01:33:19 · John Coogan
[01:33:19] S0: Do you have an idea of thesis particularly around why America is currently so strong in data center construction? It feels like when I when I count up the the gigawatts, America is doing fantastically. We we are in the lead. Of course, I agree with you on everything. We gotta hold the lead, but I wanna retreat to the source of strength that got us here in the first place because when you tell the story of there's gonna be a new industry, it's gonna be industrial, and it's gonna really start in 2000 and run until 2026. If you had said shoes, I would have said that's gonna happen in China. You know, if you had said any other any other piece of the economy, I would have said that absolutely would have been pushed offshore, pushed out of The United States, and yet we got here to this place where we do have, you know, Virginia and and and, you know, AWS and GCP and Azure and so many massive campuses. And the question is really just, are we going to go bigger? But why didn't we build the data centers abroad in the last two decades before there was all this discussion about near peer competition, geopolitics? It feels like that would have been the natural choice, and yet we did choose to build here throughout the last two decades.
01:34:40 · Josh Zoffer, Deena Shakir
[01:34:40] S7: Yeah. I think it's a great question. I I think it really comes down to two things. One is that, even today still, latency really matters. Right? If you are trying to serve customers that care about getting their data or their inference Mhmm. As quickly as possible Mhmm. You need data centers to be sort of, you know, close to to the customers they're being served. [01:34:59] S9: Yeah. [01:34:59] S7: I think there's a real chance for what it's worth that that matters less and less over time as, you know, as as inference, workloads become more agentic. Right? Like, agents don't care if they have to wait five seconds. If you're away from your computer for two hours running long a long query and, you know, it takes a little longer because the data center is somewhere else, you're not gonna notice that. And so I think to me, what that means is these supply chains are gonna be formed today. Right? The data centers are being built today. The the the factories that supply them are being set up today when speed and reliability and industrial capacity matter. And so I think there's also a real risk that if we wait and those supply chains are built somewhere else, it's gonna be really hard to reverse this. And the point I'm trying to make with the rare earths comparison is, like, we know how devastating that can be. Let's think really hard before we make that mistake again.
01:35:46 · John Coogan, John McElhone
[01:35:46] S0: When you think about the rare earth, situation, it feels like a lot of that stems specifically from China's leverage over The United States. I I think it would be a very different discussion if it was Canada or, you know, Germany. And when I think about the AI supply chain, I go to Zeiss and Trump and ASML and TSMC and SK Hynix, and I'm in Western countries. Do you think that there is a that there is a discussion to be had about the the the the sphere of influence beyond America but in the West where we can partner on data center construction reliably? Or is this, is it is it actually critical to have, the vast majority of data centers built out within the the 50 states? [01:36:37] S5: I don't think they
01:36:38 · Josh Zoffer
[01:36:38] S7: have to be, in the 50 states, for sure. I think there's definitely room to collaborate with allies both in where the physical data centers are located as well as, in the supply chain. So we're not gonna make every single thing that goes in every single data center, in The United States. But but I would say maybe two things that concern me. One is, you know, it came out a couple of weeks ago that the Chinese government is reportedly planning a $295,000,000,000 fund for data center construction. So the Chinese government, unsurprisingly, because they run this play before, they get it. Right? They understand how important all of these technologies and supply chains are. The second is today. There are lots of important pieces of the data center supply chain from the GPU manufacturing at places like TSMC to the ASML EUV lithography machines that really can only be made in, you know, Western allies, like Taiwan and The Netherlands. I think if the history of industrial development tells us anything, it's that we can't rely on that being true forever. And so, you know, taking action today to make sure that, you know, while there's a chance to take advantage of this capacity, and this demand to build next generation technology, industrial capacity in The United States when these technologies aren't commoditized and can't be bought more cheaply from a place like China is kind of you know, it's it's an important window of opportunity. And if we miss that opportunity, again, I think there there's a a world where we really regret it. You know, that doesn't mean just we should build all the data centers. We should have no no restrictions, no regulations. I think that would be a big mistake, and I think that is a way to politically get wrong you know, wrong footed and and generate a lot of opposition, which is why it's important that, you know, some of these energy costs be moved onto onto developers, as well as, you know, things like environmental restrictions being put in place to make sure that we do this the right way. I think it's important that, you know, that to me, the lesson of the rare earth example is you've gotta balance these things. Right? It can't be all of one or all of the other.
01:38:30 · John Coogan, Josh Zoffer, Jordi Hays
[01:38:30] S0: Last question. You, you wrote this in the Financial Times. It clearly will land with a geopolitically minded audience. But while I agree with a lot of what you said, I can't imagine it translating very well to everyday American voters who see AI as very deeply unpopular. How are you thinking about the the jumps and the translation layer that needs to happen to actually get, you know, American voters on board with this plan? [01:39:02] S7: Yeah. I mean, look. To me, first of all, completely agree. I think there there is a real opposition to data centers in in part because, you know, yes, they're they're big and they can be kinda ugly. But the reality is I my view is a lot of that has to do with the fact that they are the physical representation of Yes. Of AI. Right? Yeah. Yeah. [01:39:19] S0: If you look at the Instagram comments, like, you can very clearly tell that the vast majority of people that are, like, commenting on, like, data center bad content have never seen a data center. Like, they're nowhere near one. They might live in a major city that they don't even see them. But [01:39:34] S1: Yeah. You could see data centers putting up signs that say no AI served here. [01:39:37] S0: Yeah. But they don't like the idea of they're losing a job or losing out financially or watching companies get even bigger than they already are, all of that. So, yes, it's a referendum. But then how do you bridge that gap?
01:39:50 · Josh Zoffer
[01:39:50] S7: Yeah. I I think it's really, a few things. One is when a data center is built, right, making sure that, you know, if we have policies in place that, you know, push data center developers to build their own clean generation and ultimately connect it to the grid, that having a data center could mean your energy costs go down. Right? So that's point one is make the localized costs go away, and that's something that's that's very achievable. Yep. Point two would be, you know, I think when we talk about, what the effects of AI on the labor market are going be, I think taking a more reasoned view than just saying, oh, it's going to wipe out 50% of all white collar jobs by some date that's not that far in the future. I think the evidence is pretty and the jury is still out, and we have a lot of chances to use public policy tools to make sure that that's not the case and to tell a sort of a more balanced story and ensure that the American people know that it is not going to be job apocalypse and that it's going to be the kind of transition that other large technology transitions have been where, over time, the composition of the labor force changes, but ultimately, it's a source of growth and a source of new job creation. And then the last one, and this is really the point I'm trying to get across in this article, is there are lots of parts of a data center from the steel to the transformers that need to be made somewhere in a factory, and somebody's gonna be working in those factories. And there is a real opportunity to make sure that those jobs are American jobs. And that's something that I would also, you know, be in favor of using public policy to encourage, not just because, you know, those jobs themselves are valuable, but also because the industrial capacity they represent