Grid utilization vs expansion: The 100 GW debate

So our analyst, Nick Zenkin, just published a white paper Breaking Them Down, the first comprehensive look at how these tariffs are actually being designed across the country. And what stood out to me is how quickly the industry has moved to protect against risk, long contracts, demand minimums, big financial backstops.

But we're not really designing these tariffs to capture the upside either, the flexibility that data centers could actually provide to the grid. And that trade-off is gonna matter more and more over time. So if you want to understand how this next wave of load is gonna get built and who's gonna pay for it.

I'd really recommend taking a look at this white paper. You can find it at latitudemedia.com/slash research or just find a link in the show notes. I think it's gonna be worth your time, particularly if you're in this market building right now. On to the show.

system we've built sits idle much of the time. A new report from Braddle suggests the grid is only being used about half the time and that better utilization could unlock 100 gigawatts of capacity while saving ratepayers tens of billions of dollars. So the question is, do we build our way out of this moment with more transmission, more generation, more steel in the grid?

Or do we use what we already have more efficiently, more flexibly, more intelligently? The obvious answer is that we need both, but there's a surprising amount of disagreement on this front. and we're gonna wade in. A look at grid utilization versus grid expansion is coming right up.

And because FlexGen's solutions integrate with a variety of hardware, teams can adapt as technology and energy strategies evolve. Discover how FlexGen is powering more strategic energy systems at Flexgen.com.

uh did not entail the complexity that we have today. And the focus during that decade of the cloud build out in the 2010s was really around energy, right? It was around megawatt hours. and, you know, procuring, you know, via, you know, PPAs, largely vir virtual power PPAs, uh, as many megawatt hours as you could get. capacity was an afterthought. No one was really thinking about capacity enablement, time to market, um, resource adequacy. That just wasn't part of the the calculus.

And I I remember, you know, this is around twenty twenty. This is even before AI took off. Just looking at growth rates through this decade, if it if you we just kept the same growth rate that we had been on in the prior decade. It was pretty clear that by mid decade we were gonna have a problem, that the size of these additions every year was going to continue getting larger and it was gonna start to cause constraints. Of course.

Then, you know, Chat GPT three comes along in November of twenty two and then it becomes really clear, you know, in the first half of twenty three that that we have a real problem. Um and I remember having a discussion with some folks at Microsoft early that early in twenty three and

There was a debate about, you know, whether we can run out of ships or run out of power first, right? And it was I'm just sitting there in the back of the room chuckling. I'm like, guys, we're run out we're gonna run out of power before we run out of ships. That's that's where the constraint's gonna come.

at best we get a two year planning cycle for where we're going um with load and you plan power over thirty year cycles. The two just never matched. And I think we were all sort of in that struggle from I guess 2021 to 2023, those of us who could see what was happening to try to explain the timeline by which

Power, which is not a real commodity, you know, and the way that the rest of the tech world thinks about commodities works. Um, you know, and then Brian left and just solved the problem on the other side. I would wait. And then I left and I'm trying to solve the problem on the other side too.

global carbon conversation, you know, our sustainability role, our role in driving innovation, um, in spurring new technologies, in creating a decarbonized grid. And we miss the local conversation about what our Fit Print was doing. And so I think that marginal, large margin between this sort of theoretical abstract conversation about what our capital was going to do in a decarbonization rhetoric versus what our capital needs to do at a very local level in the power system in the grid.

That is the leapfrog that I've seen over the past two years, really. Right. So I would say three years ago at Sarah Week, it was still much more a global conversation driving global financial markets driving new technology. Last year I saw it sort of with the onset of the nuclear conversation, I think, it became a little more grassroots. this year it's very clear that this is about local impacts, right? And and interestingly enough, you're that's juxtaposed against

What's going on in the Middle East right now, which is a global geopolitical impact, right? So it's it's a very different change. I would say that. Ви за текду. fail to get the local conversation right, which is why it's great that Brian is doing what he's doing, because his entire focus is the local community. Uh, we we fail to get that local conversation right, not because we're doing the wrong things or because we don't care, but because we've been

stuck between these two stratospheres, I think, for quite some time, you know? And I I I'm hopeful that the tip of the spear is going to be more about the local community, more about the local power impacts, and less about sort of the global theoretical. space moving forward. Um, but you can see that transition and that conflict on stage big time here this week.

good support at the community level. Uh and this has always been a challenge. Uh and we we've seen it, you know, most notably in Europe over the years of getting this stuff built. Um, but but clearly that's become, you know, the topic du jour in the United States as well. is, you know, going to ask these communities and and and really having to have a frank conversation with them about why they should want this, you know, in their neighborhood or in their uh county and um

And I think there's a lot of misconceptions out there about the benefits that these bring. And, you know, the the my my favorite story from the last several months is, you know, we were in a fairly rural county uh i in Georgia. And we were talking with the the head of the county commission and he said, you know, he's we're a rural county and we want to stay a rural county and that's why we want a data center.

Because if, you know, someone comes and builds an autofactory here, there's gonna be five thousand people that move to town and we're gonna have to build a new high school. And if someone builds a distribution center on this site, you know, we're gonna have traffic twenty four hours a day. He goes, But a data center is perfect. You know, it brings jobs, but not too many jobs that we have to build a new high school. The high school we have just gets better.

Uh it brings tax revenue. Uh and so I I think a lot of this is about educating communities about the truly the benefits that these things bring in terms of jobs and and tax benefits. Um, and then also working with the communities to find out like w where is and is not the right fit for these things. Cause not every community. Is the right fit for a data center, but we have to build them. So it really is about the kitchen table politics and and I mean literally kitchen tables.

The Cove Relief team sits at kitchen tables every week, uh, speaking with landowners and and community leaders about what sort of benefits this AI infrastructure build out can mean for these communities.

Sarah Week is the Super Bowl of energy conferences. Let's think of transition AI as like the NFL combine or something where these ideas get tested uh and brought into the real world.

Hold on, are we as short as people say we are? Um I want to interrogate that a little bit. But um, you know, this report concludes that like, you know, the grid is only used half of the time. And so before we spend trillions of dollars rebuilding it, maybe we should step back and ask. whether we're leaving capacity on the table. So we've of course built the grid to serve peak demand, a handful of hours out of the year, which means a huge amount of infrastructure sits unused.

a lot of the time. And so the argument that Bretel is making is if you add load in the right places at the right times or shift demand to better align with existing capacity, you can spread those fixed costs across more usage And that puts downward pressure on rates. And so Braddle's modeling shows that like that this could unlock another hundred gigawatts of.

capacity and potentially save a hundred billion dollars over a decade. And so I don't think we're talking about eliminating new infrastructure. It just changes kind of how much you need when you need it. So I want to interrogate this analysis. First let's just start with the hundred gigawatt number. Again, Brian, this is literally what you do. Go out and find ways to unlock capacity. Does that seem accurate to you?

um, then you can add a lot more capacity to the system. And so I think this focus on utilization and this conversation is super important. And, you know, part of that work was I think spun up by the Utilize Coalition, which has just formed Uh that's focused on this issue. But when you you have we have this conversation about affordability, like the only answer to affordability is improving utilization. Because with the current capital cost and you know inflation impact.

to building infrastructure, there is no way if we keep utilization the same that we're not gonna drive rates up, regardless of, you know, how we structure you know, rate design and how we, you know, isolate costs to different customer classes, um, you're still gonna face up a pressure on rates if you don't do something about utilization.

the regulations or the policy structures to think of that as an asset that needs to be delivering speed to market, needs to be, you know, utilized above sixty, seventy percent. It's it's not built that way. And I think we're As a country with all this new load coming on, at real risk of just building a bunch of loops and building the same old system we've had for a long time. And you can ring fence those transmission costs. And you can ring fence the generation costs for for data center growth.

But that's only forty percent of where grid costs are gonna go, overall energy costs are gonna go over the next you know, five years. And so when you think about bringing on electric vehicles, when you think about new CNI, new residential growth, you know, fifty percent of all rates have been subject to increased distribution costs.

That's a huge aspect of as you know, as Brian said, upward pressure on rates that we're just ignoring, right? And so I think that there are regulatory ways to ensure And and I will say this by also saying is Brian being the architect of one of these and me being the architect of the other, we never went to market as hyperscalers trying to get residential

you know, customers to pay for for our system. Um, but there's a regulatory process to ring fence that. There isn't a regulatory process in place right now to make sure that utilities and Um, you know, overall D and D owners are doing absolutely the best that they can to ensure that that is the most utilized asset and that they're investing in the lowest cost way to get customers online and to make the the grid interoperable.

And I think some utilities are doing it on their own. They don't need it. You know, they see they see the light and they're and they are putting that customer sentiment forward. And some aren't, right? But it's not uniform across the grid.

As power demand rises, flexibility and control have never been more critical. Discover how Flexgen is powering more strategic energy systems at Flexgen.com or click the link in the show notes.

So I think some folks seem to believe that this obsession with utilization and distributed resources is kind of a small ball idea for a historic moment to build out new transmission, reform markets. I wonder, I'm sure you guys know like people who are involved with the Utilize Coalition, et cetera. I mean, do you think this is a fair characterization of the utilize proponents that like it's it's small ball?

Especially the, you know, vertically integrated utilities. I mean, their default is, well, I want more capital spend and more rate base. Like that's how I make money. So when you come and bring me a DLR project or a VPP or something like this. They look at that and go, well, I don't really make money on that. And you know, our response to that is, well, what you make money on is load growth long term.

And so if you are able to say yes to more customers, because you can take some of these lower capex more efficient solutions that lead to greater utilization, you're still going to get long-term capex investment in your system. It's just you're now bridging to when that capex is sort of realistic to deploy because the fact of the matter is

If your answer to every new customer is, well, I'm gonna go build a combined cycle plant and I'm gonna go, you know, upgrade a 500 KV transmission line, your answer is always gonna be five years out or seven years out or 10 years out, right? And that's not good enough for this market. You know, I I still get calls from CEOs of, you know, some of the big tech companies and foundation models and everyone else saying, Hey, do you have any power twelve months from now?

Do you have a site and I was like, well, no one has power 12 months from now. Like, you know, like, but it it happens all the time because they they that's sort of as because Caroline said earlier, like their planning cycles are still eighteen to twenty four months out, right? They're they're not, even when we see all these big announcements about I think the latest one I saw yesterday was or this week was OpenAI and Helion doing a hundred gigawatts by twenty thirty five. Great.

Um, I'd take the under, but uh, you know, it's th all those things are not solving the problem that they actually have today, which is they are trying to turn on a lot of GPUs in 2026 and 2027 and 2028. Um, and what what I think this whole utilization discussion helps with is hey, there is available capacity if if you know where to look. And it's it's difficult and you know, at least for the

sort of the non u energy person to think about how we would orchestrate those solutions. But they are there.

of demand for the AI transition. And I would challenge the conversation to be, how is this country going to take the capital that presents itself in this load growth? And invest in the system that is best for everybody else moving forward, best for the communities, best for the residential customers, best for the other C and I customers. And I think if you take that posture first.

Um, you start to look at to Brian's point those foundational elements that we absolutely need, irrespective of whether it's a hundred gigawatts or seventy gigawatts or thirty gigawatts, right? I I think one of the problems that we always had at Google, and Brian, I'm sure you shared this, was that I was always looking for a way.

To get our energy strategy decoupled from what was an erratic load growth projection from where we were going with cloud or where we were going with AI, right? And so, and in part that's what you're you're doing with Cloverleaf. But I also think we should challenge ourselves as an ecosystem to say, what's the grid that we want that gets us out of this frenetic well we're 50 megawatts short on that node, so we can't interconnect you for six years. Whereas if you were

two hundred megawatts less, we could interconnect you tomorrow. Or we have one solution for interconnection right now. And we it's pretty old. It's the same thing we've been doing for fifty years. And we haven't

thought about or really at scale deployed alternative solutions. And we don't have any solutions for a you know coming society that has mass scale electric vehicles, that has the smart home, that has, you know, rooftop solar and storage at scale, you know, which is all coming down in price and all coming down in you know, accessibility. And so I think the conversation needs to be more about that in general. In general. But

whether this is hype or not, I don't think it's hype. I think it's a both end to you said, Steven. Like we need we need to build new power for the long term. And and actually the problem with the the the um scenario that Brian set up is that Hyperscalers have a really hard time in planning for more than five years. So we get to like, what do you have in two years? Well, I don't have anything. And then we come back two years later and say,

Well, what do you have for the next two years? And you come back two years later and say, what do you have for the next two years? As opposed to saying, well, what should I be building now? And of course, you're going to be building new power plants absolutely for the next five years out. But in this race to gigawatts, what is the best thing we can actually do to one, create that bridge power that's clean and affordable?

And also is going to leave the grid in a better place, right? Everyone benefits if we utilize the assets we have and everyone benefits. if we redirect our focus into saying, how do we get the most capacity and and and the most savings out of out of the existing system? And that's gonna be through deploying things like

you know, standalone storage behind behind the meter assets of the CNI and residential space, dynamic line rating, you know, gets, non wire alternatives, whole things. That's how you're gonna get the most out of this system. It's a win win because Is to Brian's point. We're still gonna build those power plants. We still have to build those power plants. You know, that's going to happen.

Um, but if we don't do this now, we're gonna build a Frankenstein grid and we're gonna build a bunch of loops to nowhere. And and that's ultimately that the capital waste I think that could happen. I'm not as worried about generation capital waste. I think it will get there because I think training will continue. I'm much more worried about grid capital waste.

Uh Southwest Pennsylvania, four gigawatts at uh I forget the 17 billion. Um you've had uh the NScale project, you had Fermi before that. I mean you go down the list and you can quickly get to sixty, eighty gigawatts of these sort of mega projects. uh that I'm not convinced are even financeable because what's notable in all these projects is the one missing piece is the actual customer signing up for a 20-year off take.

Sure, you have a developer who wants to build it. Sure, you have a big bank or sovereign that wants to lend money to it, but you don't see the customers right now signing up. And as Caroline's saying, The hyperscalers really don't plan that far out for this stuff, such that they could underwrite a$33 billion project. Not to say that none of those will get built, but it's a lot harder than people think than just a press release or a groundbreaking to get a project like that off the ground.

Um, I mean I can drive, you know, an hour from my house out to the Washington coast and find the old Satsub uh cooling tower that was supposed to be a nuclear power plant that's now just a giant concrete edifice that you can see from the road. Um, which is a fun little fun fact when you're driving the family out there and you can point that like, hey, we tried to build a nuclear plant and we failed because it's actually really hard to build big things in this country.

So I think people are underestimating how difficult it's going to be to really execute those mega projects. that, you know, this path of least resistance as Caroline was just talking about, is is far more elegant, it's faster, it's cheaper, and it's more sustainable. So it is the right solution. Again, not that we're not going to build power plants, we will, but you know, we should build a whole lot less of those uh if we're focused on this issue of utilization.

the timeline for building out transmission into central Ohio is longer than that. Right? And so I think what what's gonna happen is There's only a certain book of turbines that can and will be built in this country because there's only certain companies, one of them I work for now, who actually know how to build them and the workforce and the labor to do it. And if those are built

Responsibly and done well and integrated well, then that's going to be good for long-term reliability. But what's going to happen for everyone else?

which is not good for the community, not good for the environment, and ultimately, you know, financeable in the short run, but I think long term has some some real pain points. It's not good.

the energy community, um, I think articulated by um Xiao Wang, who's a research scientist who's very active in these debates. Um and his argument is basically that like not all capacity is interchangeable, right? If you've got a So say a one gigawatt data center showing up in a specific location, you can't necessarily solve that by turning down A gigawatt of load somewhere else on the distribution system. Like it doesn't actually solve your constraint. Does that resonate with you?

Because it's not true that you can't connect any data center for ten years in this country. There's lots of places you can connect them. So understanding the transmission system first, and this is how we do our planning and siting is we do transmission load flow analysis as we look at withdrawal points and then we start to study what resources

are available and I w we'll be talking about this at the Transition AI conference around you know how we specifically have done that with level ten. Uh but there this is where the I think the lack of sophistication Of a lot of the folks that are trying to develop these assets comes into play. Because yes, it is true. If you just take a dart and throw it

somewhere on the transmission system and then complain because the utility won't connect you for ten years. It's like, okay, well, if that's the level of sophisticated planning you're doing, then you're right. You you you need to build your own gas plant behind the meter. But I I just don't think it's true that those resources that we've been talking about. can't contribute a material amount to enabling load growth if it's done strategically.

of where things are placed, where their demand is, where their load is going. So just that visibility into their system. So yes, I think that's true. I would but I would I would push back on that a little bit. in saying that I think that's an easy way to explain why we don't invest in the market signals to grow the VPP as a legitimate commodity. Right? Because I think with a really strong interoperable visible system, you know, and a good partnership, um, you can work with the utility to say,

Where should I go where congestion is going to be top of mind? Where can I go where you are looking at a place? that distribution relief or distribution investment is going to free up transmission capacity. And the problem with that is not only just planning, but it's also that we don't have markets.

like value signals for this. Whether you're in an whether you're in a vertically integrated space and they do their resource planning and VPPs or, you know, DERs aren't given the platform to have capacity value within the resource plan or whether you're in PJM.

And you're not allowing aggregated VPPs to play in a special auction, or even, you know, in ERCOT, where there actually isn't a price signal for an aggregated VPP yet, outside of just, you know, an energy signal. And we're working on that in ERCOT. But I I think it's a also a a market signal situation and I think it's sort of a cop out for not investing in that because

you know, like what Paige is doing over at Tapestry, they're actually trying to figure this out right now, right? Like how can you actually create the market signals for placement where maybe it's not a one-to-one swap, but it's probably at least A two to one swap, right? On on something like this, because you actually know how to operate your grid so well.

the the barriers to entry from a market standpoint are are fairly low there, at least conceptually. Again, I challenge on the financing side. But the ability to deliver that actually is a lot harder than people think. uh, because of the financing, because of the scale. Whereas the market barriers around VPPs and other solutions are actually higher, as Caroline was noting.

But the ability to deliver them is actually really easy. I mean, there the is, you know, it's there's very little friction in actually deploying those solutions in the real world. you know, from a physical standpoint. Um, so it's it's not surprising that the market sort of coalesces around what seems like the easy path on the front end, which is just build big power plants.

and is not paying enough attention to, you know, this this harder path uh because of the way that markets are designed, but but really is in the long run the faster, easier path if we can coalesce around getting the market signals right here.

That's the really irresponsible thing that's happening right now. And that's the better com that's the better data point for me in terms of if you're if you're looking for the argument.

I think it is gonna be a both and on on building new power. And I think that what we're missing is sort of the the real trade-offs long term for the community notally um and from a financial perspective between those two that Frankenstein approach versus like, hey yeah, you could get I mean, I remember sitting with and said utility at one point and we couldn't interconnect because of Two hundred megawatts.

Right. This was a multigigawatt project. We couldn't interconnect because of two hundred megawatts. And finally they looked at me and when we had this conversation through and and the gentleman said Actually, if you invested in X, Y, and Z on these three distribution units, I think I could get you 200 megawatts. But that conversation had never happened before in an interconnection conversation before, right? They had never been forced to get there.

So I think that what if we had not had that conversation, if we had not pushed on that, what would have happened? We would have, you know, considered. And we never would have done it at Google, but we would have considered one of these like Frankenstein approaches, you know? And that's the trade-off conversation that I think needs to be more relevant and and happening with utilities.

Uh coal is back. Bye.

You know, like I mean, I'm a huge believer in decarbonizing the grid and I also believe that fifty years from now we're still gonna have gas plants online. Like we we cannot operate this system without it. I I think it'd actually be foolish to to try to do that, um and not, you know, for very extreme events, you know, like even the heat wave we're seeing right now in the middle of spring, um, you know, have have some amount of of flexible gas to lean on, at least for the foreseeable future.

So I think really the the debate should be around are we building new baseload gas plants that we're gonna have to run at a high utilization? Or we can can we lean on this issue of flexibility and build more more peaking plants?

So I and I know we hear that a lot. I mean, I've heard that a lot out of the current administration that, you know, and others have echoed that, but it's just not based in fact.

then you're not really using the right tool to solve that problem. Right. And so they will pay a lot of money to go faster. And it's about and this is where that this whole utilization conversation comes back around,'cause if those solutions can deliver that bridge. Th there's again a lot of price insensitivity there, which means there's a lot of solutions we can bring to market because there's a willingness to pay. But I I don't think a lot of people are going to be signing up for

$150 a megawatt hour power for 30 years. Like that's just there's not a lot of appetite for that.

across the board that we were used to, you know, for the past five years. Uh or ten sorry, ten cents of KWH, sorry. And and um if you think about that and what the value of compute was, compute is at least 200% more valuable than the cost of of of power at that time, at least. And now it's probably even more. Yeah. So there's no way that they're that

the cost of power is ever, ever gonna become that demonstrative of a barrier in terms of continuing to wanna grow and and chase market share. And we're in a market share. You're not gonna be able to say, hey, We're just not gonna run for the next couple of years. No, no, then your business is over. You're not in the game anymore, right? And so I think that's where why these utilization and and VPP um solutions

We've just never explored what's the actual strike price for that. We're just assuming that the strike price is what we've been putting forward for the past couple of years, which is like maybe 5% above avoided cost. No, no, no. I think the strike price is much, much, much higher in terms of getting those on the grid. Uh but I think the LCOE does matter to residential customers and to CNI customers.

that, you know, are going to be part of this system for a very, very long time. So we should invest in the things that are gonna make their lives better um in the short term.

Um, but part of the way we're gonna have to get it back is, as I mentioned before, a focus on affordability and doing this the right way. Because if what we're doing is just driving up everyone's energy rates. then that's gonna be even further headwind for getting big things built in this country.

Um, and that's that's, you know, another reason why I say I power plants are gonna get built. So solar, wind, gas, it's gonna get built. and storage is gonna get built and it it needs to get built and we and we want it to get built. But um There's only a limited amount of workforce that knows how to do that, right? So what can we do with our existing system when that is a it is a real factor? And I'd like to see a lot more going into developing that workforce.

And for more in-depth stories on all the topics we cover, go to latitudemedia.com and you can get transcripts of the show there. Thanks everyone. We'll catch you next week.