Jasmine Sun on What the AI Industry Got Wrong About the Public Backlash

Odd Lots - Bloomberg

In this episode of Odd Lots, hosts Joe Wiesenthal and Tracy Alloway discuss the physical and political challenges behind AI's rapid expansion, focusing on the backlash against data center construction. They

Key takeaways

• AI's growth is constrained not by capital but by physical limitations: power, land, water, and regulatory approval.
• Data centers have become a bipartisan political issue, with opposition emerging from both left and right due to environmental, aesthetic, and economic concerns.

Main topics

• Physical constraints of AI infrastructure
• Data center opposition in the U.S.

Notable quotes

It feels very much like data centers are the physical manifestation of a new technology that is in some ways very hard to sell to America.

Conclusion

While AI is often discussed as an abstract digital force, its real-world expansion depends on massive physical infrastructure—data

Transcript preview

Speaker 5 (0:00) Get the latest news from around the world delivered to you in an instant. Subscribe to Bloomberg News Now and get news when you want it on your schedule. Speaker 4 (0:08) These are short audio reports on the day's top stories that you can listen to in just a few minutes. Speaker 5 (0:13) Make Bloomberg News Now your trusted source for the latest global headlines with context. It's available 24 hours a day anytime you need it right on your smartphone or smart speakers. Speaker 4 (0:24) Subscribe to Bloomberg News Now today on Apple Podcasts, Speaker 5 (0:27) Spotify or anywhere. you listen. Speaker 1 (0:33) Hello, Odd Lodge listeners. I'm Joe Wiesenthal. And Speaker 6 (0:36) I'm Tracy Alloway. Speaker 1 (0:37) We're the hosts of the Odd Lodge podcast, and we've got something exciting for you. Speaker 6 (0:41) That's right. So one of the best parts of hosting our podcast is we get to actually meet and interact with our listeners. And we know we have some listeners over in Los Angeles. Speaker 1 (0:51) That's right. So if you're in LA, we're going to be recording a live show, some live recordings at the Vermont Theater in Hollywood on September 17th. We Speaker 6 (1:00) have some really exciting guests lined up. have some really great conversations planned. So go ahead and get your tickets. You can find those over at Bloomberg.com forward slash oddlots or click the link below in the show notes and come and say hi when you're there. Bloomberg Speaker 7 (1:19) Audio Studios, podcasts, Speaker 5 (1:22) radio, news. Speaker 6 (1:35) Hello and welcome to another episode of the Odd Thoughts Podcast. I'm Tracey Alloway. Speaker 1 (1:39) And I'm Joe Weisenthal. Speaker 6 (1:40) Joe, there are a lot of interesting angles when it comes to AI. Speaker 1 (1:44) Yeah, you think so? Speaker 6 (1:45) Yeah, just to put it mildly. But one of the most interesting things, I think, is that we think about it as this like ephemeral digital technology. You know, in some sense, it's like the SaaS of yours. That's right. There's a window that you open on your browser and you type stuff into the internet and the internet. spit stuff out at you, hopefully not literally. But there is also this very physical component to AI because it involves huge amounts of power, huge amounts of things like memory chips, and of course, put those two things together, huge numbers of data centers. Speaker 1 (2:20) No, this has been like what is sort of fundamentally different about this, at least from a markets perspective. Yeah. Right. Which is that when we had the SAS boom of the prior decade, et cetera, like by and large, the successful companies, they didn't need to go into debt to build data. They were building. In someone's Speaker 6 (2:39) garage or whatever. Yeah. They were Speaker 1 (2:40) like building data centers, but like they were not so. Speaker 1 (2:45) computationally, they were not so massive that it was like they had to borrow money. They funded them from free cash flow quite easily. The company's building on top of that. They just plug into there. There was always availability on an Amazon cloud. There was not like some compute scarcity. You're not like using Slack one day and then it's like, oh, your four hour window is up. It will reset in a few hours. Like, yes, we are physical. constraints to the air build out, which basically in the history of the Internet is totally novel. Speaker 6 (3:15) Right. And I think that's really emerging as in many ways the prime constraint, because it feels like investors still are very happy to more or less fling money at the hyperscalers, at data center projects, at all of those things. There's no Speaker 1 (3:29) capital shortage. Right. There's Speaker 6 (3:30) no capital shortage. What there is is a shortage of, again, power. land, regulatory and political approvals for actually building these things. Speaker 1 (3:40) Totally. And, you know, this is like such a big country that it's like I always think like data centers are great. Just like put them out in the middle of nowhere. And then I'm like, oh, that's literally nimbyism. Right. It's like that's literally what everybody has that thought. Oh, there must just be somewhere out in the middle of nowhere to put it. But, A, it's going to be in theory near somewhere. And if we could just solve all of our issues by putting everything in the middle of nowhere, there wouldn't be an issue. There are reasons why they're located where they are. They have to be ideally near some sort of power infrastructure, some water infrastructure. These are things that also attract people. Speaker 6 (4:17) It's not hard to imagine that the middle of nowhere might not have a lot of energy and water infrastructure. Right. Exactly. Otherwise, there'd Speaker 1 (4:24) be a bunch of people there. Speaker 6 (4:25) Right. The other interesting thing about data centers and AI is that it's become one of the very few. bipartisan issues, I think, in America. And you have people both on the left and the right now who are saying that they don't like data centers. They don't want the data centers in their backyard. We interviewed Kathy Hochul, the New York governor, about her moratorium on building new data centers. So there seems to be a lot of, let's say, feelings. when it comes to data centers. Speaker 1 (4:51) And, you know, it did not get as much attention as Kathy Hochul, I imagine, because she's a Democrat. But Texas Governor Greg Abbott, just in the last few days, also announcing curbs on the state. People were a little quieter about that one. I have some reasons why. Maybe it didn't map to people's view of Texas as a state that's just open for business in a way that big blue states aren't. But as you say, it's bipartisan. And then I would say, like, one other thing. that I think is interesting about it is like, so we know the data center controversy, but AI in general is sort of like riven with anxiety on multiple levels. Obviously, for years, people have been talking about, you know, mass job laws. Lately, we've been talking on the podcast about, you know, what people call like X risk and stuff like that, like truly rogue AI. And in my mind. I feel like the people inside the labs are very anxious about those other two. These sort of like effects on society and again, literal like safety issues. Are they going to be responsible if everyone loses a job, et cetera? But on the ground, it does not feel like those things are the fears, even though I'm, you know. anxious, like that are really like driving the backlash politics. If Speaker 6 (6:05) I could sum it all up, it feels very much like data centers are the physical manifestation of a new technology that is in some ways very hard to sell to America, right? Like here's this thing that is not only going to take your job, but like maybe it'll kill you too. Well, I Speaker 1 (6:19) was going to say you don't like data centers because you're worried about the noise complaints. I don't like data centers because I'm worried they're going to kill me. We are not the same. We are not the same. All Speaker 6 (6:29) right. Well, that is the perfect way to put it. And we have the perfect guest to try to talk through all of these various things. We're going to be speaking once again with Jasmine Sun. She is an independent writer covering AI and Silicon Valley culture. And she just published an amazing piece of reporting called No Data Centers in My Backyard on her sub stack. So Jasmine, thank you so much for coming back on OddLots. Speaker 7 (6:53) I'm so excited to be here again. Speaker 6 (6:55) So congrats on the piece. We were saying earlier, it kind of sounds fun just to hop in a car with a few friends and research assistants and go on a road trip through middle America looking at data center sites. Speaker 7 (7:06) Yeah, I mean, it was fun. I was trying to. decide where in the country I wanted to go to do this data center road trip. I remember this was the last question I asked Claude Fable before he was taken away from me for several days. And so it was either going to be Michigan, Wisconsin, which is where I ended up going, or it was going to be Texas, or maybe the southeast like Memphis, etc. And to be entirely honest, part of the reason I ended up doing Michigan, Wisconsin, the upper Midwest is because it sounded a lot more pleasant to spend July by the Great Lakes than it did driving across the Texas desert. I'm sorry, Greg Abbott. Speaker 1 (7:39) No, I think that's a totally good reason. The upper Midwest is absolutely lovely this time of year. It's beautiful and there's lakes and all of that stuff. When we think about data center activity, I do tend to think more, I guess still in my head, northern Virginia, though there can't be that much space left out there. I do think a fair amount about Texas in part because of sort of like the Elonverse, et cetera, and how much. activity is happening there. I probably do think a little bit about the southeast. How active, though, is that area? Like how much interest is there in Wisconsin, Michigan in general about data centers? And what is it about those locations that, OK, it's like these could be attractive geographies for them, given what they need? Speaker 7 (8:27) Yeah, I mean, so Michigan and Wisconsin sort of rank near the middle to the upper middle of the pack when you look at how many data centers are being built across the 50 states. I think each of them have maybe around like 50, maybe 200 projects or proposed projects in the state. So it's not an incredible amount compared to, again, Virginia and Texas, where you have 600, 800 active data centers or data centers being built. The thing that drew me to Michigan and Wisconsin, one is that I had heard that proposals for new data centers were ramping up. And that's because one, Wisconsin, for example, has a sales tax exemption on the chips and on some construction costs. So that made it a somewhat attractive state. Gretchen Whitmer, governor of Michigan, who has always taken a pretty pro-growth standpoint for Michigan, how do we bring business back, was also sort of actively courting data centers. And so she did a very controversial photo op with Sam Altman at the Stargate site in Saline, which has become one of the big data center fights in Michigan. And so there are both things that these two states did in order to... explicitly attract data center investment. I think there also was enough, like you said, grid access and water access to make these reasonable places to build data centers, like an hour out of the cities, an hour out of Detroit, an hour out of Milwaukee, for example. And then the other thing that was really interesting to me personally as a reporter was I wanted to know how the data center story intersected with... A, the history of deindustrialization. I think this is part of why a lot of the politics in those states is about bringing business and bringing industry back, bringing the skilled trades back. Even in some of these very specific sites where there are data centers being proposed, they were like, you know. the site of a former GM plant in Janesville that sort of devastated the community when thousands of jobs disappeared during the financial crisis. Do we now want to build a data center there? And will that be the way to revive the community is a question that people are asking. Or, you know, in Mount Pleasant with the Foxconn site, there was a huge fiasco where Trump said they were bringing all these manufacturing jobs back. They weren't built. So the deindustrialization history was really interesting. And then the second thing was the primaries, which I think have now both come to pass. But I thought it was going to be a really interesting place to look at, as you mentioned, the bipartisan politics of data centers, the fact that in both states, you were seeing roughly 70-30 opposition to data centers from both Democrats and Republicans in a very, very polarized battleground states where Dems and Republicans rarely agree on anything. And then AI was also showing up in the Wisconsin governor's race where Francesca Hong was running on a data center moratorium. or in the Michigan Senate race, where Abdul Sayyed was talking a lot about data centers as sort of one of the examples of money infecting politics. Speaker 6 (11:06) Yeah, so we're recording this on August 12th, and Francesca Hong just lost the Democratic primary in Wisconsin to David Crowley. Which is kind of interesting because one of the people I remember you spoke to on the ground was someone who was saying she explicitly was going to vote for Hong because of her policies around data centers. And I know one primary does not a trend make, but do you get the sense that data centers can be like a winning political platform? Speaker 7 (11:35) Yeah, this is one of the things I was trying to get a handle on because I was wondering, like, will the local or state politics be a bit of a microcosm or a bellwether for the national politics of AI? And my sort of summary here is that I think that if you live in a community where a data center is being built or proposed, you don't like it and it's very high salience. It's very important to you. Like the fact that there is a gigantic construction project that will bring in traffic and noise and whatever for two to six years nearby. That really, really bothers people in the community. When I went to Port Washington, for example, which is another city in Wisconsin, and just talked to random people at Fishgiving, which was this like funny fish festival that we went to in order to talk to the locals. I Speaker 6 (12:16) had Speaker 7 (12:16) never Speaker 6 (12:16) heard about it before. Yeah, we both want to go to Fishgiving. I kind of want to go, even though I don't like fish. Speaker 7 (12:21) No, it was a real event. It was amazing. There's an Elvis impersonator. Obviously, it's Fishgiving. Speaker 1 (12:26) America is so good. Speaker 7 (12:28) That theme was life, liberty, and the pursuit of fish. It was amazing. It was on Lake Michigan. But every single person we talked to there had an opinion on the data center. Because when there is a gigantic, I think 700 acre facility being built near you, you will care about it. And so salience is super high if you live nearby. But then if you don't live near a data center, like if you're in a zip code where there's nothing being proposed, the polls still show that 70-30, you're roughly against data centers. I talked to Charles Franklin at Marquette who had done this explicit polling. But the salience is much lower. Data centers does not rank in like the top five issues electorally unless you were literally living in one of those