BRITTANY LUSE, HOST:
Hello, hello. I'm Brittany Luse, and you're listening to IT'S BEEN A MINUTE from NPR, a show about what's going on in culture and why it doesn't happen by accident.
(SOUNDBITE OF MUSIC)
LUSE: What if I told you AI slop, deadbots and chatbots are all part of an ongoing global AI arms race?
EVAN HALPER: So right now the artificial intelligence industry is moving really quickly to innovate. And the worry is that this has become a national security imperative - that whichever country dominates the AI industry will dominate all kinds of sectors going forward and possibly dominate, you know, not just the economy but militaries.
LUSE: That is Evan Halper. He's a business reporter covering the energy transition for The Washington Post.
HALPER: And so the U.S. has gotten, over the last couple years, very concerned that it keep up with particularly China in this race because China is also doing a lot of AI innovation. And the worry is that if we do not get in front of this and that we do not dominate this space that China will take it over.
LUSE: In order to compete in this race, tech companies have started to invest more and more into AI development and research. But AI requires a lot more than just money to work. It needs a lot of space and energy to operate.
(SOUNDBITE OF MUSIC)
LUSE: Enter data centers.
HALPER: So a data center is just this giant warehouse-like building where all the computers that do the AI training models exist. And they've been around for, you know, decades. They're part of everything that happens online.
LUSE: According to a report from the MIT Technology Review, from 2005 to 2017, the amount of electricity going to data centers remained quite flat thanks to increases in efficiency, despite the construction of armies of new data centers to serve the rise of cloud-based online services, from Facebook to Netflix. In 2017, AI began to change everything.
(SOUNDBITE OF MUSIC)
HALPER: What's happened now is as we're ramping up AI and doing more and more and asking more and more of this technology, it requires just immense amounts of energy. Not only do you need this immense amount of energy, but you need an immense amount of space for all these computing chips.
LUSE: For instance, Mark Zuckerberg's Meta is planning a data center nearly the size of Manhattan. And data centers of this size, also called hyperscalers, could potentially start to use more energy than entire cities, and they need to run 24/7. So then you have to wonder, where could they possibly get all this energy? Well, there's the power grid, which has been overdue for upgrades and repair for quite some time. There's solar and wind power, which is cheap, but still requires some battery power to store power when the sun isn't shining or the wind isn't blowing. But there's another option. Some companies are looking to nuclear plants, which are expensive, and natural gas and coal plants, which are polluters. And some of these nuclear, coal or natural gas plants were on the verge of being shut down.
(SOUNDBITE OF MUSIC)
HALPER: I mean, if I can give you an example of how much power, I was just visiting, you know, a data center that's getting built in Pennsylvania on the site of what used to be one of the largest coal plants on the East Coast. They imploded the coal plant. And now they're going to build seven data centers on this campus, and they're building seven gas plants.
LUSE: You can imagine, with data centers projected to be the size of Manhattan, the construction and revitalization of these plants probably doesn't bode well for the environment.
HALPER: These tech companies, before they started needing these massive amounts of energy, were on the forefront of making climate commitments. But I think we're going to see, very soon, them start to change the way their goals work. And now the rationalization is, if we get AI right, it could solve climate change.
(SOUNDBITE OF MUSIC)
LUSE: This is AI + U. As artificial intelligence becomes more prevalent in ways we can and can't see, I'm zeroing in on how AI shows up in our daily lives. For the final episode in this series, Washington Post business reporter Evan Halper joins me to break down how this race for power affects our climate goals, our communities and our wallets.
(SOUNDBITE OF MUSIC)
HALPER: So basically, when you need energy for AI, there's two ways to do it. You can do it on the power grid, or you can do what's called behind the meter, where you just kind of, like, build a power plant next to the AI machines. And, you know, you're not connected to the power grid at all, and just the power plant connected to the data centers just powers that. It sounds like a tidy solution, this behind-the-meter idea, and something that would be easy. But it's actually really complicated because power plants go down for maintenance all the time. And if you're not on the power grid - these AI data centers need to run 24/7, and then if your power plant's down for maintenance for, you know, a month, two months, like, you're just out of luck during that time if you're not on the grid.
So because all these companies ideally want to be on the power grid, they need to look for places where they can do the interconnection and all the infrastructure is there to be able to bring lots of power in really quickly and possibly send power out if they're building power plants on-site to power their AI.
LUSE: So the infrastructure is there. It's not like they have to start from scratch.
HALPER: Exactly. And the coal plants especially - I mean, those things were workhorses, and they were these big generating stations. And often - you know, like, this one I was at in Pennsylvania, it just happened to be built in an area where, like, it was an interconnect for, like, several regions. And so there's all kinds of power that can come and go from that area 'cause of the way the poles and lines are set up.
LUSE: Oh. I wonder, though, about, like, the effects on the community. What have been the effects on residents in these communities? Some of the companies you've reported on have said, you know, this AI advancement would actually be beneficial to the environment. What's the basis to that?
HALPER: You know, it varies widely depending on the community, the part of the country, like, what the economics are. Last year, I was out in Omaha, where it was - I believe Meta and Google had built very large data centers. You know, and, you know, they said, look, we're matching all of this with clean energy. This is powered by clean energy. But it doesn't really work that way because what they do - these tech companies - they bought, like, a bunch of clean energy, solar energy or wind energy, like, a thousand miles away but said it's still on the same power grid.
And lo and behold, to power the data center, this coal plant that was supposed to shut down in Omaha stayed open years longer than it was supposed to. And, you know, it's a - it's in a front-line community, largely low-income, largely nonwhite population. And these - the people who live there have been - you know, they've been fighting to get this thing shut down because, like, you know, all these communities around coal plants, you know, higher rates of asthma, higher rates of other diseases.
LUSE: Oh.
HALPER: And, you know, a commitment was made that this thing would shut down by - I think it was 2023. You know, it's still open.
LUSE: We reached out to Google and Meta for comment on this. A Google spokesperson sent us a statement in response to clean energy matching in Omaha, Nebraska and on their AI environmental impact. In regards to using carbon-free energy, or CFE, in Nebraska, Google said, for our Nebraska operation specifically, you'll see in our latest environmental report that in 2024 our 24/7 CFE score was an 88% in Southwest Power Pool. This means that 88% of our hour-by-hour, round-the-clock operations were powered by a combination of utility provided clean grid mix layered with our clean energy investments. They also mentioned that they've worked with all their utility partners, including the Omaha Public Power District, to bring new, clean, efficient energy to the grids where they operate. Meta has not responded.
HALPER: And, you know, when you do the math, the only way you can sort of reconcile this is, well, you know, Omaha wanted to bring in these data centers. These data centers need a huge amount of energy, and suddenly they couldn't shut the coal plant they were intending to shut down. However, you know, I was just up in Pennsylvania at this old coal plant called Homer City. You know, there's some people who live right near there who were unhappy with what direction this is going in and at least want to know what the future's going to be and what's going to happen. What's construction going to be like, and how's their property going to be affected? But by and large, this is a community that has a big labor force that needs work - right? - boilermakers, steelworkers. And this is a huge opportunity. And, you know, in western Pennsylvania, where the economy is shaky, this is really welcome.
LUSE: OK. I remember reading about a community, I think, in Tennessee - like, either in or just outside of Memphis - where there were a lot of concerns about, like, water usage. Do you have any insight into how that affects communities?
HALPER: So there are lots of parts of the country where, you know, they need to cool these data centers, and so they need, like, large amounts of water. Yet, at the same time, the data centers are being built in, like, very dry parts of the country. So they're building these big data centers in, you know, Arizona, where there's also concerns about the water table being depleted, you know, Texas as well. So water is a big concern in a lot of places because a lot of community activists are getting very alarmed. You know, they're digging into this. They're looking at how much water is being used. You know, they're concerned about their water quality They're concerned about, you know, water availability, and it's becoming a flash point. And so what we've seen is a bit of a shift to try to use less water. But then when you use less water, you still need to cool the place, so it takes more energy. And so you're doing, like, more air conditioning and stuff.
LUSE: There's a lot to be concerned about as a person who might be living in a community that either has a data center or is planning on a data center. We actually got a question about it from a member of our audience. Listener Alexis Gallardo (ph) - or Gayardo (ph), sorry about that - asked, how do AI data centers affect our energy bills, and what amount of consent are we allotted when it comes to where these centers are built? Isn't this something that communities should vote on?
HALPER: So it does. In some places, communities have decided we do not want this here and have really come out in force, and they've rejected applications based on a strong opposition from the community. But what we saw happening a lot and maybe happening less so now but - was these data center companies would come in very quietly because they're competing against each other. And they move very stealthily, and they start talking to these economic development departments. But they do it, you know, using these code names, and it's not clear it's Google or who it is. And so, you know, I was hearing from a lot of communities that by the time they found out what this project was, the tech company was already in there talking to all the officials. Like, the thing was, you know, pretty far down the field. And so there's a lot of frustration about that.
You know, when we raise this with the tech companies, they come back and say, well, you know, this is really competitive stuff, and we have a right if we want to go to a community, like, not to have to broadcast to our competition that, you know, we may be about to build a billion-dollar project here. That seems like it's been a big barrier to consent because people just don't know what's going on on the front end. I think it's becoming harder and harder for tech companies to do it in such a stealthy way as there's more public awareness about these projects and more public concerns about what it means for your community when one comes.
(SOUNDBITE OF MUSIC)
LUSE: Coming up...
HALPER: You know, 'cause in parts of the country, bills went up. Power bills went up 20% over the last year. And these are parts of the country that had these data centers.
LUSE: Stick around.
(SOUNDBITE OF MUSIC)
LUSE: So it's not so clean-cut, basically.
HALPER: It's happening so acutely. So, you know, 'cause in parts of the country, bills went up. Power bills went up 20% over the last year. And these are parts of the country that had these data centers. And so you had this perfect storm of, like, neglect of the infrastructure and then suddenly these data centers just needing a huge amount of power very quickly and then local officials wanting to bring the data centers in 'cause they think they're going to bring tax revenue. And then the grid just gets kind of not overloaded but suddenly needs a lot of upgrades, and someone's got to pay for them. And, you know, these formulas are still being worked out of, like, what is fair and how much should the data center pay for.
LUSE: But I also have seen instances where, you know, either plans for a data center were scrapped or halted at the very least, if not stopped altogether, by community members coming together. Like, I don't know. What recourse do people have if this is something that they don't want in their community or if it's something they have serious reservations about?
HALPER: Yeah. I mean, generally, if you're building a data center, you need to make some zoning changes. You know, it's like any other big project that comes to a community. And they do eventually need to go through a process, even if the whole government there is on board and likes the idea. And that's where the community can step in and pressure the zoning board, you know, the planning board. We see lawsuits filed all the time. And some might call it NIMBYism, but it's the same way you would stop any other project.
LUSE: I was going to say - I'm like, is it different if it's a data center versus, like, affordable housing? Maybe so (laughter).
HALPER: Yeah. I mean, we've had that discussion in the newsroom. It's like, at what point - you know, so the people who are fighting these data centers, are they any different than, you know, your community group who's fighting an expansion of the airport or, you know, fighting the shopping center getting built? You know, I mean, these projects have become so big that I think it's a different category.
LUSE: Yeah, I would agree with that.
HALPER: You know, it does raise that question.
LUSE: Where are most of these data centers located right now?
HALPER: They started in, you know, northern Virginia because, like, you know, the sort of internet grew up there, and it's near Washington and that was kind of the place. People wanted data centers in Silicon Valley, of course. But now they're just getting located anywhere there's power. You're seeing them located in like - you know, suddenly, New Albany, Ohio, is, like, a major hub of data centers. Texas has lots of power, so they're just - they're opening up all over the place in Texas. And so you go to these parts of the country that, you know, we never thought of before as synonymous with tech, and these just kind of hulking, you know, structures are being built there because there are parts of the country that, you know, are not tapping out their power supply, like, you know, like California is struggling to, you know, generate enough power. But Oklahoma probably isn't.
LUSE: Do the construction of these centers disproportionately affect certain communities, whether that's people of color, poor people or those who live in rural areas?
HALPER: Yeah. I mean, because they're being built around a lot of power infrastructure, and the power infrastructure was initially, you know, especially if it's an old coal plant, like we were talking about, more affluent people don't want to live near the coal plant or they don't want to live near the, you know, high voltage power lines or the kinds of things that data centers like. Already, it's self-selecting, and you wind up in these communities that tend to be poorer, that tend to, you know, be less represented in government. I mean, what you call environmental justice communities. It's very common for data centers to pop up in them, but not exclusively.
LUSE: Not exclusively. How do these data centers and the efforts of these tech companies complicate or delay our climate goals? I guess I'm trying to figure out, like, if we are indeed hustling backwards, how far and how fast?
HALPER: The climate issue is a huge issue. These tech companies, before they started needing these massive amounts of energy, were on the forefront of making climate commitments. They all have net-zero goals and very aggressive ones. I think several of them say they're going to be, you know, net zero, meaning they'll wipe out all their emissions, probably using building clean energy to cancel out the dirty energy they're using, but they say they're going to be net zero by 2030.
LUSE: That's coming up.
HALPER: It's coming up very fast.
LUSE: Yeah.
HALPER: And at the same time, their emissions are exploding because, you know, they're using all of this energy. They still say that they're keeping these goals, but I think we're going to see very soon them start to change the way their goals work. And now the rationalization is if we get AI right, it could solve climate change. And so, yeah, we need to build lots of gas energy right now, and that does pollute, but it's worth it because we need to beat China and, you know, that - and we're going to save the country through this, and we're going to save the planet because the AI is so sophisticated, it's going to be able to solve the climate change problem in ways humans haven't been able to do on their own.
LUSE: OK. So given this conversation that we've had and all the reporting you've done, what personal equation do you use to determine what kind or amount of AI usage is right for you?
HALPER: Basically, if you're a journalist, at least my employer, The Washington Post, has made very clear, like, we need to learn how to use these tools. They're going to be a big part of the future of creating the product, of the user experience and things are changing. Outlets I've worked for, you know, over my 30-year career in journalism, like, no longer exist. And, you know, as a 52-year-old man who's trying to figure out, OK, like, I really like writing, I like doing - I like informing the public, I want to stay in this career, I do feel like I need to learn how to use these applications because people are going to be expecting to engage with them in the journalism that they receive. We're expected to engage with them in producing the product.
And so you have to stay on top of it, even if every time you use one of these things, it's nagging at you, especially someone like me. And I know, OK, this is, like, taking a lot more energy than I wish it did. So my policy is use it as judiciously as I can. But to give an example, like, I was working on this really complicated story on recycling spent nuclear fuel and the processes that are involved in that are, like, super complex. And, you know, I did several interviews, but I was still trying to, like, work through and make sure I explained to readers and got this right, so I don't have to have to do, like, 10 corrections on the story.
I found AI very helpful in sort of explaining to me, like, you know, this complicated process called pyroprocessing and electroprocessing and just, like, created a chart for me of, like, OK, these are the, you know, four different techniques and how it works, and this is how different companies - and, like, that saved me days of work, you know? So it's using energy, but it's also creating these efficiencies, which I found very helpful.
(SOUNDBITE OF MUSIC)
HALPER: It's easy with AI because it's just right there, and you start typing in things, and, you know, you start using ChatGPT to plan your, you know, menu for the week, what you're going to, you know, make dinner for your kids. But I think we all need to ask ourselves, OK, how much do we really need to be using this, at least right now, when it takes this much energy?
LUSE: I so appreciate the conversation that we've had today. Am I shook? Absolutely. But I really appreciate you coming on and talking with us. This is really, really important.
HALPER: Yeah, and it's wild how fast things are moving. So, like, if we were to have this conversation six months or a year from now, it can be totally different. So it's - you know, fasten your seatbelts.
LUSE: Evan, thank you so much.
HALPER: Thank you for having me.
LUSE: That was Washington Post business reporter Evan Halper. And I'm going to put on my influencer hat for a second and ask you to please subscribe to this show on Spotify, Apple or wherever you're listening. Click follow so you know the latest in culture while it's still hot.
This episode of IT'S BEEN A MINUTE was produced by...
ALEXIS WILLIAMS, BYLINE: Alexis Williams.
LUSE: This episode was edited by...
NEENA PATHAK, BYLINE: Neena Pathak.
LUSE: Our supervising producer is...
BARTON GIRDWOOD, BYLINE: Barton Girdwood.
LUSE: Our executive producer is...
VERALYN WILLIAMS, BYLINE: Veralyn Williams.
LUSE: Our VP of programming is...
YOLANDA SANGWENI, BYLINE: Yolanda Sangweni.
LUSE: All right, that's all for this episode of IT'S BEEN A MINUTE from NPR. I'm Brittany Luse. Talk soon.
Copyright © 2025 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.
Accuracy and availability of NPR transcripts may vary. Transcript text may be revised to correct errors or match updates to audio. Audio on npr.org may be edited after its original broadcast or publication. The authoritative record of NPR’s programming is the audio record.