Is your electric bill about to go up? with Asim Z. Haque

[00:00:02] It was a Monday afternoon in August. Baltimore. 90 degrees. The kind of humid that makes the air feel solid. At 3.52 p.m., the largest grid operator in the country made a call that had never been made before in this region. They ordered utilities to cut power to rotating pockets of central Maryland. Not because of a storm. Not because of a downed line.

[00:00:31] The inciting incident was the failure of a utility substation earlier that day. The sheer amount of electricity people and businesses were pulling from the grid at that moment had pushed past what the local system could carry. They managed it, activated backup reserves, asked large customers to voluntarily pull back. Kept it from being a mass outage cascading across the Mid-Atlantic.

[00:01:01] This time. This is what happens when a grid is overloaded. When energy demand outpaces capacity. And the crisis was a year ago. Since then, AI has become more ubiquitous than ever. And so have the data centers that power every single request we type into these systems. Unlike your home, which peaks in the morning and again in the evening, a data center has no off switch.

[00:01:30] They're humming at full capacity all day, every day, adding more pressure to our energy grids little by little. PJM Interconnection manages the grid for 67 million people across 13 states and the District of Columbia. They're the ones who handled the overload last August. They project that data centers will account for 94% of its peak load growth through 2030.

[00:01:59] Not some of it. 94%. And here's some important math that doesn't add up. According to today's guest, you can build a data center in two years. The power plant to fuel it takes seven. That gap is where Baltimore is. That gap is where your electric bill is. And that gap is where we're going today. This is TED Tech, a podcast from TED.

[00:02:30] I'm your host, Sherelle Dorsey. We've spent this series tracing the physical cost of the AI boom. The water, the land, the communities that never got a vote. In the final episode of our special series on the costs of data centers, we're going to the grid itself, the infrastructure underneath all of it. My guest for this episode is Asim Haq,

[00:02:55] Senior Vice President of Governmental and Member Services at PJM Interconnection. His job at PJM is to help keep the lights on for those 67 million people who count on the utility, all while the demand for AI and the power that fuels it grows at breakneck pace. In our conversation, Asim is breaking down what's at stake when our grids reach capacity

[00:03:21] and what needs to happen to sustain the hyperscale data centers moving into our communities. Then we'll hear a talk by physicist Varun Sivaram, who argues the crisis doesn't have to end in blackouts. Instead, data centers can become part of the solution rather than just the source of the problem. Two questions run underneath all of today's episode. Who's accountable when the grid runs short?

[00:03:50] And is there time to choose a different path? We'll get some answers after this quick break. So, Asim, most people aren't really thinking about where their electricity comes from. You know, what are we really looking at when we say the grid? So, let's begin with the most basic, basic thing.

[00:04:17] Okay, you're sitting in your house and you're watching TV. That electricity via electrons was carried to your home through distribution poles and wires. Those are the lines that are outside of your house. You have a terrible winter storm, for instance, and sometimes ice develops on those lines. And, you know, you might have an outage for a day or two, hopefully not longer than that.

[00:04:44] But the distribution system delivers power to your home. Okay? Now, where does the distribution system get power from? All right? It gets power from what's called the bulk electric system. The bulk electric system consists of two components. It consists of the creation of the watt. We call that generation.

[00:05:14] So, the creation of the electron. And that can come from a variety of sources. That can come from solar, wind, a battery, a gas plant, a nuclear plant, a coal plant. The other component of the bulk electric system is transmission. High voltage transmission that take the watt that's been created and transmit it to the distribution system. Got it. Okay. Okay.

[00:05:42] So, really quickly, PJM, we have nothing to do with the distribution side. Where we operate is really the coordination of the generation and the transmission so it can get to your distribution substation and then deliver to your home reliably. That's our whole business. We are the largest in terms of the demand that we serve, the consumers that we serve, and the amount of supply and how big our market is. Thank you so much for that framework.

[00:06:11] So, you know, considering that you all are the largest in this space. I'm really excited to kind of get your thoughts on this AI boom. So, you know, we're going to have to move on. We're going to move on.

[00:06:47] We're going to move on. We're going to move on. We're going to move on. We're going to move on.

[00:07:11] We're going to move on. We're going to move on. So, we need a lot more power.

[00:07:41] demand that's been brought to us by data centers and the need to sort of meet this moment with AI, it's slowly but surely kind of chewing up our reserves that we've had on the power system. And what does that mean? What is it in terms of the reserves? Oh, yeah. What is reserves when it relates to power? Yeah, it's a great question, right? It really is two real world impacts for consumers.

[00:08:05] Okay. The first is, you know, if you have less power to deliver across the power grid, you are at greater risk of having what's called a reliability event. Okay. And what that means is that you could experience some disruptions in your service. So that means blackouts, like neighborhood blackouts for a period of time?

[00:08:30] Yeah. So it could mean a lower voltage event where not necessarily you are blacked out, but you could have sort of a lesser amount of voltage going through your home or your city or your region, which is, you know, some people characterize it as a brownout or characterize it as a browndown event.

[00:08:52] You could experience actual what's called load shedding in various pockets of the footprint. What that means is not full-scale blackout, but you could have smaller regions that would have to be without power for a period of time. We would try and manage that from a grid operator's perspective, try and rotate those outages.

[00:09:16] So I don't want to get the folks at home who are listening to this, you know, deeply, deeply concerned that there will be a blackout, a prolonged blackout because, you know, part of our job is to try and minimize that impact. But again, the first is reliability. It's if you don't have enough resources, if you're resource inadequate to meet the demand that's on the system, then you could have a reliability event.

[00:09:40] Now, the hope is that would only occur in our most strained times on the grid. So think of the grid as a superhighway. You've got multiple lanes on your average nice weather day. If it's a 10-lane highway, only five of them are being used. But if it's super hot or super cold, okay, all 10 are going to get used. And it's only during those periods where, you know, we are a little concerned as this reserve margin gets chewed up.

[00:10:08] And then the only other thing I want to say, Sherelle, is you talked about reliability. In any marketplace, when demand starts to outstrip supply, what is the result on price? It increases, okay? So that is in part what we are seeing today. And what also appears on your bill is the necessary infrastructure buildout in order to plug in these data centers. So does this mean expansion of infrastructure or? Absolutely.

[00:10:38] So expansion of the, we already talked about it, the distribution system as well as the transmission system. So when you plug in something the size of a small to large city, okay, you are going to have to build infrastructure around it in order to transmit those electrons. Yeah. When did the growth of data centers stop feeling like maybe background noise and really start feeling more like,

[00:11:04] okay, this is an inflection point in our system and in our business that we're going to now have to meet the demand with greater infrastructure or figuring out how to increase those lanes on the highway without having these potentially catastrophic reliability events. First thing to say is that, you know, we operate as effectively a nonprofit, okay? We don't own any power plants. We don't own transmission lines.

[00:11:32] Think of us as sort of an air traffic controller. It's the best kind of, you know, metaphor for what we do for the power system. Part of that air traffic control mechanism is running competitive wholesale markets, okay? So we not only are moving electrons across the grid, and we're not only saying, hey, here's where infrastructure needs to get built out in order to meet demand, but we also run these markets.

[00:11:57] A few years ago, we ran one of our competitive wholesale markets, and we saw a pretty massive uptick in pricing in just one year, okay? We saw an uptick in pricing from what was $28 a megawatt day to about $270 a megawatt day, okay? When we saw that uptick, we said to ourselves, this is a different phenomenon than anything we have experienced here.

[00:12:26] For people, the downstream impacts are potential reliability event and increased cost, okay? The only way to deal with those two things from purely a supply-demand perspective, macroecon, okay, is either you have to choke off the demand, meaning data centers, you can't come onto the system. Because if you come onto the system, you are going to create reliability and cost problems for everybody, okay?

[00:12:55] It's either that or you're going to need a lot more supply. The reality is we're just a grid operator, right? We're air traffic controllers. So if we are not going to choke off the demand, if we're not going to prevent demand from getting onto the system, we're going to need a lot more supply, okay? Either way, that costs money, all right? Remember, generation, transmission, distribution. Transmission infrastructure also has to get built. Distribution infrastructure has to get built. Take those three things, pull it all up into the bill.

[00:13:24] It's going to cost more money. This is all costing money, okay? So then the question becomes who pays. Before you continue, as you mentioned, who pays is the ratepayer. For the most part, currently, yes, okay?

[00:13:39] However, however, all of the major hyperscalers have come to the table, specifically to President Trump's table, and have committed to via ratepayer protection pledge that they are going to pay for all of this new infrastructure. Now, that's a great commitment. How that actually gets accomplished in practice is the thing that the industry is trying to unravel right now.

[00:14:10] I was going to say, because that doesn't sound – it sounds like a nice thing to say for PR purposes, but potentially difficult to enforce. Seeing it in action is a little difficult, okay? And from the grid operator's perspective, we don't send bills to everyday consumers. What we do is we send costs to a zone or a utility.

[00:14:35] And then it is up to them and their state regulators to determine – state legislators, governors – to determine who then pays the tab for those costs that have been sent over to the zone or the utility. And so it is not as simple as, oh, you know, let's just send the data centers a huge bill.

[00:15:00] Well, this has got to, you know, get work through the very sort of complicated paradigm that federalism creates. There are federal authorities, there's federal authority, and then there's state authority. And so I think the industry is trying to work through that right now. I think that is the kind of topic du jour right now.

[00:15:22] Yeah, because it sounds like there's just a lot of sort of unforeseen growth in this space through the years and still lots of layers and complexities around the calculus of how this gets executed, how the grid at large facilitates this new power load. And I read, Asim, that the data centers now are accounting for 94% of PJM's projected peak load growth.

[00:15:51] Can you put that for us in plain terms, what that might also mean for everyday people on the grid in terms of not just the maybe potential rate increase, but are there any other implications as well as the peak load growth expands? Yeah. So think of us as a 180,000 megawatt system. Okay.

[00:16:15] Our load forecasts are projecting that, you know, in just a decade that we'll be at a 220-something thousand megawatt system. It's pretty substantial growth in a pretty short period of time. Okay. Okay. Now, the implications are that there will be, unless we get a lot more supply, less power to push around. Okay. Which again creates this reliability concern.

[00:16:44] And then the other implication is all of this growth is going to cost money. Okay. Okay. So whether it's the new supply that has to get built out or the new transmission line that has to get developed. Okay. All of this is going to cost money. And so ultimately, what's the impact for consumers is, you know, the who pays concept.

[00:17:13] The other facet of this that I will say that is kind of a hidden issue, but that Congress is working on and some others are working on, is it is very difficult these days to get infrastructure actually built out. Especially in areas that are dense residential areas. And even in areas where there is not residential density, let's say it's farmland.

[00:17:42] And let's say there are six or seven owners. I would say that the grassroots organization against the build out of infrastructure has reached a bit of a fever pitch. The acronym that folks utilize is NIMBYism, the not in my backyard. So remember how I said we're a 180,000 megawatt system? Yes. We have studied over 50,000 megawatts of resources.

[00:18:10] Most of them are renewables that could plug into the grid right now. Okay. They are done with everything PJM oriented and they could plug into the grid and start providing power to consumers. Many of those projects are delayed and may never come to fruition. The number one reason we are given is permitting and siting in states.

[00:18:39] Siting means the actual you're given permission by a state body to actually construct in a given location. Okay. Interesting. Okay. So it's not just the crawl back on dollars, federal dollars for these projects to expand. Yeah. Yeah. That's right. It's the permitting process. Absolutely. There's a hold and or potentially a sabotage, if you will, of getting these technologies.

[00:19:06] Because when we're talking about kind of interconnecting into the system, this would be immediate, I assume. Yeah. So this isn't like waiting a decade for these things to be built out. Nope. Nope. So these are projects, again, once you get through our processes, right, you're then left to your own devices as a developer to obtain parts, to obtain fuel,

[00:19:31] and to get all of your permits and all of your siting certificates in order to build out your resource in any given jurisdiction. A lot of this happens at the state and local level. And in many instances, this stuff is not getting built out. I'm not talking just about generation plants. I'm talking about transmission as well. A very simple story.

[00:19:54] Is that in the state of Maryland, PJM has said, we need to build out a bulk transmission line in order to meet reliability. Some of that reliability need is certainly being driven by the data centers that have found their way around Maryland and a little bit in Maryland. But it's got a pretty congested area, especially in the Baltimore area.

[00:20:21] And the entity that is trying to construct this transmission line is experiencing a lot of pushback from folks that are along the line's trajectory, a lot of it being in more farming-based communities. Now, you want for everybody to have a voice. But as long as everybody understands that unless we build out some of this infrastructure,

[00:20:46] you're not going to reduce the two risks that I outlined for you, which are reliability and affordability. Right? Right. Unless we all mutually agree as part of the compact of being an American that we are going to have to get some of this infrastructure built out in order to, again, deal with the reliability and the affordability issue. We're going to struggle.

[00:21:10] And I think we're struggling right now, which is why actually this is something that Congress is currently taking a look at, sort of citing and permitting. And are there ways that, frankly, we need to have the federal government step in and say, no, no, no, this shall get built. We hear you. We understand you. There might be a process associated with this. We might move the line here, move the line there. But this shall get built.

[00:21:33] Where is the through line for responsible build out that is sensitive to communities concerns about potential environmental impact and also creating some assurances that we're going to build this line out responsibly. We're going to do feasibility studies, potentially.

[00:21:55] We want to do this the right way so that we're not disrupting the environmental and health welfare of communities. Because I feel like that's the chief argument. I don't know that it's that residents don't want to see this. I think that there's been sort of this almost bulldozing in certain communities and or maybe lack of understanding of how will this impact our community, our neighborhood, our what have you.

[00:22:21] Okay, so let's begin with this idea of community engagement with build out of infrastructure. So I am a former regulator. I used to regulate all utilities in the state of Ohio. Part of my duty was to chair the Ohio Power Siting Board. And we talked about those concepts of siting and permitting. Okay.

[00:22:44] And as part of my responsibility as chair of the Ohio Power Siting Board, you needed to make sure, frankly, that the developers, whether it's developer of generation or development of transmission or development of a pipeline,

[00:23:01] that they had done all of the necessary community outreach and had listened to the community and in many instances moved the line or moved the generation resource or moved the facility to avoid some specifics. Right. I mean, I think that there's so much nuance with people's property that, you know, moving something five feet can make all the difference conceivably.

[00:23:30] And so it's up to the developer to make sure that they have spent time with all of the landowners and have done their level best to accommodate the landowners' concerns. Okay. So when you see the sort of grassroots organization against infrastructure build out, okay, if you don't want data centers, that is a different grassroots movement. All right.

[00:24:00] But if you don't want electricity infrastructure built out and those data centers are coming anyway, the reliability and the cost impacts are going to find their way to your household. That's a really important piece of this puzzle. Okay. And as far as the individual landowners, developers got to work with individual landowners to listen and understand their concerns. As far as the resource types, the sources that can power data centers, a few things there.

[00:24:29] First of all, it's important to recognize that not all resources have the same operational reliability characteristics. Okay. And what I mean by that is, and we've actually got, you know, a metric that we utilize in order to say, frankly, here's how reliable all of these resources are during the most challenging hours of our most challenging days of the year.

[00:24:59] All right. And what I will tell you is the gold standard is a nuclear resource. All right. It has, we'll just characterize it as 95% reliability. Okay. Okay. The resource that's at the bottom of this stack is a solar resource that is a fixed solar resource, meaning it doesn't track the sun. Okay. And that's roughly in the single digits, like 9%.

[00:25:28] Okay. So, this is something that has to be understood about, and then everything else is in the middle there. Right. So, really high reliability resources that are clean. I would say long duration battery storage still has its challenges economically to deploy, but a very high valued reliability resource. Offshore wind, high value reliability resources.

[00:25:55] And then you've got your traditional coal and natural gas that are actually quite reliable. One piece of the puzzle that has been missing from this, like, the environmental component of generation resource types is that all of these resource types, okay, have their operational capabilities.

[00:26:15] And we have to take that into account when we're planning for a system in the future that is going to require, A, a lot more power, and B, reliable power. 24-7, these data centers are running, right? It's not like our homes. Homes, you've got peak. You have two peaks. You have a morning peak, and you've got an evening peak. Right. Because that's when we're all doing our stuff, right? For data centers, it's all peak. Right? And that is a different animal than we have had to deal with.

[00:26:43] And now, based on your experience as a regulator or sim, what are you seeing local governments or states doing that are right or at least headed in the right direction? Yeah. Awesome question. Okay. So, the first thing is, the more streamlined but still listening to your consumers, you can make a citing and permitting regime. You've got to try and do it. Okay? That's the first thing. The second thing is, this is a bit wonky.

[00:27:10] But if we're talking about who pays for all of this infrastructure build-out, that very kind of complicated question is something that your state utility regulators have to unpack.

[00:27:26] But you have to take costs that occur at a wholesale level, and you've got to then say, on the retail side, who should pay and at what rate? And that is for, again, this sort of conversation around federalism, right? That is for states to handle. All right?

[00:27:51] And so, as a former state regulator, I'd say, if these data centers are coming, we're going to need infrastructure. Get a streamlined citing and permitting to the best of your ability while still listening to the landowner and listening to the consumer. Okay? You've got to work out these formulas to make sure that residential rate payers are not being overly burdened by this data center growth.

[00:28:16] What's the one thing you wish tech leaders who are building the next generation of AI infrastructure understood about the grid that they clearly don't at this time? It's a really fascinating question.

[00:28:32] I think that the data center companies that we interact with now have a realization that the power system and all of the regulatory frameworks that surround the power system do not move at the same speed that they move at.

[00:28:56] And because of that, there is going to be this period of time where you can build, for instance, a data center in one to two years. But it'll take you five to seven years to build the power plant conceivably to power that data center. And that in that period where we're trying to manage a grid, we can get a data center on the system quicker than you get a power plant or a transmission line.

[00:29:22] There's going to have to be some collaboration and some accommodations made in order to keep the system reliable and affordable for everybody else. Thank you, Asim, for joining us. Thanks, Sherelle. Really appreciate the time. Asim Haq is not in the business of tech innovation and AI growth. He's in the business of keeping the lights on.

[00:29:51] And what he's telling you plainly is that right now, those two things are in tension. The data centers are coming. The federal government has said so. State governors have said so. The hyperscalers have said so. But the grid, the infrastructure that was never designed for this moment, is being asked to absorb all of it faster than it was built to move. And when the bill comes, who's going to pay it?

[00:30:19] And here's the other question that sits underneath everything he just said. If the regulatory system can't move at the speed of the build and the infrastructure can't move at the speed of the demand, is there anything that can close that gap before the math stops working for everyday people? Varun Sivaram thinks there is. He's a physicist and an energy researcher.

[00:30:47] And his argument is one of the more counterintuitive things you'll hear in this series. That the technology creating this crisis might also be the most powerful tool we have for solving it. Not by slowing the data centers down. By making them smarter about when and how they draw. We're going to take a short break and then come back with Varun's talk.

[00:31:19] On a blistering hot day in Phoenix, Arizona, as a million air conditioners drove up demand on the power grid, a cluster of energy-hungry artificial intelligence servers bucked the trend. They actually helped. For three hours, these AI computers at an Oracle data center dropped their power consumption by 25% to provide perfectly timed relief during that day's peak demand.

[00:31:49] And, critically, the advanced NVIDIA chips continued to meet the stringent performance requirements of their tasks. Training, fine-tuning, and using AI large language models. Our team at Emerald AI orchestrated this first-of-a-kind demonstration of flexible AI computing. And we're not alone. Google's also made impressive strides. Scaling up our technologies across the country

[00:32:19] and around the world could help solve one of the biggest challenges of our time. Powering the AI revolution. While also advancing a more reliable, affordable, and clean power grid. Far from undermining it, AI could actually help save the grid. To understand why, we need to reimagine the challenge of powering AI. And so I've reinvented my own career.

[00:32:48] For 15 years, as an energy executive and as America's lead clean energy diplomat, I focused on building more clean energy. But energy supply is just half the equation. And so I founded Emerald AI to focus on the other half. Demand. Helping AI intelligently use energy, support grids, and unlock massive, stranded power capacity that already exists.

[00:33:17] Without this capability, we face an impending crisis. A historic collision between two multi-trillion dollar networks. The network of AI data centers that's rapidly growing and an aging electricity grid utterly unprepared for all this new demand. That's bad news, folks, for multiple reasons. First, America risks falling behind in AI. In Virginia, the data center capital of the world,

[00:33:47] it takes up to seven years to connect new data centers to the grid. Second, power prices are soaring for communities. Just in 2025, as we built new grids and new power plants, data center demand drove up the average annual household power price in Columbus, Ohio, by $240. And this is just the beginning. As data centers surge from 4% of U.S. power demand today to 12% by 2030,

[00:34:16] it's like adding another Germany to the U.S. power grid. And third, fossil fuels are set to power the boom in AI data centers, which require reliable power supplies today. In the United States, natural gas is powering most AI growth, and countries like India will see rising coal use increasing global carbon emissions. But it doesn't have to be this way. The biggest new user of electricity

[00:34:45] could actually be our grid's greatest ally. The key lies in something deceptively simple. Flexibility. That's distinct from efficiency or using less energy overall. Rather, if AI were just a little more flexible in when it uses energy, it could consume vast amounts of otherwise stranded power on today's grids. Think of our electric power system

[00:35:15] as a superhighway that faces peak rush hour just a few hours per month. Think of that hottest day of the summer in Phoenix, Arizona when air conditioning demand peaks. On those days, grids risk being overwhelmed by these massive new data centers that may soon consume more than a gigawatt or more energy than the state of Vermont consumes. But most of the time, power plants are running well below their full capacity and transmission lines are carrying less power than they could,

[00:35:44] just like that highway. On average, throughout the year, half of the power system's capacity goes unused. What if during those peak rush hour periods when the grid is truly stressed, AI data centers could dynamically reduce their power consumption and take advantage otherwise of all that spare capacity throughout the year? It would be like briefly taking 18 wheelers off of that road to let the remaining traffic flow smoothly.

[00:36:13] Well, it turns out that if AI data centers were just modestly flexible, just less than 2% of the year, trimming demand by a quarter, just a couple hours at a time, America could fit up to 100 gigawatts of new data centers on existing power grids across the country. That's $4 trillion of AI investment unlocked today without waiting years for new infrastructure.

[00:36:43] Now, to be sure, America will need even more energy to power our growing economy as data centers, factories, and other users of electricity join. But by making AI data centers flexible, we can prudently expand our grid and buy ourselves time to build clean nuclear or geothermal power plants. And what's more, with flexible AI data centers

[00:37:12] acting as giant shock absorbers on the grid, we can integrate intermittent but cheap solar and wind power driving down the cost of energy for AI. So, that's what I do. My team and I are building the software brain to give AI data centers this crucial flexibility. It's an AI for AI. We call it the Emerald Conductor. It works by harnessing something we call

[00:37:41] spatiotemporal flexibility. That's a fancy term for a simple idea. Let's break it down. First, temporal flexibility. Not all AI jobs are created equal. Some workloads, like training or fine-tuning an AI model, conducting deep research, or running a massive scientific simulation, are what we call batchable. They're incredibly important, but they don't have to be completed right this second.

[00:38:11] Software can intelligently pause or slow these workloads briefly when the grid is stressed and then speed them back up when there's plenty of power available. Then there's spatial flexibility. Think of your query to a generative AI chatbot. You can't pause the job of responding to that query, but you can move it across the country at the speed of light. So even as we struggle to build electric power transmission,

[00:38:41] we can take advantage of virtual transmission or the network of fiber optic cables that crisscrosses the country and the planet to move AI workloads from a data center in a city where the grid's currently strained, let's say Phoenix on a hot day, to a data center in a region where there's presently abundant power, say the windswept Great Plains. The AI workloads get done, but the grid gets a break right when it needs it most, and the user never even notices because behind the scenes

[00:39:10] there's an AI orchestrating AI. Data centers become smart, cooperative partners to the power grid, and we know it works. Remember that demo I told you about? It happened. In May 2025 in Phoenix, Arizona, we took a cluster of 256 GPU servers and we ran a mix of AI workloads, some highly flexible, others entirely inflexible, and many in between. One hot afternoon, our software received

[00:39:39] a signal that the local utility was going to reach its peak demand, And so Emerald Conductor gracefully reduced the AI computational power load by 25% for the exact three hours requested by the grid. We proved that AI data centers can flex when the grid is tight and sprint when users need them to, but proving the technology was just the first step. The hardest part

[00:40:08] will be to convince the enormous energy and AI industries to cooperate and to change the way that they operate. For over a century, electric power utilities have assumed that their users can't simply reduce their power consumption when the grid faces peak rush hour. Sure, in limited situations, a utility may request homes to adjust their thermostats or large industrial loads to dial down consumption, but these interventions

[00:40:37] are typically tiny and marginal. But AI data centers are fundamentally different with a transformative potential to be flexible. They're massive energy users compared with tiny household loads that need to be aggregated. They respond faster and more gracefully than large manufacturing facilities, and they can move their workloads around the country at the speed of light, which no other energy user can do. That's why I'm so excited

[00:41:06] about initiatives that bring together the energy and technology industries like EPRI's DC Flex. In upcoming demonstrations in the United States and with National Grid in the United Kingdom, Emerald will showcase how AI workloads can flex and move across regions and will prove that software like Conductor can orchestrate a symphony of AI workloads in concert

[00:41:36] with on-site energy equipment like batteries to deliver even more flexibility to power grids. And with our partner NVIDIA, we're building a reference design for next generation data centers or AI factories to be power flexible so that utilities that see the certification can more swiftly connect a grid-friendly AI factory. So, where does this all leave us? Well, it means rather than wait years for grid upgrades,

[00:42:06] we can build all the AI infrastructure we need right now to sharpen our competitive edge. And far from crashing the grid, flexible AI data centers can provide relief before the grid hits a breaking point, avoiding rolling blackouts. Rather than increasing power prices, they could actually go down as flexible AI data centers more effectively utilize the existing energy infrastructure, deferring

[00:42:35] expensive upgrades. And rather than goose demand only for fossil fuels, AI's soaring energy needs could encourage more clean energy onto the grid at home and abroad. Solar today is the cheapest, fastest-growing power source on the planet. Imagine flexible AI data centers capable of ramping their energy consumption to match daytime solar peaks or shifting

[00:43:05] their loads so that they better integrate clean energy onto the grid. The AI revolution is here. And I believe we can have it all. Breakneck innovation, massive investments in AI, and abundant, affordable, affordable, reliable, and clean energy for all. An AI for flexible AI infrastructure could be a linchpin

[00:43:34] for our future energy system. Thank you. That was Varun Siviram at TED Countdown. Here's what Varun and Asim Haq agree on. The grid is not going to hold at its current pace. Something has to give. Either the demand slows down or the supply catches up. Or the people who never asked for any of this keep absorbing the difference in their bills,

[00:44:03] in their air, in their water, in the reliability of the power coming through their walls. Varun's argument is that data centers don't have to be the villain in that story. That the same technology driving the crisis can be retooled to stabilize the grid, flexing its demand in real time, filling in gaps instead of creating them. It's a compelling vision, one that might allow us to balance innovation with core values of community,

[00:44:33] equality, and sustainability. But technology doesn't have values. It goes where the incentives point. And right now, the incentives point towards speed, towards scale, to where the next data center breaking ground before the last one has figured out where its power is coming from. A smarter algorithm won't change that, but accountability can. Along with the decision, made by regulators, by legislators, by the companies building this infrastructure,

[00:45:02] that the people at the end of the line matter. That the rate payer in central Maryland, the farmer along the transmission corridor, the community that never got a vote, are not just line items in a cost model. That decision hasn't been made yet, not fully, not in a way that holds. And here's why that matters right now. The rules are still being written. The interconnection standards, the grid protocols, the state permitting frameworks, the utility

[00:45:32] contracts, they're all in motion. The decisions that get locked in over the next two or three years will shape how AI infrastructure gets built for the next 30. That's how infrastructure works. The rules made when a technology is new tend to stick. But there are people who are fighting to have a role in that decision. You've met some of them on this series. And they'll keep fighting because technological innovation is only valuable when people can truly benefit from it.

[00:46:01] And that concludes our final episode in our data center series. But it won't be the last time we talk about this topic on the show because we'll be watching this fight closely. So stay tuned. You can check out the rest of the series in the TED Tech feed. Thanks for listening. TED Tech is a podcast from TED. This episode was produced by Rahima Nasa. Our editor is Alejandra Salazar, and the show is fact-checked by Julia Dickerson. Special thanks to Constanza,

[00:46:31] Gallardo, Daniela, Belarreso, Maria Ladias, Tanzika Sangmanivan, and Roxanne Highlash. If you're enjoying the show, make sure to subscribe and leave us a review so other people can find us too. I'm Sherelle Dorsey. Let's keep digging into the future. Join me next week for more.