Taylor Lorenz’s Power User - Inside the Towns Saying Hell No to AI Data Centers w/ Jasmine Sun

Episode Date: September 2, 2026

Small towns across America are revolting against massive AI data centers. SUPPORT MY WORK: Buy a paid subscription to my newsletter at https://www.usermag.co         Support my work on Patreon: http://patreon.com/taylorlorenz To listen to the full episode and get access to more bonus episodes, an ad-free listening experience, and my weekly newsletter, subscribe to my Patreon or Substack 👇🏻Buy a paid subscription to my newsletter at ⁠https://www.usermag.co⁠         Support my work on Patreon: ⁠http://patreon.com/taylorlorenz⁠ In this episode of Power User, journalist Jasmine Sun breaks down her on-the-ground reporting across Wisconsin, Michigan, and New York to uncover why Republicans and Democrats are joining forces to halt Big Tech’s infrastructure expansion. From secret NDAs signed by local officials to skyrocketing electricity bills, strained power grids, and broken economic promises, we examine how the demands of generative AI are affecting local communities. We also dive into the real environmental impacts (from water consumption myths to the real problems related to natural gas) and discuss how the current compute buildout is creating a speculative bubble. Jasmine gets into how distrust of Big Tech is shaping the AI backlash, why previous development failures like Foxconn still influence Midwestern communities, and what the United States can learn from China's approach to AI infrastructure.

I think it's a pretty big deal that Texas and Greg Abbott even did a narrow version of a data center pause because there's no more pro-billed state than Texas. So if they are regulating, you know something's gone super, super wrong. All across America, communities are revolting against the massive, hyperscale data centers powering the AI boom. The backlash has become one of the rare political causes that unites both Republicans and Democrats. But what is it really like on the front lines of these fights? What are people in Silicon Valley still not understanding about this anti-Data center backlash? And what might these activists be missing about the way Silicon Valley thinks? Jasmine Sun is an independent journalist who spent 10 days reporting directly on data center

buildouts and campaigns in Wisconsin, Michigan, and New York. Today, I'm so excited that she's joining me to reveal what she found and what she thinks anyone who cares about the future of technology, infrastructure, and building a more equitable world should know about this fight. Hi, Jasmine. Welcome to Power User. Thanks so much for having me. So you went on this trip to see these data centers firsthand and report on the backlash. I have not seen one of these massive hyperscale data centers yet in the wild. Can you describe what these things are like? Yeah. Oh my gosh. So I did 10 days in Michigan and Wisconsin and I visited four sites that either had data centers that were proposed under construction or already running.

And a little bit ago, I also went to Northern Virginia to see some of the data centers there. And the first thing I'll say is that data centers are, in fact, extremely ugly. Like, they are, like, gigantic gray boxes. They have no windows. They're very flat. Like, you don't build them up. They're like one-story buildings, maybe two-story buildings. And they kind of just feel like they go on forever.

Like, you might be in a fairly rural area with a lot of, like, forests or cornfields or whatever. And all of a sudden, you'll start to see a bunch of power lines pop up. And it's like, ooh, something's going on. And then you just will drive past a stretch of like what feels like one giant infinite building. And that is the data center. And so in places like Port Washington, which is I think like a 700 acre site, people would time how long it takes to drive past the data center on the highway going 70 miles per hour. And it takes like a minute and a half, which is pretty crazy. Why do they build them horizontally and not vertically?

I feel like these things, maybe if they were towers, I don't know if that would be worse or better. but it does seem like they're taking up so much land. I mean, I think it's just because inside of these big boxes, it's a lot more technically complex than like an Amazon warehouse. You have these really complex electrical systems, these heavy, heavy racks of chips and servers. And so it's just the literal weight and complexity of it means that people haven't figured out how to build data centers vertically yet.

I mean, I kind of hope there's like some research going on just because I feel like that might be better to like make them smaller. But at least for right now, it's just the electrical systems inside are too fragile to be able to build them upboards. I feel like we constantly hear about the noise of these data centers. What did they sound like? If you stand like across the street from a data center, it's kind of like there's a very loud AC or fan that's near you. So it's this kind of like white noise, humming, whirring, whirring. Every once in a while you'll get like a rattle or something.

It sounds quite mechanical and inhuman, and it is pretty incessant. Like, it doesn't really go away. Imagine hearing loud hums or white noise like this right here. I would say if you're like a block or two out, you're not going to hear it anymore. So it's not like going to ruin the whole neighborhood. But if you happen to be one of the few houses that is directly across the street from a data center or you're walking past, you are going to hear it. And it's not a super pleasant sound.

Yeah, I feel like it's the type of sound that could maybe drive you crazy after a while. Well, where were these data centers being built? I mean, you mentioned Wisconsin kind of out in rural areas, but what were the communities like around these data centers? Yeah, so what's interesting is data center citing is complicated where the first thing that a data center developer does when they're looking for a place to build is they actually look for electrical capacity. And so because it requires so much power and so much electricity, one thing people ask is,

like, why don't they put these things truly in the middle of nowhere? And the answer is because they need this power infrastructure. They do have to be close enough to an urban center or a suburban center that you can do the power connection. So even some of the more rural areas we went to were maybe an hour out of Milwaukee or an hour out of Ann Arbor, Detroit. And so these are mostly in either rural or sort of suburban areas or like smallish towns thinking like anywhere from I think 3,000 to like 60,000 people approximately. Some of them were built on areas that were already industrially zoned. So like in Janesville, which the data center there actually got pause, but basically there's this 250-acre lot in the middle of town that used to be an old GM factory that was there for like 100 years. And it was in the middle of town because there's so many workers who would like commute to and from the factory that shut down in 2008 during the financial crisis.

And it's kind of been laying dormant and blighted until a data center developer showed up and offered like, hey, we can do some cleanup and we'll come in. And part of the controversy with that site is it's super close to downtown. So I do think it would be a little bit more invasive or in people's way compared to some of the data centers that are in rural areas. But it was an industrially zoned area. In other cases, they're actually taking agricultural lands like cornfields, farmland, whatever. And the data center developers are usually buying it out at multiple times market values, like millions, tens of millions of dollars even from farmers in order to then build a data center on it. So in Port Washington, driving from the city's like downtown area to the data centers took about 15 minutes. It was there were not a lot of houses around.

It was mostly farmland. So in these communities, there's a lot of activism. You mentioned the Janesville project, which I feel like is sort of a centerpiece of your reporting. And you talk about the kind of activist communities that have mobilized to block projects like that one. What types of people make up this anti-data center coalition? One of the most interesting things about the data center fights that really came across to me, even before I went on the trip, is that they are unusually bipartisan. And they do cross a lot of typical divides that you would see in politics.

So in Wisconsin, for example, you see roughly 70, 30 opposition to data centers from both Democrats and Republicans, which is super, super unusual in like a really polarized state where Democrats and Republicans rarely agree on anything. So like Charles Franklin, the pollster explained to me that say with solar farms, which are actually also quite controversial and got a lot of opposition from rural conservatives because of land use. Democrats, meanwhile, were like, oh, we love renewable energy. So it's become kind of a 50-50 partisan issue. But data centers, everyone right now kind of hates them. There's like no pro data center faction going on. Nobody wants them is sort of the line I kept hearing. And so the people I met, I would say there are somewhat more Democrats than Republicans,

but we saw everyone from stay-at-home moms to like a retired banking executive who got into data center activism to veterans, to some retired folks, to some more professional activist types. But I think the main thing is that I had this like very memorable conversation with three women in Celine in Michigan, this town of, 3,000-ish people. These women I talked to, they were liberal Democrats, but Celine itself is very polarized. It's a very purple town. And they were telling me that data centers were the first issue in 10 years since the Trump election that they had been able to productively talk about

with their Trump voting neighbors. And so they almost felt a sense of political optimism, actually, in the agency in crossing party divides that they never, ever thought they could cross and working hand in hand with their conservative neighbors in order to oppose the data center. And Port Washington again, which was another of the towns I visited, that's a pretty conservative Republican area. I was seeing Tom Tiffany signs in all the front yards. He's a Republican candidate for governor or nominee at this point. And those folks were just as angry about the data center as the liberals who I talked to.

I mean, this is something I hear a lot as somebody that covers the quote unquote child safety movement as well, where I feel like anger at big tech or sort of like these tech oligarch billionaires is such a cross-party issue. You mentioned solar farms, and I've heard you say this before, of just like, there is no pro data center lobby. Like I've seen people make comparisons to nuclear energy as well. How do data centers kind of differ from those other industrial projects? And why aren't there more people, I guess, like outside of the AI industry that want these projects built? Yeah, I mean, this is a really interesting question to me because I was trying to understand I would see these polls that are like heat map news did some good pulling.

Like if you ask people, do you want a solar farm near? you, a nuclear plant near you, a transmission line near you, a coal plant near you. Data centers are literally more unpopular than I think everything except coal. So, like, people would rather have a nuclear plant in their backyard or a car factory or a lot of things that, frankly, pollute a lot more than a data center does compared to a data center. And so there must be something going on that's different. And one of the hypotheses that I had after going on the trip is that, like, with some of these other industrial developments, yes, maybe it's noisy and there's construction and it's

annoying. So there's some opposition. I'm not saying the other ones aren't controversial. In fact, Sodor had a lot of controversy in Michigan and Wisconsin, but there is some pro case for it. It's sort of like, well, we all need energy and we know that renewable energy might be a good way to achieve that. And so, yes, it kind of sucks if it's in your backyard, but like you got to take one for the team or like at least somewhere in my state I want it built. That's the classic NIMBY line. It's like, we want housing. We just want the housing somewhere else, right? Whereas with data centers, what's super interesting is that people are not like build the data center in the town over. they're like, don't build the data center, right?

The opposition to data centers in the polling holds, even if you are not in a zip code where a data center has been proposed or is being built. And so you see the same fury at it because I think the thing is that people don't really understand why they need AI. They get why they need energy, housing, even like factories producing cars, people kind of intuitively understand what that's doing. Whereas I think with data centers, it's like, okay, and it produces chatbots, which are like these widgets that are kind of funny, I guess, but not essential to my life.

or it's like coding agents for people in San Francisco, but that doesn't really help me in my life. And so people weren't like, oh my God, I hate AI. It's going to kill us all and take all their jobs. It was more just like, this thing doesn't seem that useful. Why do I in my community have to bear all of the material and physical costs for some industry that really isn't helping anybody except a handful of people who are already rich who live on the coasts? Yeah. I feel like there's this meme, too, of AI is being shoved down our throats.

We're being forced to use AI. And I'm curious if you heard any rhetoric like that. do these people engage with AI products at all? Because it's so interesting to look at this anti-data center backlash, which is so pervasive. And then also see things like, I know people are skeptical because Gemini has been integrated, but, you know, Gemini just hit a billion users. ChatGPT is one of the fastest growing, it is the fastest growing consumer product of all time. The consumer use case for this product seems to continue to be made as people keep downloading

these apps. So where is that disconnect? So I did ask most of the activists I met, like, do you use AI? like, what do you think about it? And I would say a couple things. So one is some people did use AI, but they were saying, I can make some memes with it or I can help write emails a bit better, but they're like, it's not worth it. Like, if you could promise me that there'd be no data centers in my backyard or anyone's backyard and also you took away chat CBT, they were kind of like, that's an okay trade to me. Like it's kind of helpful, but it's not essential to my life in the way that

maybe a car or energy or housing is essential to my life. With Uber, for example, which did a lot of political lobbying around labor laws, like one of their most effective lobbying tactics was Uber had actually baked itself into the necessities of like getting around so that their consumers were much more avid advocates of Uber. Like to them, it really felt like a cost to take Uber away in a way that I think people, unless you're like in software or something, don't really feel with chat GPT. The other thing I heard a lot about was like this kind of AI ambivalence and I do kind of get that right.

I use my phone and I'm happy my phone exists, but I think I have a complicated relationship to my phone like a lot of people do. A lot of people spend time on social media and they kind of regret it. And with AI, one thing that I would hear sometimes from people is I'm using AI, but I'm trying to use it less, right? It feels like a vice. Or I talked to a bunch of public school teachers in Madison and they were telling me that their students are actually very anti-AI and the younger students are more anti-AI. That doesn't mean that they never ever use it. Like if they're in crunch time and they have homework due, people will cheat on their homework by using chat GPT.

But like the kids don't feel good about it. and they would prefer not to, right? Or I think of those polls that I'm sure you've seen where people say, I would pay 20 bucks for TikTok or Instagram to be removed from everybody in my school's usage. But given that my friends are using it, I'm going to use it too. And so the thing I really felt was AI, some people are using it, but maybe it hasn't been helpful enough or like definitively positive enough and definitively useful enough that people are actually willing to advocate for it.

Yeah, I feel like people just still, I don't know, I was reading the Zuckerberg Manifesto on Super Intelligent. And it's like painting this vision of the world that sounds amazing and great, but it's like a lot of people are just struggling with pretty basic needs, especially in this country. And so I totally understand why it feels frivolous. Or I used to work at a call center after college and I am in touch with some friends that still work and basically those types of industries.

And AI is kind of a nuisance or it's being, you're basically training it to replace your job. So there's a lot of negative sentiment. You also get into some of the environmental concerns. And I feel like this is something that we also hear, constantly every I'm sure it's going to be in the comments of this video people are very frustrated rightly so about sort of the environmental effects of this so first let's talk about the water stuff because I feel like people are starting to realize that maybe it's a little overblown but you know how bad are these data centers for our water systems I would say that

water is not the primary environmental impact that I would be concerned about I do think that a lot of the use is overblown so just to give some stats a chat to be query like a normal question you ask is going to take, sure, a little more water than a Google search because Google searches use almost no water, but it uses way less water than a YouTube video. Or if you look at a data center as a whole, right? So both in Janesville and Celine, the proposed data centers are right next to golf courses. And I did some research to try to estimate the amount of water that the golf course would use versus the data center. And it's very similar. And still today, the total amount of golf courses in the U.S. use four times as much water as a total amount

of data centers. And a lot of this is actually the trend. Yes, there are more data centers, but The water usage is not a huge problem because all of the new data centers, for the most part, are using closed-loop systems, kind of like AC, where the water just recirculates rather than having to constantly, constantly draw out water. And the water is being taken from municipal water systems. The data centers are paying customers. And we're not really in this sort of freshwater crisis in the U.S. So I think so long as it continues to be true that we use closed-loop, non-evaporative cooling systems, water is not going to be a super big threat in terms of the environmental cost of data centers. I just want to say that if you're getting annoyed by the amount of ads in this episode, please, please, buy a paid subscription to my substack news that are at usermag.com or support me on Patreon via the link below. Supporters of both my substack and Patreon get full access to every episode and bonus episodes completely ad-free.

I have no long-term brand partnerships on this channel, so I rely on your support to keep making this show. Please buy a paid subscription below. And again, that will give you access to the ad-free RSS. Thanks so much. Now back to the show. What I hear constantly for people as well, and I was just listening to the news this morning where they were talking about this issue in LA of rising energy costs. I saw some chart on Twitter that I need to find, but it's just like this terrifying chart of how electricity costs have effectively doubled in the past 20 years. And their point was to say, like, you can't blame this all in AI because look at what's been happening for 20 years.

But still, given that rise, given the amount of power that these data centers use, it seems like people are rightfully concerned that their power ability. is going to go up, right? Yeah. Honestly, I'm a lot more worried about electricity, both as an environmental and an affordability concern than I am about water. And part of the reason I want to shoot down the water myth is because I'm not saying there are no environmental costs of these things. It's just we got to focus on the right problems, right?

And so the thing about electricity is these things do just require tons and tons of power. The chips and the servers inside the systems are all connected to each other. They're all talking to each other super fast or performing these super complex calculations. And I think it's something like 7% of all electricity in the US now is going to data centers. Some of the big Stargate centers that are being proposed, like the one in Michigan, it's supposed to be like a 25% increase just without one project over DTE, the monopoly electricity provider's normal load. And so if you are putting that much strain on the grid, if you are trying to add a 25% of your existing capacity,

you are going to need to build out new power plants. And that's going to be more demand on the system. And if nothing is done about it, rates could definitely go up because we're just not building the power plants and the gas plans fast enough in order to keep up. And so I think in that case, One is unless there are ratepayer protection loss, which you are seeing more of across the country, and I think that's a really good thing. There is a risk that normal residents end up holding the bag and paying more in electricity

because of the increased demand. I think that would be terrible. I don't think anybody should be paying any money for like meta or like Google's like electricity requirements. And these data center and AI companies should be covering at least all of the electricity costs, both of the buildout and then any potential rate hikes that you see from that. The other thing is, of course, then the environmental concerns, right? And so in the near term, I think Mark Zuckerberg has said that it's mainly going to be gas plants,

like natural gas that's required in order to serve the new demands on the electricity grid. And so that is obviously not the maximally environmentally friendly option. People say, like, we hope it's going to be more renewable in the future. We hope it's going to be more green. Some of the data centers are using some percentage renewable energy, but just solar and wind is a little less reliable. It's not like 24-7. And so that's one of the problems with doing that.

And so one of the areas that I feel like policy and regulation is really important is in setting these environmental standards and holding the companies accountable to like, you've got to make sure that both from an affordability and an environmental perspective, you are doing as much as you can to both cover the cost, but honestly to like even give back even more. Like you should be investing, frankly, in my opinion, extra money into maintaining the environment, into conservation, building out the green energy grid because you know that's going to be really important.

And so I do think that the power story is a really important one. Yeah, Molly Taft at Wired recently posted a piece about how the two biggest winners so far and the sort of the data center buildout race are Chevron and Williams and how this has just been such a boon for natural gas and AI technology itself is allowing natural gas companies to actually act more efficiently and grow their businesses, which just feels like it's furthering this natural gas infrastructure. I mean, you mentioned that other forms aren't as reliable. Do you see any world where we could be powering this data center boom with clean energy?

I mean, I hope so. And like, when you talk to the folks in the AI or data center industry, they say medium term, we do want more and more of it to be green. And like maybe there's even a world where the AI companies pay for the cost of greening the grid, right? Because that's something that everybody or like a lot of folks in the U.S. and already want to do. But it's such a big infrastructure investment that the economics don't really work out.

And so one hope is that maybe AI and data center companies can actually pay for the cost of doing, that both for self-interested reasons and we would also all welcome that. But I think because there's so much short-term demand, we are not going to get all that new renewable capacity in the next one year, and there's plenty of demand for AI in the next year. So the hope is that natural gas is just a short-term thing. But like, frankly, I think, again, I really hope that policy is here to like help keep those companies accountable to making sure that that's a short-term thing. And that's something that people are worried about too. Like I was talking to some activists in Celine, and they were saying, you know, the only thing worse than a data center in my backyard is a data center and a

small modular nuclear reactor at my backyard. I was like, honestly, yeah, fair, right? So I think that in a world where our grid is already strained in a world where people's electricity prices have already been rising year on year and reliability is frankly not improving that much along with these increased costs. I think we do have to be more careful and sensitive to people's concerns about I better not be left paying more on my electricity bill just because we didn't plan this right. One thing that I actually think is interesting that I saw is someone pointed me to a policy in Ohio where AEP, the utility provider there, they basically pulled the data centers. You are forecasting X amount of demand of how many gigawatts of electricity you need, but you

need to commit to paying for 80% of that no matter what. And that ended up being really important because their forecast was wrong. They were exaggerated or they just did it wrong and it turned out they didn't need as much electricity as they said they were going to. But the data center then had to be accountable for paying for most of it anyway. Because I think a big question here again is like, this is kind of a speculative industry. We don't know exactly how big the AI build out is going to be. And so it's really important that normal people don't get left holding the bag.

And if we can extract as many guarantees up front out of these AI companies to pay their own way, I think that's really important. Yeah. I mean, I feel like speaking of extracting benefits, which I think is great. And obviously, we all want these towns to have more negotiating power. You write about the use of NDAs and how a lot of debates around this stuff happen behind closed doors, not in the public view. And that seems to have really poison the discussions, too. I mean, people are rightfully angry. It's like these things show up. They don't have a say.

Why were companies using NDAs so heavily if they didn't have anything to hide? So one of the most controversial pieces that I feel like I really underestimated until I actually want to talk to people is the NDAs really did so much to amplify distrust and break down negotiations between people, their governments and these data center companies. Like even the most pro-Jas center people I talked to where like we regret the NDAs. It was a huge mistake. It lost so much trust with people because I think the NDAs basically are a dark money and politics story of like some big trillion dollar company is coming to town, paying off your local officials, not only not inviting public input, but actually banning the public from knowing how the negotiations are going. And then

you only tell people once the deal is already done. It's again, it's AI being shoved down our throats. You have no choice in the matter. It feels very, very anti-democratic, right? And so I think the reason that the NDAs were originally being used is one, there's this process where the data centers are deciding what town to go to. And so they don't want people to know about it before they've committed to a site. But also, frankly, in a kind of self-interested way, it was sometimes about extracting the best deal from these communities. Because two years ago, there was like no data center backlash. And so it was almost this process where the data center developers would go to a small town and be like, what tax subsidies are you going to give us?

You should compete to want our data center over in your town. And so actually the local officials were kind of worried, like, oh, shoot, if we don't give them a good deal, maybe they're going to go to the town over and bring the tax revenue over there. And by handing NDAs out to all the different cities, they couldn't really coordinate with each other. And there's this sort of competitive dynamic that was going on that really put localities at a disadvantage. Or the NDAs were also used to prevent farmland prices from going up. So in contexts where the data centers are like buying up tons and tons of farmland, they're already paying quite a bit of money, like well above market rates. But if people know it's specific,

specifically for a data center. Everyone knows how much money the AI companies have. So people are going to charge way, way, way more. And so again, data center companies were trying to save money and save hassle by forcing all the city officials to sign NDAs. But that actually just ended up really burning people's trust and making them feel like something shady is going on. This is not a fair negotiation. It's extremely asymmetric. And so now you actually have a world where like Microsoft, for example, I said we're not doing NDAs with local governments anymore. We regret that. I think other companies might follow suit. And I think that governments should not be. signing these. I think that it is not worth it. And like actually one thing I heard from local

officials was it meant that we couldn't get ahead of the story. So when rumors started spreading about someone coming to town or is this going to happen, is that going to happen? Local officials couldn't even talk about it. They couldn't even get ahead of the social media gossip because they had signed this NDA and then they couldn't educate and communicate with people in their community and have the actual public conversation about it. And so I think the NDAs were a huge, huge mistake here. Yeah. It seems like there's just anger in not being able to have these conversations. see these videos, right, of somebody getting dragged out of a city council meeting. And that just goes viral, right? And it fuels it and fuels it. I feel like, you know, when I talk to people

in the AI industry, and I'm sure you hear this too, or people that are sort of building these developments out, they say, well, you write about this in your piece like, oh, we try to correct the misinformation, but these people won't believe us. And no matter what we say, they don't believe us. I think this is something that you encountered as well, right? People just saying, I don't believe them. So people have really been burned by these corporate promises before. And that's why they would say, I don't believe them with regards to the data centers. And in Wisconsin, there was no example I heard more about than the Foxcon deal. So what happened there was in 2017, Trump during his first term, was like, I'm going to bring back high-tech manufacturing back to the U.S. away from China.

And Foxcon, this like big Taiwanese electronics giant makes a deal with Mount Pleasant, a city of about 28,000 people to open up a big factory and create, I think, 13,000 jobs was what they originally said. Trump announced it nationally. There was an NDA. Initially, it's supposed to be this, like, exciting thing about reindustrializing the Midwest. But they messed up the deal terms. Foxcon pulled out and after the city and county took out over $100 million in loans to pay for this infrastructure to build out roads, build out electricity, to eminent domain a bunch of land, Fosson was like, never mind, labor costs are too high. We're going to make 1,000 jobs instead. They just ditched. They're still like paying back this gigantic debt requirement. And everyone in the area is just like, how could we have been so stupid as to not do our due diligence, not to

write a better contract? We are never going to believe some tech company that promises stuff ever again, right? And so I think that when AI companies say, oh, but our data center is going to pay so much in property taxes or we're going to create this amount of jobs or the environmental impacts are not going to be that bad, people just look at them and say, we have been screwed over before. Why should we believe you this time? And so the data centers are showing up in this environment of extremely low trust among people who feel like they have been screwed over before. The GM plant in Janesville again, when GM went bankrupt, they not only laid off all the workers

at the plant, but they left $30 million in contamination in the ground there. And they never cleaned it up because of an EPA loophole. And so on one hand, you would be like, oh, let's replace GM and put a data center there. On the other hand, the reaction of a lot of people in Janesville is, dude, we know what it's like when we invite a big employer to town, their business goes bust and they just leave us to deal with all of this environmental ways and they never take accountability for it. And that risk is something that I think a lot of people are worried about with the data centers too. Yeah. I feel like people don't trust, I mean, the tech industry for so many reasons,

but there's also this idea that it's a bubble, right? And that this technology, they've seen it before. AI is quite frequently compared to like crypto or, you know, NFT type things. Like Silicon Valley does engage in this hype around kind of seemingly every new technology. And so I guess like are those concerns validated? Because we have seen this bubble before and the tech bubbles generally always burst. And so isn't it smart for these people to kind of be concerned or seek to slow things down? It was really interesting to see how much that showed up in my conversations with folks is like, even if there's only like a 20% risk that AI is a bubble, that's still a 20% risk that like this

gigantic ugly thing is going to be left in your community. and you're never going to see any of the tax benefits. And so I think the perception that AI is a speculative in high-risk industry, and we don't know what's going to happen in four years. Are they still going to need all of this compute? Is a big question. And it's so interesting because, again, in San Francisco, when I talk to people in like startups or whatever,

they're like, we're so desperate for compute. The reason that the AI subscriptions are so expensive or Claude has all these limits is because we want so much more AI that we can even serve. And so that's why they're so desperate to build out these data centers. I think it's very hard to say, is it a bubble like so many economists are debating? this question and I'm certainly not that level of expert, but I think that A, people are right to want to secure the downside, right? If you're some local official or some person thinking about

whether you want a data center, you need to consider the risk. Maybe it's not the most likely outcome. I don't think AI is likely to be a bubble, but I certainly think that in the case that it is, you should have security against holding the bag in that instance. And that's where, again, I sort of come back to the power asymmetry between governments and corporations and also, are you able to extract these guarantees to make sure that you don't end up with that GM scenario where GM just like leaves the brown field there and is like, you guys clean up the environmental waste later. If they don't need as many data centers as they thought, which is possible or even like there's this one AI company and maybe the whole bubble doesn't pop, but one AI company says goes out of business and they're the

ones who built the data center. What's going to happen to that? And so those are sort of the questions that I think are worthwhile. I actually became way more open to say Kathy Hochel's New York moratorium after doing this trip. I think at the beginning, I thought that it was kind of blunt. And I'm pro-AI regulation, but I wasn't sure that data centers were the right place to regulate AI. But the thing that actually made me much more amenable to the moratorium, if it's done right, is I was like, yeah, I actually think it's worthwhile to figure out at the state level. What are all of the guarantees and regulations?

Environmental regulations, financial guarantees, all of these different downside risks. Like, it is government's job to secure normal people against any downside risk. And maybe that only needs to be figured out at the state level, which has more resources, negotiated and then towns can sort of borrow that set of frameworks instead of having these like part-time mayors who are getting paid $5,000 a year somehow negotiate with Open AI, which is like never going to be a fair negotiation. Although I do think they have a lot more leverage now, but I totally agree. I have been very skeptical of the moratoriums just because I'm not sure that they will ultimately accomplish. Hopefully they do. But I think like it's also different if you have

an indefinite moratorium versus like a year or whatever. And I think like it makes sense. I mean, when you talk to people too, especially in these. communities I feel like a lot of people and there was a proposed data center down here in Southern California recently it ultimately didn't go through but they were saying I just wish you could slow down like I just don't have time you know I'm trying to learn about this stuff that you know we've seen Bernie Sanders now call for this AI pause open AI is saying they're like slowing down development on I think one of their models because of cybersecurity concerns on one hand it seems like even the AI industry themselves want to kind of slow down but then they also are sort of like

rushing to build these data centers is it maybe good that these more or tombs are in place, well, that sort of forced this slowdown that they claim that they kind of want. Yeah, it's so weird, right? Because in no other industry, I feel like you have this dynamic where even the people who are working at the companies are like, we wish it would all go slower. But then they're like working like 100 hour weeks trying to like build AI anyway, which I find so strange. Because like it is true that if you ask anthropic or open AI employees, raise your hand if you wish that AI didn't pause, but just like went a little slower. I think everyone would raise your hand or the majority of people would raise their hand.

Because again, like you said, people feel like there's just not enough time. But I think because of the very competitive dynamics where it is still a competitive race between Anthropic and Open AI and then somewhat meta and Google and whatever, people are so worried that if only one company slows down, we're going to lose. And so the capitalist dynamics of competition mean that the race surges ahead, even if every individual person is kind of like, I wish not that it paused, but that it would go a little bit slower. I think the speed is also what is making people angrier. That is why they are not having these public discussions. That's why people aren't really getting educated before they have time to sign this contract in NDA right now.

One thing that really pissed people off in Celine Michigan is DTE, the electricity provider that, again, everyone already hated because they were kind of terrible and expensive and there were brownouts all the time. They actually demanded to skip all public hearings for their big Stargate Center because they were like, we need to start construction on this data center in early 2026. And this AI project is so essential that we don't want to have almost any public discussions. And so they just skipped a bunch of stuff and they successfully did that. And I was, you know, trying to compare data center approvals to say housing approvals, which have been slowed down an incredible amount by just infinite sort of public discussion and meetings. And data centers get approved much, much faster than housing projects.

I'm not sure that that's right. I really feel like in this case, maybe slightly more public input and like negotiation would help everyone involved. However, at the same time as for the question of, will the moratoriums actually slow down AI? I'm not sure. Maybe a little bit on the margins. Like, I think it's a pretty big deal that Texas and Greg Abbott even did a narrow version of a data center pause because there's no more pro-billed state than Texas.

So if they are regulating, you know something's gone super, super wrong. I do think people are much more aware that there's a risk with domestic politics. At the same time, these AI companies, they have backup plans. They have already started building data centers in Canada, in Australia, in Europe, they're starting to talk about to in space, I guess, is Elon's big thing. And it's literally partly because they're like, well, no one's there to stop us. in space and that technology is very, very speculative. But I will say that the AI companies are so desperate to build out their compute that they will find someone willing to say yes, they will make the checks big enough that eventually some people will. And I don't know if it's going to be

stopped just because we do a moratorium. Do you wish you could just hit skip on the worst parts of your life? You know the same way you can skip an ad? I get it. I'm Siaia and I live in Ice Cove. I've made some questionable decisions that didn't end up the way I planned. And today, I'm still figuring it out. Somehow, things usually get worse before they get better. Apparently, that's how I roll. So bundle up and come along for the bumpy ride. Stream a new episode of North of North Tuesdays on CBC Gem.

I have been seeing a bunch more of these TikToks, Instagram reels, of people on the left talking about China's AI landscape and kind of how why China's doing AI not in the evil way. And China can build out AI, but it works for the people. And it's been so interesting to me because one big thing they talk about is like this idea of clean energy. China is maybe has some sort of technology where they're like harvesting seawater and using it in two plants. I know I was kind of reading about it. I don't think I've seen that one.

I have not heard of that either. But China is undeniably relying a lot more on clean energy. But they also don't seem to have as many data centers. This is a very tech optimized interconnected country that I think a lot of people look at and they're like, wow, this is like the meme of the future if, you know, like they see. They see these amazing cityscapes of like Chongqing or whatever. And it looks incredible. And they're like, well, hey, they have this futuristic utopia where everyone is housing theoretically.

But they don't have all the data centers. So how are they doing it? And why can't we do that? That's so interesting. So like on one hand, yeah, China is ahead on clean energy for sure. And they are better at batteries. They're better at solar. China's better electric vehicle.

So there are some dimensions of green technology that I do think the Chinese government has just made a way bigger investment in. And I think that is really commendable. I think the U.S. should learn from their electric vehicle investments, for example. At the same time, they don't have nearly as much compute as the U.S. and actually when you look at the AI industry and people who see this as a race, the U.S. is compute advantage. Having more data centers, having way more chips is sort of the main thing that people think

is going to make America win the AI race. It's not just about the models or on talent we're like kind of falling behind because there are way more scientists and engineers in China. But the compute factor, the fact that you need these gigantic data centers to run AI and China just hasn't built as many and they also don't have the chips because of expert controls. That is seen as like a big disadvantage for them. The other thing is China is building out tons and tons of data centers. It's not like the government does not know that they are falling behind in this capacity.

They are trying to build their own domestic chips ecosystem. They are smuggling chips from the U.S. to get access to the banned ones. And in places like Guajo province in the southwest or some places in the north, the Chinese government is building out tons of data centers. I think the difference is that as an authoritarian state, you're just not going to going to see the public backlash because there was no expectation or system ever that there's any way for normal people in China to say, I don't want a data center. When the Chinese government tells you to do something, when they eminent domains some land or they tell you we're building

a giant factory where your farm used to be, go move over there, people don't have a choice because, again, it is an authoritarian state. I do think that the same characteristics politically that allow China to say build bridges really fast or to build a lot of skyscrapers really fast, that also goes hand in hand with not having any system of public input, which means that people cannot really protest if data centers show up. So like China is investing in the data center build out, and I would not hold it up as sort of this utopia because that's even less voice than you see in the U.S. Yeah, I was getting into it in the comments of one of these videos because these are the same

people saying there's not enough public accountability, right? And then they're like, but China's doing it and there's no dissent. And it's like, please. It's like dissent is not allowed, actually. Like dissent is illegal. You cannot organize. an anti-data center protest in China, the cops will show up and pepper spray you and take you to jail if you do that. It is a single party state. Yes, there are some applications of AI in China that I think are pretty interesting. Like I think that a lot of the smart city stuff with climate and like green buildings and

things like that, China's certainly investing in those applications of machine learning. They also have the world's most sophisticated surveillance system, right? And a lot of the reason that China has been ahead in computer vision AI for a very long time is because they have built out this gigantic surveillance network, both of cities and the internet and everything else. And so not all the applications of AI and China are ones that I think that the US should be modeling. I completely agree. You went to this rally for Abdul Al-Sayed, who's now won the Democratic primary in Michigan, where Bernie and AOC, I think, was there too. And you mentioned kind of how they've used data centers as this rallying cry and it's becoming this political message. This is something that

the Democrats have sought to really lean into, especially how do you see this like AI popular? populism reshaping the electorate, what candidates get elected, etc. It's also kind of interesting too because I think Francesca Hong was sort of seen as the anti-data center candidate and she just lost. So I don't know if it's as clear cut as we might think. I mean, I think there are a lot of things going on in both races besides AI and data centers. So like it was not the central central point of either race that both of the candidates did talk a lot about AI.

I also went to a Francesca Hong event and to be honest, like I mentioned this in like a footnote or something, she's just much less charismatic than Abdul Sayyad and I thought she was just less sharp. seem less knowledgeable in a lot of these issues. So I was not super surprised when she lost and I think she had other vulnerabilities and related to data centers. But the thing that I took away from the politics of AI in both Michigan and Wisconsin, it's less that people uniquely care about AI.

Like some people do, but that's still a little bit niche unless you live literally in the neighborhood where a data center is being built, in which case you do really hate it and it really matters. But most people, they're like, I don't like it, but it's not my top five issue. What AI does do is it adds fuel to the fire of really fiery anti- elite sentiments. Like I think, you know, the Abdul-Syad rally I went to, it's kind of an extension of the Bernie AOC oligarchy tour, right? The story there is we want to get money out of politics. Your politicians are not working for you. They're working for all these corporations and

billionaires. The chant that I really recall being quite powerful from Abdul's rally speech was when DTE buys a Senate seat, they are buying the ability to keep you quiet while they like raise your rates and take out your power during a storm. When Blue Cross buys a Senate seat, they're buying the ability to make pharmaceuticals expensive more and more expensive. And when big tech buys a Senate seat, they're buying the ability to put a data center in your backyard. So the AI and data center story was much more like fuel on the fire of your politicians today are not really working for you. They're actually working for these big corporations.

And that, again, it's like very similar to what I would then see in these local fights where it's like, yeah, your local officials signed in an EA. They made a deal without asking you, promising your land and your community's investment. to support this thing and people really have this sense that these people are making these backroom deals without consulting us. And so that's kind of how I expect AI to mainly show up in the election. The exception being if we start to see like actual job loss at a really high level, these kinds of more material impacts, then I do think that AI will become like a more central issue.

It's like if AI intersects with the things that people really care about, affordability, their jobs, you know, their lives, then AI will become a big issue. AI in and of itself as like, oh, will there be like cyber risk or something? I think it's something people care about, but it's not going to be a top five voting issue for them. I feel like your piece was so good because you really explained. I think to a lot of people in Silicon Valley kind of where this opposition is coming from and like really got into why people feel the way they feel and it's not all just maybe

Facebook misinformation or a Chinese sciop or whatever they claim. I'm curious spending so much time with these activists and also being in Silicon Valley and embedded in this culture yourself. What are things that you think these activists are maybe misunderstanding about the Silicon Valley world. There's so much distrust and like lack of understanding on both sides. I do think that these AI companies are willing to pay a lot of money and provide a lot in community benefits in order to build a data center. And I think it is not for every community worth making that deal.

And you do have to be really careful to make sure you're not providing all these carveouts. But like, I think that some of the activists are underestimating the scale of economic and community benefits they could get. if they approach the conversation in a little bit better faith. Like, again, in Louisiana, Medi got 50K bonuses for public school teachers. And that is life-changing money. Like, that's pretty amazing. And these companies are willing to pay for a new park, for money for the schools, for whatever.

And so, like, if you are able to, again, negotiate a good deal where you do the right payer protection, you use the least environmentally costly form of the data center built out, you have good paying jobs, whatever. and you make sure to identify what are the actual parts of our community that could use some extra funding and have the AI company pay for that. I think that is sometimes a deal worth making. And I think like the level of skepticism, I would say I get it and I'm empathized based on the past history of corporations breaking their promises. I guess like I feel that people are foregoing a lot of opportunity if they just say no, no, no, no, no. There's no way in which this deal can work.

Well, Jasmine, I'm so grateful for your reporting. Thank you so much for coming on and breaking all of this down. And I can't wait to see what you cover next. Thanks so much for having me on. This is such a fun conversation. All right, that's it for this week's episode of Power User. If you like my work, please, please, please buy a paid subscription to my substack newsletter at UserMag.co.

That's Usermag.co or support me on Patreon. I have almost no advertisers, certainly no ongoing advertisers for this channel. And I cannot continue to produce this podcast without your support. So please, please, please support me via the links below. On both my Substack and Patreon, you get access to bonus episodes, monthly Q&A live streams, and more. I'll be back with a brand new episode of power user next week. See you then.