A major fault line runs through the data center debate: how many will actually go off-grid?

There’s a lot of enthusiasm in Silicon Valley for breaking free from the grid. But even the companies pushing hardest for off-grid projects now admit they want a grid connection eventually. So how much behind-the-meter power will actually get built in the next few years? And what fills it?

SemiAnalysis, the influential chip and infrastructure research firm, recently projected over 40 gigawatts of behind-the-meter power at U.S. data centers by 2028, most of it gas turbines, reciprocating engines, and fuel cells.

This week, Tim Hade, SVP at Voltus, joins the show to interrogate that forecast. He thinks that timeline is a fantasy, and that the real story for the next three to five years is load flexibility driven by batteries.

We dig into the SemiAnalysis numbers, the disagreement over how much headroom the grid has, and whether declining marginal value for batteries undercuts the case for storage. We close with a new framework from Rhodium Group that scores how these choices help or hurt the energy transition. 

Read Jigar’s ode to Lorenzo Kristov, mentioned in the show.

Get your ticket to Latitude Media’s Flex Summit in Austin, Texas on October 14-15.

Credits: Co-hosted by Stephen Lacey, Jigar Shah, and special guest Tim Hade. Produced and edited by Stephen Lacey, Sean Marquand, and Anne Bailey. 

Open Circuit is brought to you by FischTank PR, an award-winning climate and energy tech, renewables, and sustainability-focused PR firm dedicated to elevating the work of both early-stage and established companies. Learn more about their PR approach and how they can support your company’s messaging by visiting fischtankpr.com.

Transcript

Stephen Lacey: From Latitude Media, this is Open Circuit. Data centers have created several schisms in the world of energy. Some see them as an emissions nightmare. Others see them as a historic opportunity to leverage clean resources. Some see them as the reason your electricity bill is climbing. Others see them as a deep pocketed customer who can modernize the grid. Some see them as a destabilizing threat to the electricity system. Others see them as a sophisticated tool for flexing load and using the grid better.

And one of the more interesting divides is over off grid versus grid connected data centers. Should they island themselves to protect rate payers and get online faster or should they use their load to be better grid citizens? That has evolved a bit as most companies pushing for off grid data centers admit they want to connect to the system eventually. But it’s brought us to a more interesting secondary question.

Just how much behind the meter power gets built in the next few years and what fills it? SemiAnalysis is out with a projection that we could see 40 gigawatts of behind the meter generation at US data centers by 2028, mostly gas. So is that where we’re headed and is it really the fastest path to power? This week we’re wading into the debate with someone who’s betting batteries or a much bigger part of the answer. That’s coming right up.

I’m Stephen Lacey. I’m the executive editor at Latitude Media. Thanks so much for being here. Jigar Shah is a clean energy investor, co-managing partner at Multiplier. How are you, sir?

Jigar Shah: Doing well, doing well.

Stephen Lacey: And here with us as a guest co-host from the American Riviera is Tim Hade. He’s a senior vice president at Voltus, which is a new title after the company he co-founded Brightfield AI was acquired by Voltus in June. How are you?

Tim Hade: I’m doing great, although I realize this is the first time I’ve ever been on this show without Katherine here to protect me. So I’m a little nervous, but I’m ready to go. Let’s do this.

Stephen Lacey: I’ll provide a buffer.

Jigar Shah: The real question is whether you can see the oil derricks off the coast of Santa Barbara.

Stephen Lacey: Right. I heard the Trump administration is trying to push for the reopening of an oil pipeline there off the coast.

Tim Hade: Yeah. I mean, we could do a whole episode on that. If you guys want to Google the Sable pipeline, it’s quite controversial here in my local community and a lot going on around that. But the answer to your question, Jigar, is yes, everyone in Santa Barbara can see the oil rigs out in the ocean and it’s I think a lot of people in the community’s mission to get rid of them. So everyone’s working on it.

Stephen Lacey: Fun fact, the first offshore oil rigs in the US were off the coast of Santa Barbara, right?

Tim Hade: That’s right. That’s right. Yeah. It all started here. So you’re welcome world.

Stephen Lacey: So Tim is a familiar voice. He was on the show last year. He was on the show with us 10 years ago and he was recently on Jigar’s other podcast, Energy Empire. And you’re actually going to be digging deep on C&I batteries with Shayle Kann on our sister podcast, Catalyst. This is starting to, you’re on all these shows that we produce. It’s starting to feel like those circular deals that tech companies are signing with each other to fund AI.

Tim Hade: Yeah. We’re all trying to figure it out together. So apartly all my friends have podcasts now, so I love being here.

Stephen Lacey: No bubble here, just supply meeting demand. Well, that gets us quickly to the Voltus acquisition. So what were you doing at Brightfield with C&I batteries that fit into what Voltus is doing?

Tim Hade: Yeah. So I guess the high level story here, and I’m excited to talk more about this, is look, I think in the last 18 to 24 months, what’s happening with the electric grid and just load growth generally has really catalyzed and come into focus. And so again, we’re in this world in which over the next three to five years, I think the foundation of the next generation of economic prosperity in the world is going to be determined. And a huge limiting factor is power. And so depending on who you talk to, something like a hundred gigawatts of net new demand is forecast over the next five years. And the way we’re going to get there is primarily load flexibility in my opinion. And so a piece of load flexibility is deploying storage where you can deploy storage. And so about a year ago, I was thinking specifically about commercial and industrial battery systems.

And one of the big problems with deploying commercial and industrial battery systems for reference, we’ve deployed a lot of batteries in the US over the last decade. Roughly 90% of them are utility scale batteries connected to the transmission system. About 10% of them are residential batteries, so home batteries, Tesla powerwalls and things of that nature. And 90 plus 10 is a hundred. So roughly 0% of those batteries have been deployed in commercial facilities despite the fact that commercial and industrial facilities are roughly 60% of demand on the electric grid. So no commercial and industrial batteries. And so I was thinking a lot about that problem. And one of the things that keeps coming up is roughly 25% of the cost of a typical C&I battery project is what I’ll loosely call transaction costs. So development, building financial models, permitting, interconnection, all that type of stuff.

And so at the same time, me and a few friends were playing around with AI and learning everything we could learn about sort of how to build agentic tools. And I think the idea was basically can we build agentic tools that will help reduce the friction in the transaction process for C&I batteries? And the answer to that’s yes. That’s actually a really good application of AI. And we did that on the commercial battery side of things and so many other companies are looking at that as part of their value stack. And it’s working really, really well. You guys know all this, but forever, one of the arguments against DERs has been soft costs or transaction costs are too high. And I think AI is really a huge part of the solution. And so we built a company to try to address that and we’re still in the process of building that technology.

But when this opportunity to join Voltus came along, I think now is the moment. And so it was an opportunity to be part of something bigger than ourselves and we’re really happy to be here.

Jigar Shah: I see merch. I see merch. The soft costs are too damn high.

Stephen Lacey: That’s a great sticker. I would put that on my car. I would wear that t-shirt. Jigar, why is C&I… I mean, Tim just sort of explained the problem with transaction costs. Why is the C&I market so damn hard?

Jigar Shah: Because fundamentally the challenge with C&I is that on the one hand, you have people who are way too sophisticated. And so they ask you way too many questions. And two years later and 14 meetings later, you still don’t have a contract. And you’re saying, “How much more time am I going to invest in this damn relationship before I get dollars out of it?” And on the other side, you’ve got 90% of commercial buildings that are less than 50,000 square feet and they’re unbankable. Think like the local church, the local food bank, the laundromat, the dry cleaner, the whatever. And so they’re not bankable in the traditional sense. And the last thing I’d say is when you look at residential, residential is funded as a pool using FICO score. And so you have a predicted amount of failure within that portfolio that you model upfront.

C&I has never been funded that way. C&I has always been funded on a deal by deal basis. So the bank wants to know that every single deal works and none of them fail, even though of course some of them do. And so moving to a portfolio wide financing mechanism is the key to unlocking C&I, knowing that 1% of the portfolio is going to fail every year. It’s fine because you’ve charged everybody an extra 0.3 cents per kilowatt hour for the power and you’ve got some reserves in there, but nobody has done it that way.

Tim Hade: What Jigar said is 100% right if you look backwards at sort of the history of the industry. But I think one of the things that is really, really exciting about this moment is for a lot of commercial and industrial customers, reducing energy costs has gone from being a vitamin to a painkiller. And so if you go back sort of 10 years and you would go out and you would meet with the facility manager of a grocery store, a cold storage facility or a data center. Again, everyone was interested in reducing their electricity spend. Various companies had different sustainability targets and resilience and reliability. All these things mattered, but it definitely wasn’t a top priority. And I think the shift that’s happened over the last 36 months has really been it’s become for a lot of those customers an absolute necessity. It’s impacting their bottom line economics.

And so I think we’re in this moment now where having conversations with C&I owners is a lot more productive and it’s moving a lot faster than it’s ever moved. But the key to this is really aggregation. And that’s where Voltus comes in. And so any single C&I facility, 200 kilowatts, 500 kilowatts, something like that, independently that doesn’t really have system wide benefits. If you can aggregate those loads together, then you can really start to play some games on the grid and unlock opportunities for large loads and economic growth. And so that aggregation piece is really I think the key to how successful we’re going to be in this next phase of the energy transition.

Stephen Lacey: All right. So let’s get into the philosophical divides around data centers. I think in two parts, one is just the status of off grid data centers, the debate around how many fully islanded data centers we’re actually going to se. And then I want to talk about the behind the meter generation projections. So I think we know where Jigar stands on this. Jigar’s been highly skeptical. And I want you to expand on it a little bit Jigar. But I mean, over the last year, there has been this recognition that there will probably be few islanded data centers and that many of the off-grid data centers are going to be seeking some kind of bridge to the grid, but we have seen some very large projects announced or under development. At our Transition-AI conference in April, Tim, you predicted that off-grid data centers, fully islanded data centers will be a specialty service because the engineering challenges are so high. So where do you think we are with practical implementation?

Tim Hade: Yeah, look, I mean, I think the thing a lot of sort of market analysis misses is how hard it is to build an off-grid data center. And so in essence, people think about this as like you build a power plant here and then you build a data center here and like voila, you’re off grid. And that’s the furthest thing from the truth. In fact, I would argue that 80% of the difficulty of building one of those projects is the power electronics that sit in the middle. Or another way of thinking about it is you’re essentially building your own electric grid. You’re building a microgrid. And there just aren’t a lot of people that know how to do that. And so you can build models and you can say like the gas turbine costs X and the core and shell of the data center costs Y and the lead times or this, that and the other thing.

But at the end of the day, you have to be able to hire people who know how to do really, really, really sophisticated electrical work in order to connect those two things. And there just aren’t enough of those people. And so again, I think off grid data centers are a thing. And there are a handful of companies. Our friend Sheldon at Intersect and the Crusoe guys and my former colleagues at scale who all have these incredible teams that are built to do this thing. And I think a lot of those folks are going to be wildly successful. But if you sort of put all that together, we’re talking about a handful of gigawatts of net new capacity added over the next three to five years. And this three to five year window is really the critical point. And so again, I think it’ll be a thing, but it’s not the answer to the problems. Or put another way, people should listen to Jigar.

Stephen Lacey: Jigar, how many fully off grid data centers do you think will get built? Either in number or capacity terms.

Jigar Shah: I think Tim’s right. But I think that the thing that was really shocking to me about the news and the news cycle that’s come out in the last month is just how tone deaf it is. Today it’s very clear that you choosing to go off grid not protecting rate payers from rising electricity costs. In fact, it is making things worse. You are making supply chain tighter. Everything is going up in price because of you. You are not adding more kilowatt hours of sales to the overall shared system. So then you’re not helping bills go down over time. And so the level of selfishness in the current news cycle is just shocking to me. They only care about their own tokens and their own compute. They don’t actually care about all of the people who are protesting data centers around the country and those people who are voting for candidates for office who are saying, just shut it all down. We’re just like tired of the extreme selfishness.

Stephen Lacey: I want to dig into that a little bit more. I think you’ve made this point on a previous episode. Basically what you’re saying is that they’re not bringing the added load to the grid and potentially benefit the grid system that it’s connected to. And instead they’re creating supply chain constraints that make it more expensive for everybody else. Is that what you’re saying?

Jigar Shah: Yeah, because they’re buying power transformers, they’re buying batteries, they’re buying gas turbines that are short in supply. They’re buying all super capacitors, all sorts of stuff that ultimately was not built for this purpose. The supply chain is not in place to build 50 gigawatts of off grid data centers. On top of that, the people that Tim suggested are the experts here are the experts and I love them to pieces. But they’re finding that every time they think they’re close to completing their design, something else is missing. And they have to add another $10 a megawatt hour to the cost of the system just to make sure this edge case is taken care of. Initially it was like, we can just run these things off of backup natural gas generators. Then they’re like, crap, we actually need to have 50% redundancy in natural gas generators. Oh wait, where’s the workforce to actually maintain these natural gas generators?

Wait, we need lithium ion batteries to buffer the natural gas generators. Wait, we’re adding 15 cycles a day. Lithium ion won’t work. Let’s go to Vanadium redox or zinc halide. Wait, that’s hurting the batteries. We need super capacitors between the batteries and the data center. Now they realize they need to move to an 800 volt bus because they can’t actually just take this stuff and make it DC. So then you got Heron Power putting out their 800 volt bus strategy of which only 20% of it exists in real life. And the other 80% is in startups that are trying to put it together. And you’re like, wait, what is the final project going to cost? And all along the way, they’re buying esoteric components that are very essential for people in different industries like semiconductors, et cetera. And they’re making those supply chains short. And so then those suppliers are tripling prices.

And so the thing is that, look, I love an engineering challenge. I do. And so the fact that these people are building an off-grid data center is such a heroic feat. And I think the world of the engineers that are working on it. But the notion that you’re going to do this en masse across the country is a deeply selfish decision. And the notion that people don’t understand that is just shocking to me in this moment.

Stephen Lacey: I think that many of them have really convinced themselves that they are doing right by the rate payer though. That they are insulating themselves and protecting rate payers if the data center doesn’t materialize or the data center is underutilized. I think that there’s a real core belief that they are doing the right thing.

Jigar Shah: Yeah, you’re right. People can make up lots of things, particularly using AI that likes to agree with you to actually make you dumber. But the notion that you’ve gone to experts who actually understand how the grid works. Those experts are in charge of those grids in those states. In Utah, they do not believe that off-grid data centers will reduce rates for everybody who lives in Utah. In Texas, they absolutely don’t believe that, which is why they’re creating their batch zero process. That is not true in any place across the country where an expert actually lives. You need an N-1 framework by which to power a data center. And I guarantee you that even these data centers that come online that Tim’s talking about, that within the next seven years of operation, they’re going to be like, oh crap, this whole thing sucks. And we’re going to abandon this compute and we’re going to move to other places because it’s too damn hard to keep this thing running.

Tim Hade: Yeah. I mean, look, I think what Jigar’s talking about though is I think when you put all this together, the reason that I’m so optimistic about the moment and I think that there’s a really good outcome here. So the one bone I’ll pick with the boss here is I do think the hyperscalers care. Certainly that’s been every experience I’ve had with a hyperscaler is they are thinking really, really, really hard about this. And I do think that there’s a recognition that the way to get AI infrastructure built is in large part based on meeting the communities that they’re working in where they are and delivering tangible benefits. And then look, I think over the past 12 to 18 months, what’s emerged is a clear pathway for doing that, which we can loosely call load flexibility, but we can talk more about what that actually means.

But the idea is basically that you can connect data centers to the grid in a way that has social benefits for everyone in that local area. And so again, I think ultimately there is that intent there to be a good citizen of the grid. And there’s a way to do that. That technology exists, those capabilities exist. And ultimately what we can end up with over the next three to five years if we do this the right way is we can have a much better grid that’s cheaper and cleaner and better for the average rate payer. And at the same time, we can add hundreds of gigawatts of data centers to the grid and everyone wins. And so one of the reasons I’m so excited about this moment is I do think there’s this win-win construct in place and that’s what you’re seeing in the market.

And so again, I think that now is kind of the moment where everyone’s economic and social incentives have aligned towards this common outcome and now it’s just a matter of execution.

Stephen Lacey: What do you guys make of Microsoft’s project Kilby in Texas? 2.7 gigawatts of turbines from GE Vernova and Caterpillar. Do you think this project will get built and is it a model that we’re likely to see? They are eventually going to connect to the grid, but it’s a multi-year project that will be off grid for the most part.

Tim Hade: Yeah, maybe. And so just to put this in perspective and I think this is something people miss all the time. You sent me this SemiAnalysis report where they’re like, there’s 40 gigawatts of behind the meter data centers and we can unpack that if you want. Yeah, we’re going to definitely get into that. I strongly disagree with their methodology and conclusions. But the important thing to know here is when you talk about large data centers, right? So let’s say 500 megawatts plus is a large data center. There are exactly zero off-grid large data centers on the planet earth today. In fact, to the best of my knowledge, there’s only been one that’s ever been built, which is Colossus, which is what Elon Musk built in Mississippi and Memphis. And let’s just say that’s not a replicable model for the rest of the industry.

Jigar Shah: And it’s connected to TVA now.

Tim Hade: And it’s now connected to the grid. But that was the only large data center that’s ever been built off grid with no grid connection that’s islanded for any long duration of time. That’s a singular example. And so again, what I would say about any off-grid data center project, and I would include the Microsoft project in here is it’s a science experiment. It’s a first of a kind project and maybe it’ll work and maybe it won’t work. But I think if you’re a hyperscaler and you’re thinking about how do you build a replicable model that can get to scale? Yeah, it’s fine to try this stuff out and see what you can figure out, I guess. But again, I think what they’re going to find in practice is these things are really, really, really difficult to build. They’re really, really difficult to operate and maintain after you’ve built them.

And a far better option is doing what we do know how to do, which is build data centers and connect them to the grid. And so I think that’s what the likely outcome for most of these folks will be. There’ll be a lot of science experiments, but ultimately what you’ll see is co-located, interconnected to the grid data centers will be a thing off grid data centers, again, with maybe edge cases being an exception are not going to be a major part of this market.

Jigar Shah: Well, and the irony is that we are trying to build a transmission line from the Permian Basin into the main heart of Texas because so many of the oil and gas rigs out there are running off of backup diesel generators and so off of very thin distribution lines. And so they’re trying to beef up that entire grid just to get people off of backup generators onto the grid. And because of this exact project, all of the Republican senators and House members in Texas are so anti-data centers that they’re now trying to block the transmission line going into the Permian Basin because it’s for the data center. And I’m like, it’s not for the data center. It’s for decarbonizing the Permian Basin so they’re not running off of backup generators. And the data centers are so hated by the entire community that the Republicans who represent that part of Texas are now against the transmission line getting built into that region, which is pushing out the date for that data center to be grid connected. The whole thing is just ridiculous.

Stephen Lacey: Well, let’s bring in the conversation about what the resource mix for behind the meter generation is going to look like. You mentioned the SemiAnalysis projections. This is an influential chip and infrastructure research shop that has increasingly been dipping its toe into energy. They showed that they tracked 40,000 power plants nationwide and concluded that grid headroom despair capacity left over after covering peak demand will turn negative across the country by 2027. And as a result, they project over 40 gigawatts of behind the meter power at data centers by 2028 with equipment demand surpassing 50 gigawatts a year by 2029. So the question is, will all that get built? What fills that gap? Is it turbines reciprocating engines, fuel cells, gas, mostly top to bottom? Let’s just walk through the arguments that SemiAnalysis is making. Tim, you’re shaking your head as I’m reading some of those numbers. What’s your initial reaction to that top line number?

Tim Hade: Yeah, my initial reaction is that everyone has access to Claude and anyone who wants to can produce a report. So look, I think if you look at the SemiAnalysis paper that you sent, it’s titled US Grid Constraints Towards 40 gigawatts of behind the meter data center by 2028. So let’s keep that date in mind while we talk about this. They’re talking about 18 months from now there are going to be 40 gigawatts of behind the meter data centers. So look, I think basically the analysis they walk through is the first point they make is there’s 80 gigawatts of demand for new data centers over the next five years. And I actually buy that argument. I think that’s about directionally correct. So there’s 80 gigawatts of demand. And then they say there’s not 80 gigawatts of availability on the grid. Now two points I’ll make about this.

The first thing is Jigar had Amit Narayon on his Energy Empire podcast I think two weeks ago. And I listened to that. And one of the things I try to decipher when in a world where everyone can write a report that sounds good is who’s the author of the report? And so Amit has been doing this for 20 years. He was the founder of AutoGrid. He’s now the founder of GridCare. He has true expertise in this. He says there’s 300 gigawatts of headroom on the US electric grid. If you go look at the authors of the SemiAnalysis report and you look at their background in the energy industry, they do not have 20 years of experience of deep technical experience. So the author here matters, right? But they say there’s no grid headroom. So the basic argument is we have 80 gigawatts of demand, there’s no grid headroom, and therefore we’re going to build 40 megawatts of off-grid data centers.

Stephen Lacey: 40 gigawatts.

Tim Hade: Yeah, 40 gigawatts in the next 18 months. And so look, what they don’t make is a compelling argument for how you’re going to build 40 gigawatts of off-grid data centers in the next 18 months. So put this in perspective. The build time for one of these projects is probably 36 to 48 months best case scenario. When you’re building an off-grid data center, you need all the same power electronics. Jigar mentioned this earlier, transformers, switchgear, relays, conduit, all these things. There’s just supply chain shortages. So some of these items have 200 week lead times. Right now in the world today, you can’t get it for 200 weeks. And so there’s no way that any project that’s not currently under construction, permitted, financed and under constructed is coming online in the next 18 months. And we don’t have 40 gigawatts of data centers that are currently under construction.

And so the idea that we’re going to get 40 gigawatts of net new off-grid data centers in the next 18 months is factually incorrect. There’s just no, I don’t think, comprehensive argument for it, nor do I think SemiAnalysis makes an argument for it. They basically just say there’s 80 gigawatts of demand. We’re 40 gigawatts short and therefore 40 gigawatts of off-grid data centers are going to materialize. And that’s just not the way it works. And so the reality is we have 80 gigawatts of demand. Maybe we can satisfy that and maybe we can’t. But in a best case scenario, maybe 10 gigawatts of that will be these off-grid data centers. And the vast majority of those are not going to come online in 2028. What we have over the next three years is the grid that we’ve already built. And the good thing about the grid we’ve already built is depending on who you ask, we’re currently operating that at 30 to 40% average capacity factor or utilization.

And so we do have a lot of headroom on that grid if we can figure out how to sort of tweak things a little bit so during the peak hours of time we can flex load either at the data center or on the grid around it. And that’s the path forward. The next three to five years is about load flexibility and that’s going to be the driver of data centers being interconnected. And the vast majority, like 90% is that. That’s the challenge we have in front of us and everything else I think is kind of a distraction.

Stephen Lacey: Okay. So Jigar, over to you on these numbers. So the claim is that behind the meter will power well over half of data centers by 2028, 50 gigawatts a year by 2029. What do you think about the overall numbers? Do you disagree with the actual numbers as Tim does, or is it just the year they land in?

Jigar Shah: No, I mean, having talked to all the investors, I disagree with the numbers. I mean, just to be clear, all major investors of data centers have said that they will not finance off-grid data centers because they don’t believe that they will work. So if you go to the major investors behind data centers, they have all publicly said that they are skeptical and none of their independent engineers have validated that the data centers can actually work. So they’re not sure what the value of their collateral is for investing in these projects. And so this is the thing that Tim says, which I agree with fully, is that look, I want to like SemiAnalysis because it seems like nice people and they have a lot of Indian people there. But the thing is that it feels like semi refers to partial analysis. When you’re doing something as intelligent as they’re doing, which is generally my experience with SemiAnalysis is they’re doing supply chain analysis, right?

Particularly in semiconductor space. And

Stephen Lacey: They’re well respected there.

Jigar Shah: Totally. Where you have an extraordinary overhang of supply chain is in batteries. We’re short natural gas. We’re short lots of other things. You know what we’re not short? Battery manufacturing capacity. We have battery manufacturing capacity for days, not just in China and Asia, but also in the United States. Why? Because the loan programs office actually funded all of those plants. And so Ford is now converting their plants to utility scale batteries. So is Stellantis, so is SK, so are other people. We are going to be able to manufacture all these batteries in the United States, get the extra 10% bonus tax credit. And guess who else is all in on batteries? The OBBA. They extended battery tax credits through 2034. And so now all these people who are experts in solar financing are now becoming experts in battery financing. So we don’t have a shortage of workforce in that area either.

And so my big argument here is just that they are basically saying that the sector that has not built at this scale in over 15 years, so we have no workforce in that area, that’s why Bechtel and Kiewit are charging three times more to install these natural gas generators than they were just 10 years ago or five years ago. And all of that stuff is going to get resolved easily via Claude or why don’t we talk about ChatGPT anymore, by the way? Are they just not good anymore? I don’t know. Anyway, but the area where we have 300,000 workers where they’re completely trained and ready to go, they’ve been deploying batteries for the better part of the last five years in California and Texas. They already have a financing supply chain behind them, but those people are not even in a report. I mean, it’s just laughable on its face.

And so the thing for me is that do I think we’re going to build a bunch of behind the meter natural gas? Yes, of course we are. They’ve already been purchased. Why would you just put them in a warehouse somewhere? You’re going to install them. I think that’s great. But I just think that when you think about the fact that they don’t have firm gas capacity, it’s five to seven years to build a new natural gas pipeline or expand the gas pipeline. So they can’t get guaranteed natural gas delivery to their site. All they can do is to burn gas when available on the line. And so they will hopefully take those data centers off grid to help provide some of that demand flexibility that Tim was talking about for 200 hours or 300 hours a year, which I think is a great use case for those behind the meter natural gas plants.

But the notion that you’re going to run them off grid and then run them off grid for seven years, which is what the Rhodium report was saying, is ridiculous. And I just don’t understand how many times people who have real technical expertise like me or Tim or others can say this and people are like, defend yourself against a 28 year old using Claude please again. Please defend yourself again against a 28 year old using Claude again, please. No, come on. It’s not even serious.

Tim Hade: Just in fairness, and I have a lot of empathy for the 28 year olds using Claude because I used to be that person. And so again, 15 years ago when I started doing this, I got good at Google and I learned a bunch of things. And I would go to my boss at the time, Howard Goodman, who is my co-founder at Scale and I would be like, “Hey, I figured this thing out.” And what you learn over time is that like, well, there’s actually a lot more nuance to building these projects than what you can learn on the internet. Real experience matters in electricity. Electricity is dangerous. Safety matters. What you do matters. And so again, I think what you can learn just kind of looking at a supply chain and trying to do math on a spreadsheet and what you can actually construct in the field are two very, very different things.

And I think that’s the part that what I’ll loosely call like the tech bro community misses about this whole thing. This isn’t like build a piece of software and then you have a power plant. This is like real physical infrastructure that requires a skill set that not many people have. We haven’t been building mega projects in the United States at this scale in my life. And so the people that build these things just don’t exist at that scale. And it takes a lot of time to train them and sort of produce that workforce. So we’re starting on that. So you’ve seen like Meta and Google make announcements about investing in workforce development and this, that and the other thing. And if that goes well, like five, 10 years from now, we’ll have a new crop of master electricians who can go in and build these projects.

But again, we’re talking about the next three to five years. And in the next three to five years, that’s just not something you can make. It’s something that you have to build over time.

Stephen Lacey: So this analysis that we’re talking about specifically punts on flexibility, but says that we think behind the meter will be a more prominent solution. They said that they’re going to come out with a flexibility report, but what do you think is the missed opportunity here?

Tim Hade: But just to put that in perspective, right? You can’t draw a conclusion without thinking about flexibility. So their analysis is basically like there’s no room on the grid. We’re going to talk about flexibility in a later report and therefore 40 gigawatts of off-grid data centers. And that’s just not the right methodology. The right methodology is there isn’t currently headroom on the grid based on their calculation methodology. If you account for flexibility, there’s a lot of headroom and therefore there’s no need to build 40 gigawatts of off grid data centers that we can’t build anyway. And so missing the flexibility analysis of that is one of the critical flaws of the report.

Stephen Lacey: Make the case for why you think in the next three years, load flexibility is the prominent solution is the only way to bring data centers online.

Tim Hade: It’s the only thing we can do at scale I guess is basically my argument. And so again, if you look at the supply chain situation, if you look at the workforce situation and you look at the demand side of this again, 100 gigawatts of net new capacity to serve data centers and hyperscalers, can’t build it. So the option we have is we can use the stuff that we’ve already built more efficiently. That’s it. Now if you think about this over a five, 10, 15 year trajectory, I am really optimistic about a lot of new technologies that are coming online and the ability to build these data centers faster and better and engineer new solutions. And I think there’s a huge role for advanced power electronics and solid state transformers and SMRs and advanced geothermal and all these types of things. But in the next three to five years, the thing we have is the grid we’ve already built.

That’s it. And so again, I think when you look at this sort of practically, that is the thing that we can ramp fast and that we can scale quickly. And that is by the way, buying time for people that are working on these newer technologies to sort of ramp up and get to scale. And then their decade will be 2030 to 2040. But in the next three to five years, what we have is what we’ve built already. And the only way we’re going to get more net new capacity is if we use our existing infrastructure better.

Stephen Lacey: What do you make of… I think one other argument the report makes is that we see the declining marginal ELCC for batteries. So as more storage gets deployed, it reduces the incremental value of new capacity. How much of a limiter is that for batteries as a solution?

Jigar Shah: I think you have to think about the trajectory here. Batteries came down and cost 27% last year. You know what didn’t come down and cost 27% last year? Natural gas solutions. Those went up by 27% last year. And so will the ELCC go down for short term batteries, let’s say two hour batteries or four hour batteries? Yes, because it’s working so well that they’re taking out all of the two hour and four hour events such that now the events are five hours or six hours or seven hours. So there’s two ways of doing that. One is just de-rate the batteries. So a one megawatt battery for four hours is the same thing as a 500 kilowatt battery for eight hours. You could just de-rate it and call it 500 kilowatts. Or you can actually now build a one megawatt battery that’s eight hours.

But guess what? The costs have gone down by 50% because the cost trajectory of batteries keeps coming down in costs. But the other piece of it that I think people don’t understand is that a natural gas generator only has a few revenue streams that they can access. Whereas batteries have five revenue streams they can access. It’s not just the capacity payment that you get from these tags that are being paid by the transmission system operator like in the PJM, but you also have the rating of your building as it relates to retail choice. And so if you can reduce your peak load that your building is showing on the 5CP market or some of those things, you get paid for that. If you can buy low and sell high, then you can get paid for that. If you can offer your battery into the ancillary services market, you can get paid for that.

If the utility needs to upgrade the distribution circuit and instead would like to pay you to use your battery flexibly to be able to push off that investment, you can get paid for that. So there’s all these other reasons. And so everyone’s like, well, the ELCC is going down. Okay. But these other four revenue streams have nothing to do with ELCC. Those other four revenue streams have other reasons why people are paying you for those things. And so the thing that bothers me the most about where we are in the quality of our discourse is I feel like on this side of the tech bro world, we’re not making enough progress. I think that like you would think that those people are educatable, but I feel like after 15 months of trying, they have something that they’re pushing and they don’t want to learn. That’s what I have to conclude.

Now the good thing is that the people who run the grid in Texas and Oklahoma and Utah and New Jersey and other places, they do want to learn because they actually have a ton of bricks on top of them and they have all this presure to solve things. And they’re realizing exactly what Tim is saying is that I actually have to solve this in the next rate case. I can’t wait for 2031 and 2032, which is when the PJM is expecting most of their natural gas assets to come online. They have to solve it now. They need headroom now because there’s a snow creed event like we had in January. There’s a heat dump event like we had last week. And so they need to make sure that they restore the reserve capacity now in the face of all of this pressure from the Federal Energy Regulatory Commission to say yes to the data centers.

And so even the FERC is now saying, do whatever you need to do, but get it done. Do behind the meter gas, do like batteries, do it, just do it. And they’re even like saying, do grid enhancing technologies, do advanced conductors. All these things that were out of favor last year, but we have liftoff reports for magically have become in favor. Why? Because the supply chains exist and the only thing that was holding us back was utility and transigence. And now utility and transigence has no place in where we are in the physical grid. And so even the utilities are like, dammit, I didn’t want to do this because I hate you Jigar, but we’re going to do it anyway because I have no other fricking choice.

Tim Hade: Load flexibility is not easy. And so the difference between load flexibility and behind the meter off grid islanded data centers is one, load flexibility is exceptionally hard and the other is impossible. Just not going to happen. And on the behind the meter generation side of things, the constraints are physical infrastructure and workforce. On the load flexibility side of things, the constraints are software and market operations. And so again, I think that if we’re really going to add 100 gigawatts of net new load to the grid over the next three to five years, we definitely need to have evolution in terms of how we operate the grid. That’s happening in PJM right now. It’s happening at ERCOT right now. It’s happening in CAISO right now. Every market is going through some semblance of reform to try to figure this out. And then you also need software.

And you guys have talked many, many times on this podcast about how utilities don’t necessarily have the functionality on the distribution system that’s required to operate the grid in the optimal way. And the difference is we have aligned incentives on the market operation side. Generally speaking, and there are a lot of nuances, like the regulators bipartisan. The regulators and the utilities and our industry and the hyperscalers are all on the same page in terms of load flexibility. And so again, it’s going to be hard and we’re going to have to work out a lot of nuances and a lot of different regulations are going to have to get changed, but incentives are aligned and we’re already moving towards a better environment from a market operation side to accommodate load flexibility. And then again, on the software side of things, this has already been solved. So you go to the UK and Octopus runs essentially DSO.

They essentially run a distribution system operator in the UK. They have the software to do it. They know how to manage all these loads with pinpoint accuracy. There’s a bunch of other companies that are doing this. So on the load flexibility side of things, we need some regulatory reform. We need some market operations evolution for sure. And we need to make sure that we’re incorporating the right software, but those things exist and we can create those in the next 12 to 24 months. What we can’t do is build a hundred thousand master electricians.

Stephen Lacey: So final question on this. Let’s sort of map out an ideal scenario over the next three years. Where’s the highest value application for batteries? Where do you see batteries having the most contribution to the system?

Tim Hade: Everywhere you can put them. My general view on batteries is we need to connect batteries everywhere we can put them as quickly as humanly possible. And so again, I think one of the problems we have with utility scale batteries, which again is 90% of the market today, is they’re stuck in the same transmission queues as the large loads. So most utility scale battery systems are connected to the transmission system. So depending on what analysis you look at, right on average in PJM, it’s a six year wait time to interconnect a new utility scale battery system. I think in CAISO it’s like approaching nine years. Even in ERCOT, it’s north of four years. So utility scale stuff, again, there’s a lot of stuff already in the queue, but net new utility scale batteries, it’s just a race to try to find interconnection slots. And I do think people will find creative ways to get around that, but that’s not where the majority of net new is.

So then basically the alternative to that is the distribution system. And there are a lot of different ways to connect batteries to the distribution system. You could do them at homes, you could do them at businesses, and you could do them at community scale. And I think like everything in the electricity sector, every utility and every market and every state is going to have a little bit different preference in terms of how batteries get deployed. But the overarching goal should be deploy as many batteries as we possibly can on the distribution system over the next three to five years. That will result in us having the most load flexibility possible, which then allows us to connect more large loads faster.

Jigar Shah: So obviously agree with Tim. I’d say it a slightly different way, which is that I think that the data centers need to earn the trust of the communities that they’re going into. Otherwise, the governors are going to pass moratorium, in my opinion, in every single state. And what the Octopus Energy’s nonprofit, the Center for Net Zero found was one of the ways to earn their trust is to give them free stuff.

Stephen Lacey: Yeah. We talked about that last week. I think that’s a really interesting finding.

Jigar Shah: And so what I’m very concerned about is that we put all the batteries at industrial sites and other places where you can put one megawatt batteries and we don’t focus on figuring out how to give people residential home batteries because that would then… We talked without naming him about Lorenzo Kristov’s paper at CAISO. And he calls it designing the grid from the community up. And so I hope that we make sure that these batteries go into churches, they go into food banks, they go into FEMA emergency centers, they go into schools, they go into this long tail of places where we were supposed to serve them with the Greenhouse Gas Reduction Fund or the Solar for All money, but a lot of them have not received a lot of these benefits. And so I hope that while we’re on this journey, the batteries go to those people who provide that essential resiliency service for the community and not just to large industrial loads.

Tim Hade: I guess before we move on, Jigar actually just posted an incredible tribute to Lorenzo Kristov on his substack, which I would encourage everyone to read. But it was a reminder for me that we all stand on the shoulders of giants. And so again, the reason we’re in this position as an industry isn’t because we just made it up in the last 12 months. This has been an effort that’s happened over decades. And Lorenzo is one of the founding fathers of this stuff. And everyone who’s good at this today has read his papers and been inspired by him. And so Jigar wrote this nice tribute to him the other day and I would highly encourage everyone to read it because we won’t be here without his brilliance.

Stephen Lacey: We’ll throw it in the links and the show notes. Nice. Well, to wrap this up, I think we should turn to this Rhodium group analysis, this framework they developed that asks a very different question entirely, which is does the choice actually move the energy transition forward? We’re not just talking about speed, how fast the power gets built. It’s what’s the actual impact. And so they developed this transition acceleration framework that scores different ways of powering a data center on a single scale. So a grid connection paired with clean firm power scores the highest. Permanent off-grid gas scores the lowest. This is not neutral. This actively working against the transition in their model. So this gets us kind of back to the philosophical divide that we touched on at the start of the show. I think the purpose of this show, the three of us are thinking about how these choices about how we power data centers accelerate, enable clean solutions.

And this can get overshadowed by the speed only believers who don’t really think decarbonization utilization should be much of a factor or they’re not thinking about it. So let’s comment on this framework first. Jigar, what are the spectrum of best and worst options according to Rhodium?

Jigar Shah: So when I read the paper, I was so disappointed because this is a 2023 conversation. It literally has no place in today’s conversation. I think that what they were trying to say is that if you were to fully decarbonize the grid using the bulk power system only and you don’t look at grid flexibility and you don’t look at grid enhancing technologies and you don’t look at figuring out how to get more out of the grid we’ve already paid for, but you just focus on generation that is matched with the load, then here is how we rack and stack all of these options for meeting that load. So for instance, no, the baseline is just plugging into the grid. No upgrades, no nothing, just plugging into the grid, that’s zero. But weirdly, if you just buy racks from across the country without hourly matching, you get a 0.2 score.

I was like, oh, that’s interesting. I thought we were supposed to hate racks these days. And then when you look at like Google’s project with Sheldon Kimber, that has a 3.4 score. And then if you look at off grid, they’re assuming that even as a bridge solution, it would take seven years for it to bridge to the grid, which I have never seen in any of the projects that I’m looking at. So I feel like I don’t understand why you would pick seven years as the number when most people I’m looking at are looking at a bridge of six months, eight months, maybe a year. And so I just think that in general, one of the things that disappointed me so much about this paper was it basically was pitting affordability versus climate, which I thought we were past that, but we’re back right into the Milou.

Stephen Lacey: What was your reaction, Tim? I mean, this doesn’t factor in the DER flexibility model we’ve been talking about. Do you feel like that’s a blind spot?

Tim Hade: Yeah, for sure. I mean, look, I think the conclusion of the analysis is directionally correct, right? Which is grid connected data centers are better for society and the energy transition than permanent off-grid data centers. The calculation methodology, you can definitely nitpick. And I agree with most of the points Shagar made, that there was a lot of stuff that wasn’t accounted for in the analysis, but I think the outcomes are directionally correct. But look, I think this is the big takeaway for hyperscalers and policymakers and regulators and everyone who thinks about this is this debate between onsite power and the grid is not really the right debate. Is a data center going to be part of the system or is the data center going to be an island that sits in isolation? And look, I think the important thing about this is if we do this the right way, data centers can be part of the system in a way that benefits everyone.

So we can lower electricity rates. We can make the electric grid more reliable, more affordable and cleaner because the hyperscalers are coming in and paying a lot of money for us to do that. And so in essence, we’ve created this environment where hyperscalers are essentially subsidizing the rebuilding of the American electric grid. And that’s the potential of what we have in front of us. And so I think if we take advantage of that opportunity and we execute at a really, really high level over the next three to five years, the outcome of this has the potential to be really, really good for everyone. Not just the hyperscalers, but all the rate payers on the rid as well. And I think directionally that’s what this analysis shows is that on grid data centers that are part of the solution, whether they have onsite generation or not is not really the debate.

It’s whether they’re connected to the system and they’re members of the community or whether they’re not. And folks that are members of the community are part of the grid or part of the system. If they do that the right way, have an opportunity to make it better for everyone. And that’s the future I’m hoping to be a part of.

Stephen Lacey: Just to flag one other critique, your colleague, your former colleague at scale and friend Duncan Campbell made an argument on X. This is very similar to what Jigar has been talking about related to how we’re going to utilize all this behind the meter gas. And he basically said that most of these behind the meter systems, these data centers will get a grid connection as soon as they can. And then that’s just a bunch of gas capacity that gets privately funded acting as low capacity peakers that are helpful in managing high variable energy grids. And so if a data center gets connected immediately, you kind of push coal and gas up the merit order to serve that facility. But if you have all this behind the meter gas, then you can utilize it to manage a high variable energy grid. So are they missing that this behind the meter gas can actually be a transition enabler?

Tim Hade: Yeah. Well, first of all, I guess I’m glad you reminded me about Duncan. For those of you who are interested in this topic, DERVOS is September 18th, one month before the Flex Summit that’s hosted by Latitude Media. So we got a big conference season coming up to talk about all of this in the fall. But look, yeah, I think Duncan’s point is correct, right? Which is again, if you build onsite generation, specifically reciprocating engines or gas turbines onsite and you use that as a bridge and you do that in the right way, those assets can be essentially peaker plants for the grid once a grid connection is available. And that’s part of a load flexibility strategy. And so again, I think the way Duncan thinks about this, and I don’t want to put words in his mouth, but the way Duncan generally thinks about this is load flexibility is what we need to achieve.

And so therefore building onsite generation that could be a short-term bridge, but also part of a bigger system in the long run can be done. Now again, the argument against that is not that Scale can’t do that or Crusoe can’t do that or Intersec can’t do that. It’s just that there aren’t 40 gigawatts of those folks out there who know how to do that. But yeah, I think the model a lot of these companies are pioneering where they’re using natural gas as a bridge to grid interconnection. And then after the grid interconnection, they’re using gas as a flexibility asset is a point that has real merit. And I think Jigar’s made that point a bunch of times before is like natural gas is going to be part of this equation for a long time. From a climate perspective, we’re trying to ramp down not the capacity of gas we deploy, but the capacity factor of those gas generators over time in order to decarbonize.

But yeah, I think projects like that can definitely be a big part of the solution. I think some of the best firms in the world are working on them.

Jigar Shah: Well, we’re going to need capacity on the grid in the same way that China is building new coal plants and then designing them to run flexibly and is reducing the capacity factor of how often they run such that they believe that they’ve hit peak coal in terms of terawatt hours. So paying for capacity is not a problem for me. The thing that disappoints me the most about this report from Rhodium is that in this moment, the people that we need to inform are governors. And I don’t know what a governor is supposed to take from this report. You look at this report and they’re like, is this actually helping me with affordability or not? Is this helping me with speed to power or not? It doesn’t really address any of these issues, which is why it was so disappointing to me because it’s basically a 2023 report.

It’s sort of like if you want to decarbonize the grid faster, here’s how you do it and here’s what’s the most useful. So I find that valuable and I’m going to put it on the shelf and reread it when there’s an administration and a bunch of decision makers that care about this topic. But right now with 36 governors running for election right now, every one of them hates data centers and hates their electric utility company. And I don’t know how this report actually informs what stuff they should be doing to get elected and to keep rates down.

Tim Hade: On that point, maybe I can use my opportunity to appear on an open circuit to ask a question. I mean, Jigar, one of the things I’m really interested in is there are a lot of governors who are hyper focused on this issue right now. A lot of them have different approaches to how they’re talking to industry, how they’re talking to utilities, that type of stuff. Do you think there’s a particular model in a particular state that’s really good right now or is everyone trying to figure it out?

Jigar Shah: Well, obviously our friend Arushi Sharma Frank has been doing extraordinary work in Texas with Batch Zero and PGGR and some of that stuff. So I think that stuff’s great, but I think that fundamentally what the governors have to prove to their voters is that they’re actually protecting their voters. The voters are never going to understand all this nuance that we’re talking about here. They’re just like, “Is someone looking out for me? Is someone making sure that a billionaire or now a trillionaire is not taking advantage of me?” And as you suggested, Tim, if you wanted to get full unlock of demand flexibility, that requires new software to get deployed. It requires new cultural norms within the electric utility. It requires the regulators to do things differently than they’ve done before. And so if they’re going to force that onto people, they want to know it’s going to work.

That’s real political chips that they have to spend to get that done. So like you see in Georgia, for instance, they’re mandating that they implement a DERMS platform because they’re saying, “We need to have demand flexibility and unless you have a DERM software platform installed, you can’t do this.” In New York, they’re mandating that National Grid hire GridCare who then magically found 650 megawatts of underused capacity that National Grid New York didn’t know was there. And so when we were at a previous episode of Open Circuit, I asked would all 168 utilities hire someone like GridCare by the end of this year? And I think the answer was like, “I’m not sure, but unlikely.” That’s ridiculous. If you could just use software for 26 minutes and find 650 megawatts of unused capacity in the grid, every governor should force their utility to hire one of those six companies and find their underutilized capacity tomorrow.

And so I think that all of these individual best practices are being lifted for the governors to read and learn about, but I don’t think there’s one place unfortunately that’s done it exactly right.

Tim Hade: Cool. So one follow up so I can get my money’s worth here. Yeah,

Stephen Lacey: Get in here.

Tim Hade: All right. So we’re talking today. I think I read last night that Governor Hochul in New York became the first governor to sign a formal data center moratorium, which I think is for one year. How big a deal, like Stephen, I know I learned so much about this stuff from you, but how big of a trend do you think that’s going to be? Do you think data center moratoriums are going to be a real thing or do you think the governors see the opportunity to kind of create a better system for everyone?

Stephen Lacey: Oh, I think that they’re feeling immense amount of pressure and we’ll probably see more moratoriums. I don’t think it’ll spread to every state like Jigar suggested earlier in the show. To me, this still tracks like fracking where we saw some states limit or ban fracking, but ultimately the industry moved forward relatively unabated. And I just have a feeling that that’s how this data center story is going to go. Governors are under pressure, but they’re also under pressure to bring in economic activity. And I think many of them still privately really like this data center activity, even if their voters are really angry about it.

Jigar Shah: One small thing is I think it’s moratoria.

Stephen Lacey: Moratoria. Yeah. You’re right. Here I am. I’m the editor. Thank you.

Tim Hade: Learn something new everyday.

Stephen Lacey: The editor needs an editor.

Jigar Shah: But I also think that one of the big challenges I see is that the utility companies, when they’re filing for their 15% rate increases, are deliberately blaming the data centers for those rate increases. And so this notion that we have report after report after report that says data centers are not raising rates, data centers in fact have lowered rates like four years ago or whatever it is. That undermined by every single quarterly conference call and every single integrated resource plan that the utilities are filing right now saying, “We need a 15% rate increase because of data center load growth.” And so I just think that’s what voters hear. They hear, “My bills are going up 15% because of data center load growth because that’s what the utilities are telling

Stephen Lacey: Them. I mean, the thing that’s fueling this politically is this question about whether these data centers are being used to build something that benefits people. You can make a case that with fracking, we’re going to have cheap energy and we’re going to become more energy independent and it’s a really clear story. I don’t think the story is very clear on the benefits of AI. Even though I am a big believer in the medium and long-term economic potential, I think that that story is pretty negative for most people. And we’ve just heard the last two years of tech executives and AI labs saying they’re going to automate your job away. Half of the white collar workforce is going to go away. People are sick of slop. I think the politics of this are a little bit different or a lot different than fracking because there’s no clear story for the economic benefit this is going to bring. So that to me is the big gray area about how this is going to play out politically. What

Tim Hade: A time to be alive.

Jigar Shah: I also don’t think an 80-year-old president’s a very good articulator of the benefits of AI.

Stephen Lacey: No, but Tim was. AI for lowering the cost of commercial storage. I love it. This is what I have the most hope in.

Tim Hade: No, the one takeaway from this is like that’s definitely a potential outcome. There is a way that we can use load growth to fuel the next generation of electrical infrastructure. I’m 100% confident in that. And incentives are aligned. It’s just an execution game. So we got a lot of work to do as an industry, but I think we’re in the best place we’ve ever been for what we do. And so thank you guys so much for having me. This has been a blast.

Stephen Lacey: Yeah. So glad to have you on the show. Tim Hade is an SVP at Voltus. Really good to see you, my friend.

Tim Hade: Good to see you all too. Thank you.

Stephen Lacey: Jigar Shah is my regular co-host and the co-managing partner at Multiplier. Thanks, Jigar.

Jigar Shah: Of course.

Stephen Lacey: And before you go, make sure to subscribe on YouTube. A huge amount of our new viewers on YouTube do not actually subscribe to Latitude Media. So we’ve got a ton of you watching these conversations now, but if you want to head over to Latitude Media’s YouTube page, you can watch episodes of Catalyst. You can see all our back catalog episodes. The show is edited by me, Sean Marquand and Anne Bailey. You can find our audio episodes on Apple, Spotify, or wherever you get your podcasts and transcripts at latitudemedia.com. Thanks so much for being here. We’ll catch you next week.

The post The off-grid data center fantasy appeared first on Latitude Media.

via Latitude Media https://ift.tt/mV5cDvo