Trump grapples with the AI data centre debate - The President's Path podcast, BBC World Service

[00:00:00] Speaker 1: The Trump administration has really grappled with this because AI investment and digital infrastructure, you know, are really a foundational pillar of his agenda and the U.S. economy. [00:00:13] Speaker 2: It's such a fascinating combination of the affordability debate, kind of the NIMBY debate. So not in my backyard, because some people say these data centers are ugly. You know, why do we want them here? We don't want them in our community. Plus the politics behind AI as well. [00:00:28] Speaker 3: The China aspect is hugely important because this debate isn't happening in China. If they want to build large, massive data centers for their own national objectives, they can and are. [00:00:47] Speaker 2: Hey, guys, really good to see you. Great to have our favorite White House correspondent at Bloomberg, Courtney Subramanian, back with us. Hey, Courtney. Hey, Sumi. Hey, Bernd. Thanks so much for having me back. [00:00:59] Speaker 3: Welcome back. [00:01:00] Speaker ?: Thank you. [00:01:01] Speaker 2: Well, we should say you had that lovely backdrop, Courtney. And Bernd, I know you're in a little booth at the White House. But here we are because, you know, we teased this actually at the end of last week, Bernie, when we talked a bit about AI and the White House's AI policy. We talked about data centers and just said that needs its whole own episode because it's become such a huge issue. And I guess just to level set what data centers are, there are these kind of physical infrastructure. So these buildings that house the servers, the computers that kind of power the internet and, of course, increasingly going forward, the AI that we've been talking about as well. So I saw in that newsletter from Punchbowl that they called data centers the new bipartisan punching bag. And I think that pretty much sums up what's been happening. [00:01:50] Speaker ?: Yeah. [00:01:50] Speaker 1: I mean, it's become a real flashpoint throughout the country, right? We have lots of polling showing that Americans are really turning on data centers. There's a recent Gallup poll that showed 71% of Americans oppose constructing data centers in their local areas. And I was looking at some modeling from the open energy outlook at Carnegie Mellon, and that found that data centers could increase electricity bills 8% nationally and as much as 25% in some regional markets, which, as you guys know, and I mean, we've been talking about this and when I've appeared on this podcast over the last year, that that really drives home the whole affordability question and crisis, which has been at the heart of, you know, national politics over the last year. [00:02:50] Speaker 3: I think that's kind of why it's such a challenge for the White House to address the issue, because President Trump, of course, we were all at campaign events in 2024 and have heard him speak since then. Electricity is one of the first things he brought up on the campaign trail and bringing down electricity prices. It's something he's mentioned kind of in more recently in these events leading up to the midterms. But at the same time, the White House is very conscious that it doesn't want the U.S. as a nation to be left behind. And in their view, data centers are necessary to compete with the likes of China in terms of AI, for example. So it's really quite tricky. And I think it's really telling that, you know, some of the local state papers, I read one in a Pennsylvania paper that said Donald Trump is kind of broadly out of line with public opinion on data centers, even amongst other Republicans. So I think it's really challenging for them to kind of address this and get it right. [00:03:49] Speaker 2: Yeah, I guess we should say the reason this is so important is some of the biggest AI companies think OpenAI, Anthropic, Microsoft, so on and so forth. They are trying to build out data centers because they need the energy to power, again, the artificial intelligence growth that we were talking about in the last podcast. And that means putting these data centers somewhere, of course. And what you guys have been referring to is just people saying, we don't want these data centers in our backyards. It's really interesting if you look at some of the data behind, you know, the electricity usage, water usage, which has become another kind of controversial topic in some of the communities that could be hosting data centers or are already hosting them. Some of the data is kind of mixed. You know, I went to a briefing with a company that does build data centers, and they were saying, look, you know, there are some really important points to keep in mind, of course, you know, in favor of data centers, which is they may add to the electricity load, but they kind of utilize the network, the grid network, a bit more heavily. And that means that it flats the cost for everyone. They also point to the fact that they raise a lot of taxes for the towns where they're located as well, so that can push down property taxes. So there's some of that. But some of the data is mixed on water and electricity usage. But as you said, Courtney, the consensus is at this point for most of the information out there that they do push up electricity and water usage in the communities where they're located. And it's such a fascinating combination of the affordability debate, kind of the NIMBY debate, so not in my backyard, because some people say these data centers are ugly. You know, why do we want them here? We don't want them in our community. Plus the politics behind AI as well. [00:05:41] Speaker 1: Well, and all of those points, I think it makes it such a fascinating story. But there's also even an environmental component here, too. I was talking to Tom Steyer earlier this week who ran for governor in California, where, you know, AI underpins the entire economy of that state, right? And we were talking about a recent story in The New York Times about how Amazon is building a new data center in Texas that will incorporate a gas-burning power plant. And if it's built to specification, it will become the biggest CO2 emitter in the entire country. So, you know, this is not just a matter of electricity, of water, but, you know, it's also about how companies can do this in a carbon-neutral way. And that's something, you know, states are going to look at that very differently, depending on their politics. In California, obviously, they're very environmental conscious. But there's just, there's so many different layers here that all come together under AI development. And, you know, to Byrne's point, you know, the Trump administration has really grappled with this because AI investment and digital infrastructure, you know, are really a foundational pillar of his agenda and the U.S. economy. So they've really tried to, you know, put together a strategy that sometimes feels ad hoc because they want to be hands-off, but they understand, you know, some of the concerns around this. But ultimately, it seems like they're very bullish on data centers. The president has come out in favor of them. He's said Texas, where there's a moratorium, is making a mistake by halting the construction of new data centers. So it will be interesting to see how this plays out, you know, in November in the midterm elections as more candidates are talking about it and this White House tries to, I don't know, maybe ignore some of the pushback that we're seeing. [00:07:52] Speaker 3: I think it's quite interesting how they've tried to address some of these issues, particularly the affordability issue. And back in March, there was a pledge by some of the kind of largest tax firms in the country that in places where they built data centers, they would kind of build their own energy infrastructure as well and kind of help develop the kind of wider infrastructure of what they need that the White House kind of believes will ultimately help the other areas. I mean, the wider area, rather. President Trump has even said he thinks in some cases these data centers, if they do this infrastructure project around them, that they could actually bring electricity prices down. But the issue is there is this is a pledge. It's not mandatory. It's not an executive order. It's not something that these companies have to do with certain kind of firm fixed goals in which they would be somehow penalized if they don't actually accomplish that. So I think, you know, it's, you know, very early on in this. This just happened in March and he was it's since been expanded. But I think the results of whether that's the case will ultimately be looked back on as, you know, whether the White House kind of managed the issue of the data centers and the energy infrastructure around it adequately. [00:09:10] Speaker 1: I feel like that strikes at the heart of the Trump administration policy around AI in general. So much of their attempts to try and, you know, put together guardrails around the development of AI has been voluntary, right? And so much of the conversation is should this be voluntary where we have companies buy into these, you know, these regulations or pledges or should they be regulated, you know, by some either independent body or at a state level or at a federal level. And that I feel like, you know, has has been the crux of the conversation about how, you know, we actually, you know, again, put some protection around the development, which is happening at such a fast pace that it's hard to keep up with. [00:10:05] Speaker 2: Absolutely. And I think, you know, interesting, the point you made, Courtney, about, you know, candidates campaigning on being anti data center. These are Democrats and Republicans. What issue do you have these days where you have Democrats and Republicans actually agreeing? And you mentioned Texas, the governor there, Greg Abbott, with his moratorium, which President Trump actually said he thought was a mistake. You also have New York issuing its own moratorium. The governor there, Kathy Hochul, and talk about two states with very different politics, of course. That just kind of gives you an idea of just how widespread this, let's say, opposition to data centers has become. At the same time, I think what's really interesting is it's somehow far beyond, you know, the question of water and electricity and the noise that data centers can make, jobs. I think part of the opposition, or at least a large part of it, in my opinion, is fueled by this kind of very vague sense of what AI is going to do to, you know, the US economy or the fact that you have these large AI companies with these huge valuations, multi-billion dollar valuations, and the sense that these are kind of Silicon Valley elites that are going to benefit from artificial intelligence and that average people are going to be left footing the bill, more or less. So this feels like a much bigger opposition than simply, and of course, it's not just simply, but one about utilities, because there are other projects. Think about office parks that are built in communities where, you know, they're ugly. They take up — I hope I'm not offending anyone by saying that office parks are ugly, but they take up a lot of water and electricity as well. But this is something different, and I think it really is because AI does not, as we talked about previously, have a good, you know, reputation in polling right now, and that's because people feel like they're not necessarily part of the conversation and they're not the ones benefiting. [00:11:59] Speaker 3: I think that's one of the reasons why there's such a disconnect between what the Trump administration and the White House is saying and what we're seeing from local and state candidates, because, I mean, the concern is very different. I mean, there's the concern about not having it in your own backyard because it's ugly and the infrastructure issue. I mean, voters are concerned about what AI could mean for their jobs. I mean, there's been some very alarmist reporting about, you know, what this means for the future of all sorts of professions. Whereas I think the Trump administration also has to view this from kind of a more macroeconomic and national security levels, particularly with regards to China. I think like many things in this administration, the China aspect is hugely important because this debate isn't happening in China. If they want to build large, massive data centers for their own national objectives, they can and are. And I think the Trump administration is concerned about that because, of course, AI also has kind of massive national security implications and military implications even that they have to address at kind of a bird's eye view rather than the state and local concerns that voters and candidates might have down the ballot. [00:13:12] Speaker 1: Just to not to belabor the point, but I do think that, you know, the data center has become sort of the physical embodiment of how Americans feel about AI. They're uneasy, right? You know, are they going to take our jobs? Is this just going to further the wealth gap? We're going to see these big tech companies coming in and getting richer. And it sort of feeds the sort of populist sentiment we've seen. Right. But, you know, I one of the things I was talking about with with Tom Steyer, and I think there's been a larger conversation about is how, you know, Americans can buy into this. Right. And is that a sovereign wealth fund creating sovereign wealth funds? One of the policies he promoted when he was running for governor was a token tax. So charging a tiny fraction of a cent for every piece of unit of data processed by major tech corporations using large scale AI for worker retraining, for cash dividends. You know, the Trump administration has also weighed creating a sovereign wealth fund. And what would that look like? You know, how can people also cash in on, you know, all of the energy that these companies are using and all of the land they're taking up? [00:14:30] Speaker 2: We should say the stakes are really, really high here. I mean, if you think about we were talking in the last week on the podcast about whether there's an AI bubble with these huge valuations that we're mentioning, billions and billions of dollars for these companies, those valuations are based on these companies having the ability to power this AI. So the compute to be able to power this AI. And for that, they need the data centers. And so that kind of chain is really important. And obviously, if that compute doesn't materialize, you could see a huge crash. And obviously, the economic consequences of that would be pretty massive as well, also for the Trump administration to deal with. So it feels like it really is an existential debate. And I think, you know, you, Courtney, that's such a good point, that these data centers are kind of the modern stand in for AI writ large. So I think there's going to be a lot more for us to talk about on all this, especially to see whether President Trump, you know, his position develops on data centers and that, you know, we see some of these more populist solutions, let's say, as to how people can actually feel connected to them like they're profiting. But I think we're going to have to pick that up, unfortunately, in another podcast. So it was really great, really interesting conversation, guys. I'm looking forward to doing this again next week. Love talking to you guys about this stuff. [00:15:48] Speaker 1: Thanks so much for having me on. [00:15:50] Speaker 3: Hey, guys. See you next week. [00:15:52] Speaker 1: See you. [00:16:00] Speaker 2: If you've enjoyed this episode of The President's Path, then like and subscribe to our YouTube channel. We'd love to hear from you. So please leave us a comment below or email us on path@bbc.co.uk.