Could AI Accidentally Solve Climate Change? — With Amy Harder

0:00Hello again from Newsweek headquarters at the top of the World Trade Center in New York City. I'm Carlo Versano, and this is another edition of The 1600. This country was stone cold dead. So welcome back, everybody. We're going to take a little bit of a break from our recent beat of fire and brimstone, war crimes and blasphemy, and talk a little bit about energy and climate today. I don't know if anyone else remembers, but climate change was a major political priority, at least for the Democrats, at least relatively recently within the last couple election cycles. You don't really hear about it so much anymore, except when the Trump administration is unraveling some regulation meant to keep the air and water clean.

0:39Democrats seem to sort of abandon climate change. I think they've probably taken from their recent electoral failures that this is just not an issue that moves voters in any sort of meaningful way. That said, we have the AI boom now and we have the energy demands that go with that, which is creating some new momentum in the space, trying to figure out just sort of how to satiate all of this energy demand without further burning up our planet. So for that, to get into all of this today, I have with me a guest I've been really looking forward to talking to, Amy Harder. Amy is the national energy correspondent for Axios.

1:09She also writes the great Harder Line newsletter. Amy, welcome. Thanks so much for joining us. You're very welcome. Thank you for having me. So let's get right into it. I want to talk to you about some of the reporting that you've been doing on the data centers, as I just mentioned. Spending on these AI data centers, you know, it's off the charts, right? It's rivaling investments in entire energy sectors in some cases. There's major implications for power demand as well as emissions. This is a demand shock, really, I think, unlike the grid our grid has ever seen before. So my question to you is, can you talk a little bit about sort of the mismatch between what AI needs and what the grid can actually deliver?

1:52For sure. Well, at a very high level, I just want to underscore some of what you said in your intro, that this is just a wild time on the energy and climate beat. and everything you laid out, you know, for a while we are all focused on climate change. Before that, I was covering a lot of fracking throughout the United States, that controversial drilling technology that unlocks natural gas. And now we also have the Iran war, which is a huge energy story. And it's just so surprising, even to me, having done this for almost 20 years, that this beat can be so different in any given year or time. And the really interesting thing about the AI boom is that, yes, it's a huge deal to the energy industry, but it also matters so much to everybody on this planet because we're all either using AI or scared of it.

2:42And either way, it's going to affect our lives. And so I just think reflecting on that as a human, I think, is really important as I cover this beat. So that's just some thoughts at a very high level. Yeah, I agree completely. Coming back to your question, the electricity industry in particular has a very stubborn reputation of being a little bit slow moving, somewhat intentionally because of regulations, both at the local and regional level, but also at the federal level, and for being slow to integrate new technologies and new innovations. and now it's being compelled to move really fast and maybe break things as the saying goes in Silicon Valley.

3:25And I think what you're seeing is some of these projects, these massive multi-city level projects. And by that, I mean data centers that demand the amount of electricity that you would need for multiple cities. So that's just, it's almost, it's hard to fathom. And so some of these projects are not going to be either on time or they might just fall through altogether. And you often see that in these boom cycles where so much is thrown against the wall and not everything is going to stick. But a lot is going to stick. And the grid, which, by the way, is not just one single grid. It's actually several different ones that are connected in most cases, but in some ways not as much as you would think.

4:15We're in this rapid moment where the industry, the electric industry, is trying to keep up with the tech industry that moves fast and breaks things. And so it's a really fascinating time to be covering this topic. Hey, guys, quick ad break here, and we'll be right back. It's a point that I hadn't really thought about because, yeah, you hear in Silicon Valley that the move fast and break things mantra. And, you know, it's one thing to, you know, to crash a piece of software over the course of, you know, a month or two maybe. be. It's another thing to build a power plant. Even without the regulations, presumably, this is just the shovels in the ground to actually getting it connected to the grid.

4:52I mean, you're talking years, right? You're talking years, I think, and multi-billion dollar investments. One huge winner in this debate is natural gas. I mentioned a few minutes ago the fracking technology that really was unleashed alongside horizontal drilling and beginning in the 2010s, we have so much natural gas in this country that even though there was an extreme crisis of natural gas globally because of the Iran war, prices here in the United States are basically the shrug emoji. It's really like we are fully insulated because of the amount of natural gas we have now. And in fact, so many of these data center companies and the hyperscalers, who, by the way, and maybe we'll get to this, they're trying to stay committed to their climate goals.

5:41But it is difficult when natural gas is one of the easiest things to build out. Now, nothing is easy right now. There's a years-long wait for the turbines that you need for natural gas plants. But on the other hand, there's a years-long wait to connect to the grid, which is what you need to do with wind and solar power. So there's not a lot of fast options. Now, I will say because of that, there's this growing trend to build power plants on site right next to the data center and basically have what are called energy islands. This is exactly what I was going to ask you about. Yeah. No, I didn't mean to interrupt.

6:24But I wanted to ask you about this high stakes fight that's happening right now. We've done some really good reporting on this. Basically, the fight's about, you know, should these data centers plug into the grid or operate as what you just said, energy islands? And so there's one example that I know of where Chevron is working on a dedicated natural gas plant for, I think it's a Microsoft data center in Texas, right? And that would be an on-site power. So that would be one of these islands, right? Correct. From what we know in the public sphere as of now, I mean, they are both Microsoft and Chevron and the other investors involved are being a little cagey, as is the whole industry, because things are moving so fast.

7:03But that, I believe, is the current plan. Now, many of these companies say that eventually they will connect to the grid when the wait to connect to it isn't so long. But there may be implications of this interim period where they're not connected, where there could be repercussions. And I think we're just an unknown – we're in unknown territory. Well, that was also what I wanted to ask you because there's a big – a lot of this is happening because of this – there's a big public backlash that's brewing, getting bigger with people saying, look, first of all, we're not even sold on AI as a technology yet.

7:42All we hear about AI – and obviously I'm generalizing here – but what I've heard from a lot of people is them saying we're talking about a technology that's either going to kill us or take our jobs. At least that's how it's being framed by some of the people who are making it. But put that aside, it's also sucking up all of this energy and making our electric bills go higher. Are there any downsides to you and me or Joe Schmoe if the data centers go off-grid? I think immediately it's not clear that there would be. So the two sides of this argument are – it's a little bit nuanced. So here's how I would try to simply explain it.

8:18So at a high level, it could make sense that the industry wants to be off the grid to not burden the electricity grid with their big power demands. And so intuitively, as humans, that might make sense. But if you talk to people inside the electricity industry, what they say is that the systems operate best when you have really big power loads that can be part of the system and help balance it out. And, you know, going back to high school fractions, you know, you want a big, lower common denominator to spread the costs out. And so if you take off the grid these massive power centers, then actually the cost of maintaining the grid could be higher for those remaining.

9:11So even though it's a more complicated argument, which is very difficult to convey in our public debate, that's what the electricity industry is telling these move fast and break things AI industry. I mean, this stuff is hard, and I'm not an expert like you are in this. So, again, I say to our listeners, bear with me because I'm trying to understand this stuff too. But back to the Energy Island thing. So I was just reading this in – I forget where it was, but I think it was like a trade publication. So Meta announced recently that it's planning the largest purpose-built power system for a single corporate end user in U.S.

9:48history for this data center in Louisiana. They're calling it Hyperion. Is that what it is? I think so. It's hard to keep track of all of these. Yeah, no, I know. But that's irrelevant. So. So but Meta is putting up the build out cost and Entergy, which is the big utility down in down in Louisiana. They're they're actually building it. So Meta has a four year lease exit window, which I my understanding is that that means that four years from now they could bail. Doesn't that mean that the Louisiana ratepayer or the taxpayer, but I guess you would call it a ratepayer in this case. Does it mean that they're taking on most of the risk over the life of this plant?

10:25Or am I totally like off base here? I don't know the details of that particular project, but I do think at a higher level and generally speaking, there is a concern that these tech companies are coming in and procuring a bunch of power. And sure, in this moment, it can feel like this is what's going to be the case for the rest of our lives. But it might not be. I think we should have a little bit of humility. You know, four or five years ago, it was all about climate change. And that topic has evaporated. So I think... I mean, meta was all about the metaverse four or five years ago. They even changed their name.

11:01Right. How did that work out? You know, this is getting slightly outside of my wheelhouse, but, you know, the tech industry is used to, and you alluded to this earlier, the tech industry is used to moving fast and breaking things with relatively affordable things. But massive power plants are not very cheap or easy to move around. And so I think there is an understandable concern that if there is some sort of bubble with AI or if the demand for it doesn't materialize, what will we do with these huge power systems? No, exactly. That's a more articulate way of putting the sort of in the weeds question that I asked you.

11:43A lot of these companies, so I know Microsoft's one of them, Amazon's another, Google, Meta, I think Oracle's in there. They signed this pledge with the Trump administration to negotiate separate electricity rate structures with these utilities so that the data centers don't stick ratepayers with the bill. That's the idea, at least. That's the way that it's been pitched. That's a voluntary pledge. I mean, do you believe them? I mean, should we believe them? Would a voluntary pledge be enough here or is this something that demands legislation? Well, I think, you know, generally speaking, we should, everybody should always take with a grain of salt anything that's voluntary in a capitalistic society.

12:19That said, these tech companies have very deep pockets and they can afford really expensive energy in a way that most of us could and could not, most other electricity consumers. And so in my conversations with them, they do seem very committed to paying their fair share of the electricity. Now, how does one define a fair share? And there are, you know, the president's initiative on this has had sort of various different impacts. And one is actually, in my conversations with top executives in this space, that pledge has actually prompted some of these companies to go off grid to create these energy islands.

13:05Because, again, from a high-level, relatively simplistic perspective, it makes sense that I'm just going to stay out of the way and not burden the grid. But I think when you talk to others in the electricity space, what that pledge should ideally do is prompt a more nuanced conversation that involves things like, for example, equipping data centers to be flexible with their compute so you can flex it up and down with the various demands of a broader electricity system. There's a startup called Emerald AI that is trying to do that and is partnering with NVIDIA to make that a reality. Another data center developer, Cloverleaf Infrastructure, their whole thesis is that there's actually a lot of electricity around the country that's going unused most of the time.

13:57And how can these data centers sort of soak up that unused electricity capacity and put it into data centers? So it's these more creative and more complicated solutions that some in the industry are pursuing, but others are just plowing ahead with natural gas power plants, fuel cells with natural gas right on site. So there are a variety of ways that this this AI race is happening. Well, another thing that you've written about that I find really fascinating is carbon capture. And I know there's a handful of these projects under consideration right now that would pair the natural gas you're talking about with carbon capture technology for these data centers.

14:41Right. So I think the the the the idea is the I guess what I'm trying to say is, is this the thing that finally makes carbon capture thing? Because what we've always been told is like, oh, carbon capture is one of those things. It's not economically viable in terms as a as a sort of like climate solution on its own. The question for you would be, you know, is the A.I. boom accidentally sort of solving a problem that, you know, environmentalists couldn't, for lack of a better phrase? Well, you know, one thing that I've said in this whole debate is that during the earlier part of this decade when climate fervor was at a fever pitch, you know, back then, although it maybe not didn't feel like it in the moment, climate was a nice to have.

15:20We were coming out of COVID. We were having existential consideration of our lives and our planet. And it felt like it was really important. But now energy is a must have. And for now, most of the hyperscalers are at least attempting to keep up with their climate goals while rapidly procuring energy. And since that's the case, you are seeing several different commercial scale carbon capture projects being discussed. Now, the key word there is discussed. I would say Google is the furthest ahead. They have publicly announced one called Broadwing in Illinois. There's another in Nebraska that they haven't yet confirmed, but local media and those in the industry say is likely Google.

16:07But it remains pretty expensive. Again, but we have these deep pockets from the hyperscalers that back in the earlier part of this decade when climate as a topic was more prevalent, they didn't need to pull out their deep pockets for that. It was more of a, I don't want to say it was just marketing, but it was a lot more talk than action. Now they're actually, they have a capitalistic reason to reach into those pockets and pull out some money and put it toward not just carbon capture, a whole range of clean energy technologies. Right. And some of these are moonshots that, you know, could work and could not work.

16:44But the point is that they're at least now investing in some of this stuff that we had been told is like, it doesn't really work. Like you said, it's nice to have, like maybe we could make it work, but not at scale. Carbon capture is just one example, right? Exactly. And it's really, I think, you know, again, with a lot of asterisks because we don't know where the future will go. I think the sort of a counterintuitive take, you know, after President Trump gutted the Inflation Reduction Act and billions of dollars worth of subsidies, The AI boom may really be the capitalistic push that gets things like fusion and advanced nuclear power and carbon capture, you know, to the starting line and to be commercialized.

17:30And that would be a huge win for climate change, even if it wasn't in the name of climate change. Yeah. No, I mean, that would be that would be like the best case scenario. If Trump accidentally, his legacy is accidentally being the president to figure out how to stop climate change or at least, you know, work towards it, that would be something else. And I mean, things have happened, right? I would say that it would be more of a coincidence that the AI boom happened underneath him, just like the fracking boom was a coincidence that it happened under President Barack Obama. But Obama took credit for it until he didn't.

18:06And, you know, I think Trump will take credit for things like geothermal and fission and fusion. But I think that will be an interesting storyline if it develops. And it's a slow moving topic that might be hard to see happening until it's in the review mirror. Right. Well, let's stay on Trump's energy agenda for a minute. You mentioned the inflation reduction, how he gutted the IRA. I want to talk about that in a second. But first I want to talk about the EPA rescinded the Obama era endangerment finding. I think it was from 2009 related to greenhouse gases, which is the thing that sort of underpins, at least at the federal level, all climate regulation.

18:49When that happened, I mean, was that symbolically, obviously it was symbolically important. What's the real world impact for that rescission? You know, I struggle with this as a journalist because, yes, it is a very big deal symbolically and legally. But from a tangible everyday perspective, per your question, the lawyers and agencies and different companies have been suing each other over regulations stemming from this finding ever since 2009. And so it really hasn't ever been implemented fully. So its impact has been almost nil since then, which is crazy to think about. Now, even though I say that, you know, the impact, one, from a global perspective, the fact that the United States is pulling back on this is huge from a global leadership perspective.

19:47Right. There are a lot of other factors there that Trump has also very clearly pulled the United States back from. And climate is just one of many issues. And so I think, you know, because the levers of change are really outside of the government right now, I think these regulatory battles are less important, even if in the long run of history that they remain important. And one asterisk to that is the emissions, you know, have helped reduce emissions in the transportation sector and prompted more fuel efficient vehicles. That is a direct correlation from that endangerment finding. But going slightly in the weeds here, the Transportation Department already had fuel efficiency standards.

20:40The EPA was on top of that. Yeah, Obama didn't invent that. Exactly. So, you know, when it happened, it was a big deal. Back in Trump's first administration, I remember thinking that would be the third rail if he did that. And, you know, he's crossed a lot of third rails, this administration, that those of us back in the first term thought wouldn't be possible. Right. Well, and it's sort of the point, right? It goes back to like Steve Bannon's whole thing about like you just keep doing things and people just can't keep up with them. And I think that's what we're you know, a lot of that is what we were just talking about.

21:12We're seeing this with climate. Right. I mean, you mentioned this, the big, beautiful bill, the big reconciliation package he passed last year repealed all of these clean energy credits that Biden had shoved into the Inflation Reduction Act. And there are some projects that are already getting canceled. But you haven't really seen a lot of political backlash. I know you're not I know you're not like a strict political journalist, but from your from your perch, you know, just covering this industry, why do you think it is a clean energy doesn't really move voters? Oh, geez. You know, I'm sure people have written books about that.

21:48I think one reason why you may not have seen as much pushback in the last year and a half due to all with in response to all of the cutbacks and clean energies, because, as you said, there was a lot going on in the world. I mean, last year, all of the cuts in foreign aid was just historic and demanded a lot of attention. And so there's only so much we as humans can hold in our brains at one time. And I share that as a challenge for me as a reporter deciding what to focus on. Oh, absolutely. And, you know, that's intentional. Flood the zone. They did that in the first Trump administration. And it just keeps getting more insane.

22:30And it's just it's it's it's kind of overwhelming to think about, to be honest. And I think so that's my high level response to that. More specifically, you know, yes, Trump did repeal a lot on the clean energy front. But my hot take, which I've heard from others, and actually I got this idea from or this concept from a source, you know, take away the Inflation Reduction Act. Pretend that never happened. It was a historically large clean energy bill. Many would say too big. Now, what Trump actually retained in his bill, if you look at that just on its own, it would actually be the largest clean energy bill in history.

23:12But because he actually retained a lot, he retained a lot of funding for energy storage, geothermal, advanced nuclear power. There's actually a lot in there on clean energy. It's just it was so gargantuan before that. So it's an interesting perspective shift that I think. Oh, I like that. Yeah. And so and even now today, fast forward, you know, about a year, you know, more than a year since that bill, the Trump administration is coming around to solar. You're seeing some in the MAGA industry supporting solar. Katie Miller, Stephen Miller's wife, has supported solar. The rumor on the street is that they're getting paid, which is a reminder that everybody will do something if they get paid for it.

24:02Right. Exactly. That was that was what I was going to say. But it actually it leads into my next question, which is, you know, he needs all of this. Trump needs all of this new power to run this A.I. economy that he is promising and that, frankly, the stock market is. I mean, we're you know, we're talking on today, Thursday. day. Stock market hit a record high today, right in the middle of this war and this huge energy, you know, crunch that we're seeing around the world. Why is that? Well, a lot of it is because of this AI CapEx and all of this AI buildout, right? So Trump needs that buildout to continue.

24:31Now, he's simultaneously, you know, he's been undermining the fastest and the cheapest sources of energy, which are renewables. I'm not saying renewables solve all problems or even a lot of problems, but they're fast and they're cheap. Why is this administration, putting aside the points that you were just making, why do you think they're so bearish on renewables? I mean, I know that Trump himself has like a specifically weird thing about wind turbines in particular. He hates them. But is there more to it than that? Well, so at a very high level, yes, Trump is generally speaking anti-renewable energy of all kinds.

25:08When you go down a little bit deeper, you hear them say that maybe renewable energy is cheap if you look at it in a vacuum, but actually it's not always sunny or always windy and you need something to back it up. And the fact is they're right. Oh, sure. And wind and solar are more expensive when you factor in the system costs of backing it up. Now, wind and solar and batteries is really competitive with natural gas, but the exception is offshore wind. Now that is where everybody I talk to, you know, it's almost like offshore wind is a punchline and a joke inside the energy industry. And, you know, the other day I was at this private energy conference and people sort of muffled laughter when they'd say, well, anything except offshore wind.

26:02And I say, I'm like, hey, guys, it might be offshore wind today, but it might be something else tomorrow with the next administration. So you might want to think about it. Head your bets. Head your bets and have some empathy for this billion-dollar industry that has just been demolished by Trump. It's really, somebody needs to write a book about just the fall into oblivion that has been the billion-dollar offshore wind industry. And the arguments there, so yes, before we get to the factual arguments, Yes, the general sentiment is that Trump's golf course in Scotland, there was an offshore wind project and he didn't like the view.

26:41I don't know if there's anything more than that, but that really is where it began. The complaint about offshore wind, which is true, is that it's much more expensive than even onshore wind and solar. Now, you build the turbines, you got to get them out there. Exactly. My response on that is most new technology are really expensive. So is fusion. So is carbon capture. For whatever reason, because the reason being Trump doesn't like offshore wind, offshore wind isn't getting the same sort of grace that these other expensive technologies are. One would think that if you have enough consistent investment in something like offshore wind, the prices would come down like they are projected to for things like.

27:26Which is what happened with solar. Exactly. But because of Trump's hatred of offshore wind and the industry's reticence to do anything more with that, it's unfortunately, at least in the United States, going to stay expensive. Yeah. Yeah. It's like uninvestable right now. I think part of it with Trump is also, you know, these things, I think especially to, you know, the boomer generation, these things kind of code is like liberal, you know, kind of like hippie dippy. Oh, yeah, you're going to run your house on, you know, solar panels. But what I wanted to lead this into is to talk a little bit about nuclear, because I've always been kind of out on a limb, less so now, but in my own life, just being like, why pounding the table?