underlying trend that's been around for four or five years. There's a race to win in artificial intelligence with a lot of companies now trying to compete in this space and add a lot of data center capacity. And so over the next five years there's going to be a very large
growth in data center capacity. But just like Mark Mills back in two thousand, as we move beyond that, like into the twenty thirties, there's a lot that could happen on the efficiency improvement side and the efficiencies of the models themselves and how much compute they require.
There are as many numbers floating around as consultants, but one that I've seen kind of coming more and more is number of one hundred and eighty gigawa of US Electrictic commitments for data center for our listeners, that's one or eighteen nuclear plants by and large, knowing this about
one hundred in the US. That is absolutely massive. And if I can give another number, the hyperscalers who've invested three hundred billion US dollars in capex into twenty five, I've all announced that for Tour twenty six that's going to be seven hundred billion dollars. Now it's not just going to be powered, mostly going to be chips. Now, if you look at the capex of a data center, about twenty percent is power and eighty percent is chips or you know, you'll tell me from right or wrong.
So how do you interpret that thirst for energy?
So let's take that one hundred and eighty gigawatts number. It would result in about a thirty percent increase in demand in the United States for electricity. It is equal to like two French power systems.
That they would have to be rebuilt.
To put it into perspective, the number that we are watching most closely is what have US utilities actually committed to? And what I mean by that is a data center has made an application to interconnect with an electric utility. The electric utility has studied what it's required for it to do to be able to accommodate that demand growth.
It has figured out how much it's going to cost to add us to the system, and an agreement has been signed, and typically the data center company is making down payments to the electric utility for long lead time equipment. So that's where the one hundred and eighty gigawatt number comes from. That's what US utility have committed to at
this point. And what's important to keep in contexts and the evolution of this is that for fifteen years, US utilities have had no growth, so they didn't have teams of people to respond to all of these requests. They didn't have a culture of dealing with growth. The regulators that regulate the utilities haven't regulated in an environment, most
of them where the industry is actually growing. Industry is very unprepared for it, and a lot of change is happening as they respond to this one hundred and eighty gigawatts of commitments.
Well, one or eighty, that's for real, that's extraordinary. But as you said, the industry has lost the muscile memory of growth and they've been mostly managing their dividends and you know, making sure the trees didn't fall on the line, but that was kind of boring, and now it turns out to be there in the one of the most exciting sector are going on. And Chris, the most extraordinary
thing is you just tooking that center. So we're not even talking about the electrification of transport or electrification of heat, which in other countries are as big, if not then data center. So I don't know what numbers are going to look like because they're going to be antification of transport at some point.
Correct part of the way I'm thinking about this is that what's happening in the US right now is really the first time that we have had a deregulated electricity market experiencing substantial growth, and this is kind of a precursor with provide lessons to be learned as other parts of the globe, like Europe in particular, where electrification of transports and heating likely become a bigger part of the story what those markets are going to go through in
a few years as that becomes a bigger part of the story. But I would say in the US, though, the large loads are not confined to just data center demand. Data center demand is most of it, but we also have chip manufacturing facilities that are under construction. We have electric battery manufacturing facilities. Those are all large loads that are of a scale in size that the electric industry hasn't had to deal with in the past.
And the interesting thing is that those clients can pay because what you used to have with big industrial loads is that those who are like large volume, low margin, like you know, steel or refining, where every sense petkida whateur would count was here. The margin on computer is so ginomous, and those people are willing to pay a lot in order to get their power fast, which was not probably the case before.
Yeah, that is a super important point. The reason that I think that is so important is that you have a buyer that puts enormous value on something, namely electricity, that they're not able to get enough of right now. When you have those type of conditions, that is the type of condition the results in innovation that is difficult to predict. We don't know exactly where it's going to come from. But when you have somebody that values something so much they can't get that's when we see innovation.
That's part of what makes this time period so exciting in the electric power business. As we talk, we're starting to see some of that innovation take place.
Now. You have clearly defined the sector in the US as really been in two buckets, the regulated part and the unregulated part, which probably reminds you how you started your career. Let's dive first into the regulated part, which are regional, fully vertically integrated dutties, and then we talk about the unregulated market. So what's your take on the regulated utilities.
The first thing I would say is there's differentiation happening. There are regulated utilities that are finding ways to transfer themselves to become a home for data centers, and those utilities are actually becoming growth stocks, not something you would normally associate with an electric utility. But let us focus on the ones that are innovating in what they're doing and the types of innovations that they're bringing to kind of become a home for data centers. Three years ago,
they were getting inundated with data center requests. They have chip manufacturing facility under construction. They actually have all of the different types of loads I was talking about. They had no process whatsoever to deal with it. It's literally like a group of executives sitting around the table saying I'm going to prioritize that project, not that project. We're going to call this politician to tell them we're not doing the project that's in their district. And it wasn't transparent.
Not a great way to communicate about it. You know, another thing about this electric utility, I don't think they had a single load over five megawatts in two thousand and twenty, and probably by the time that they get to two thousand and thirty two thousand and thirty, one thirty percent of all of their demand is going to be coming from just a few large load customers. So to deal with this where they have to make a large investment to support one facility, they've changed how their
tariffs work. Historically, tariffs have been you average up all the costs of electric utility and you allocate that across all of the customers, with some differentiation in customer classes. What they're trying to do is develop tariffs that are very specific to this customer, where they allocate all the costs of serving that customer directly to that individual customer.
And unlike electric utilities of the past, they have all types of credit requirements, they have minimum payment charges that this data center company has to make to be able to mitigate the risk of those costs eventually falling on
existing customers if something went wrong. But what I found even more kind of transformational about this particular electric utility is that they're now able to analyze their system and have a conversation with a data center company, And this is a real life example where they can say, if you give us three hundred hours of interruptible power during the wintertime between the hours of eleven o'clock at night and seven in the morning, we can get you online
four years earlier because we can meet your demand from our existing resources. So that's like an analytical capability that they didn't have before. Is kind of an example of them finding a way to get somebody online even faster.
And now they've even gone beyond all of that to develop like a subscription model where they have acquired a site that they can put two thousand megawatts of gas fire generation on, so very large gas fired power plant, and data centers are basically bidding to subscribe to that capacity and to pay for the cost of building all of that capacity. So that's just kind of like one example of how a regulated electric utility has transformed itself in responding to this demand growth.
And we've seen recently in the price a lot of announcement of hundreds of billions of investment from those regulator utts. So I'm going to name a few juke ANTLG Southern Center point to a lesser extent, those guys are really boosting their investment like we've never seen before.
Yeah, And talking with one of the data center companies, they were telling me what their ideal energy supplier looks like. One of the things that they said was interesting was we want an energy supplier who can actually go into the local community and help us with the permitting of our data center and help change the zoning requirements that
we need at a local level. They want a partner who who cannot just necessarily deliver energy to them, but knows the local community really well and can deliver these other benefits. And when you think about who's best poise to do that in the energy world, it's the regulated monopoly utilities that have really strong political connections, typically within
their service territories. And you're seeing that, you know, most of the companies that you just mentioned are those vertically integrated regulated electric utilities.
So now switching to the unregulated utilities, and here I'm gonna name a few Constellation, Vistra, and RG Those guys have mostly been busy buying existing assets, somehow bitting that the price on a regulated market would go up, whether it's on PGM, COT or is where those der regulator market looks very, very different from the regulator utilities.
Yeah, let's step back a second and dissect the deregulated markets. In the deregulated markets, you have regulated utilities, but they only own the wires. They don't own generation supply. So that's how they're different from the regulated utilities group we're just talking about which are regulated. Their generation is regulated, their transmission is regulated in their service territories. The data center doesn't have a choice of who they buy from.
They have to buy from the regulated utilities. In the deregulated market, the wires only utilities. They represent about half of the commitments to adding data center capacity to the grid, but when they study the request to connect somebody to the grid, they only look at what investment has to be made in the grid to accommodate this data center.
They have no responsibility or obligation and for the most part aren't even allowed to participate in the generation market and figure out how to meet the generation supply to be able to supply power to that data center. So so it's creating a real kind of planning predicament for the industry. You have utilities in Texas, single state, that have committed to fifty gigawatts of new load coming onto the grid, and there's no coordinated plan for how supply
will be brought on to match it. And there's two fundamental problems. The first is you can build a data center a lot faster than you can build a new power plant. That creates a mismatch and being able to maintain demand and supply. And the second problem you have is a political problem, which is the regulated utility I describe is able to allocate all of the cost of building its power plant directly to the data centers and
the deregulated markets. The way it's supposed to work is Adam Smith's invisible hand and price setting mechanisms that result in new supply getting built. And what that means is if you look at the power markets right now in the US, our prices are at about half the level of what's required to provide a price incentive to build a new power plant, to get somebody like Vista or Constellation of the companies that you mentioned to build a new power plant. And if prices rise to provide that signal,
then prices have to rise for our customers. And that has become a big political problem, as affordability of energy is one of the biggest political issues right now in the United States.
So basically what you're saying is that something also we've seen all around the world is when the market are tight, government or at least a central command control is more efficient. Was when the market are laxed and over supply, This is where the market creates a good price discovery. And now as we are moving from one relatively laxed market to a more stressed one, it looks like naturally the regulated central command a more efficient into responding to this rise of demand. Yeah.
I sometimes have a hard time emotionally saying this, Laurel, because I got into this business because I saw the promise of drying in all the innovation that it could create. But the reality of what I'm seeing right now is that the vertically integrated regulated utilities, some of them, not all of them, are doing a better job at responding to the demand growth. And I think there's one critical
reason for this. It goes back to the story that I described earlier of an electric utility being able to communicate to somebody that if you give me interruptable power, I can get you online four years in advance. It's because the vertically integrated regulated utilities are looking at their system on an integrated basis what investment needs to be made in transmission and distribution and generation to most efficiently
support this customer. In the deregulated markets, there's not co ordination in the planning right now between how generation gets added to the grid and how load gets added to the grid, and that disconnect makes it a lot harder to efficiently accomplish this. In the deregulator markets plusters the political overlay.
Plus on the top of that, the price of equipment has gone Ballistick. You used to have your assistant BC at one that apple What now it's at three that apple. Wa So go get any type of LCE below one hundred that up and make what our those prices can't be done? Yeah? Sure, when we all know so, we'll always find this way. But you need a lot of land and some batteries. But so somebody don't mean the other day, I just puts it out on the roof of the data center, as my friend, you're gonna put
the five hundred megawat data centers. Okay, you're gonna copy the roof. You're gonna get ten mega? What well the four ninety coming? And of course that opens a lot of fantasies around SMRs and fusion and whatever, because everybody who supposedly as a solution is going to propose whatever in their book. Chris, how do you see that playing out in the de regulated market.
There's a couple potential paths that it could take, and this has important implications for investors in the sector and how things will work out for the companies that you mentioned, like the Vistras and the Constellation Energies. One plausible scenario is we actually see some reregulation of the market to some extent and allow some of the regulated utilities to get back into the regulated generation business, and some of
them are pushing for that. I think a second more likely possibility is what we will see emerge is a market that has one price for new capacity and one price for existing capacity, and we will see structures put in place in those markets where the new load that is coming into the market essentially has to sign long term purchase power agreements. That has to come from a
new power plant that has been built. It can't come from an existing power plant, and that has very important implications for the value of existing assets, and there will be an attempt by politicians to kind of have this one price for the new capacity, make sure there's overall oversupply in the market as it relates to all of
the existing capacity, and we get differentiation in those prices. Ideally, we would see an attempt to try to make the market work and develop structures that result in better connection between generation planning and transmission planning, but I think it's likely that politics will interfere with a plan that does that, where you would see substantial price increases for all customers. One other thing to note, just on the regulated sides too,
is just on the transformation in the innovation side. Because of the pressure on regulated rates, we're starting to see things like utilities pushed by the regulators to embrace more things like grid enhancing technology that get more capacity out
of the existing grid. It's causing a lot more pressure on developing mechanisms and structures that create more opportunity for distributed energy resources and things like that, and you can see the early signs of that starting to take hold, which I think is going to create opportunities for other players to come into the space.
Yeah, because they're old models of utility used to be what can we build, what can we build, what can we put in the rates? And probably at some point they're also going to be reregulated in the sense that they will need to account for a much more efficient system to be run and just not piling a passet.
There's so much to be built that they have a wrong growth story even as they become more efficient, and when they didn't have that growth to do it, there was things that would make them more efficient might take away any growth opportunity they had. So it's really a new kind of paradigm for the regulated utilities.
Well, Chris, from what we discuss, it looks like there is a chinomus investment opportunity at every level. Where is it going to be.
The scale of the opportunity that exists for investment in new sources of supply? We need all supply sources, So this is an opportunity for investment and gas power plants. This is an opportunity for investment in batteries, in solar and wind like all resources are necessary to all hands on deck right now, and the scale of that opportunity is larger than it's ever been during my thirty years of working in this business. But that's where the most concentrated opportunity is right now.
To my friends and listeners from the investment funds, open your checkbooks, game on exactly. Chris, thank you so much for coming on the show.
Thank you so over.
What's your thinking?
I can literally not comprehend the demand for power from the perscalars. I recently had a conversation with one of the top guys at Google and he told me our budget for two twenty six for data centers, and of course that includes chips and everything, but our budget is one hundred and eighty four billion dollar. So I said, oh wow. They no, No, you don't understand. One hundred and eighty four billion dollar for Google alone. That's more than the Russian military budget.
Yeah, I mean side, the numbers are off the charts absolutely.
So I say, yeah, but all this AI you know is the I I say, no, No, it's not about the AI. It's just normal cloud it it's way more so automous vehicle. It's YouTube, but we just can't have enough compute so er Yes, but it's just the normal cloud development that is so power hungry. And the interesting thing is that if AI does not materialize the way everybody's thinking, they will repurpose their data centers. So this is a real demand and it's happening.
Yeah, I'm not worried about the electricity side of things either, because worst case, let's assume you're in an AI bubble and a bus, that power will have other uses. At the end of the day, we are electrifying society and that power go to over the places.
Yeah, so we'll see it in China, we'll sit in the US, and we're all eager to see in Europe because right now it has been a bit of a misery in terms of demand of all those years. So hopefully, even if it's only partial, that demand from my postcalor is going to come here.
Yeah, well, it all yes coming here. I mean, at the end of the day, you're talking about Europe here, It is coming here. But the demand in Europe is more on the cloud side than it is the AI side. But doesn't that change come forward, No doubt about that.
Absolutely. And look I can tell you because I'm checking electricity map all the time the stress point. I can tell you there's a stress point right now. It's Norway. Norway used to export a lot of power and now they need to import, which is crazy. And maybe it's because the hydro season is not very good. And that's also the power of interconnectors. And of course we took about batteries morning an evening, but interconnectors are also extremely important to give resiliency to the system.
Without a doubt we need to and by the way, I think the come back to the Norway situation. What's interesting about this is maybe the Norwegians will actually realize that interconnectors are not just win laws, they're also a win win And what I mean by that is huge amounts of sort of I say, anger from Norwegian customers about the power price is going up because of the fact that Norwegian our has been exported to Britain, in Germany and now actually to the way around exactly.
So we want to think. Chris, really a great conversation and job I hope you on this is getting better and I took to you next week, look Quarters.
Thank you for listening to Redefining Energy.
Don't forget to rate the show and subscribe on Apple Podcast, Spotify, or the platform of your choice