Do Americans Hate Data Centers?

[00:00:00] Speaker 1: Hi, and welcome back to the ARDC Debate. As you can see, the band is back together, face-to-face, after a long time, and we're in the U.S. Ever since we came here, and actually the few weeks before that, a lot of conversations have been around data centers being built in the U.S., and there's been a lot of backlash about it. So today, we're going to talk about why do most Americans hate data centers? And to kick us off, let me give you a little bit of stats to get you creative juices flowing. Over 100 billion worth of data center projects have been blocked so far in 2026. Across the last two years, actually, the number is nearly $300 billion. Posts show that over 70% of Americans oppose data centers being built in their community. There's currently 800 groups in 49 states opposing 1,500 planned projects. And there are over 100 active local memorandums. We've also seen very high-profile state-level memorandums. Obviously, while we were here, we saw New York state doing a memorandum. And there's actually over 300 data center bills that have been filed. Some of those bills are different. Some of them are to remove tax incentives. There's been data center bills that would require participation in the grid that we saw a few months ago. Amidst all of this, one of the big news over the last week was that Musk and XAI have installed 59 natural gas turbines without a federal clear act permit. These will emit thousands of tons of nitric oxide when the nitric oxide is released at the point of combustion. And when it goes into the atmosphere, it becomes, uh, it becomes nitrogen dioxide. And as it happens, this is in an area where already there's very high above average asthma rates. The XAI in Mississippi Department of Environmental Quality claim that the gas turbines are mobile and as a result, they don't need a federal permit. But in addition to pollution, beyond pollution, people have heard airport level noise. We heard it. We saw quite a lot of different materials come out of there. The community is struggling. So the question really is, have we gone too far with data centers in the U.S.? [00:02:55] Speaker 2: Okay, first of all, let me say that I totally understand the concerns of American people to the data centers. Uh, I think the name by, which is not in my backyard, uh, movements is actually quite reasonable. If you think about this, um, a data center is a very big investment and someone built a data center in your neighborhoods and, uh, without, maybe it bypassed a lot of a national level or state level permits just to build as a temporary building for now. And they transport tens of, uh, gas turbines or, um, the, the engines that trans transferred, transmitted from the, uh, retired, uh, aviation engines, and which makes a lot of noise. So the data center there consumes your local power, consumes your local water. But what does it give return to the community? I think that is the, the, the, the very basic thing that people usually have. So first of all, does it bring local jobs? We more or less know that how many people do you need to run a data center? Usually for color data center, it's more than one people per megawatt. And for a massive automated hyper-secure data center, it might be less than one. So let's say you have a one gigawatt data center, which is epic in the history. You only need, you only created like less than a thousand jobs. And imagine the same investment. If that goes to a factory, it'll create tens of thousands of jobs. So there's a, uh, lack of job opportunity creation for data center to the local community. And second of all, the data centers are creating a lot of, uh, AI tokens and used by people around the world, but it's not fully used by the people in the local. So people might start to think, why should I, uh, suffer from lack of power, lack of, uh, water and, uh, polluted air and, uh, unbearable noise in the night just to benefit people, benefit people 10,000 miles from me using AI models. So I totally get, get that kind of concern. Um, but the question is why have to build data centers in the neighborhood? America is a big country. There are plenty of places where the inhabitants, the residents are not there. So you can freely build data center. But I think for now, time matters the most. So, uh, they, the desert owners are looking for places where they can build the desert the fastest. They can spend a lot of time to, to change a desert or a middle of nowhere into a desert place. But that takes much longer time in the places where you have flattened lands, you have, uh, uh, material people and the transports and gas pipes. That's where you want to build a data center within one year. [00:05:35] Speaker 1: Manoj, what do you think? [00:05:36] Manoj: Yeah. I think, uh, there are a lot of like constraints that, uh, the data center operators will have to play with. Right. And, uh, they would look for the availability of power, the availability of power source. It could be a, it could be a grid connectivity, or it could be, uh, a source of, uh, like maybe natural gas pipeline or any other, um, sources. Which is required to power the, uh, power the data center. And then obviously water and then like land, cheap land, and also many other factors like there should be people available from the surroundings to come and work in the data center to build it and, and to operate it. Obviously to operate, you need, uh, far lesser people than what you need to build it. But building is actually one of the most critical part of the, the data center. So, uh, right now in America, we have a shortage of like plumbers, electrician, optical fiber cable tech technicians, and all kinds of like labor, which is associated with the data center because of the intensity of construction, which is happening at this point in time. Uh, so, uh, uh, it, it's, it's actually a very booming period and it is pretty natural that there will be apprehensions and concerns about the growing, uh, data center constructions, uh, in, in the U S in, in, around the world. Um, and obviously it, it, it could be a bigger concern for, uh, the local communities as well. Um, but I think that, uh, this is something that the operators need to be more, um, more actively, um, influencing the local communities to make them comfortable with the situations, providing incentives to them, um, um, or like, um, to take advantage of, uh, to take advantage of the opportunity coming up from, coming up from the data centers. Uh, whether it could be like during the construction time or after the construction time or like, uh, having, uh, like working collaboratively with the community to have like, uh, um, uh, to provide other social commitments and stuff like that. So I think that, uh, um, the gap right now between the operator working in isolation, uh, decoupled from the community, I think that is actually the biggest challenge. And with every big, large build out anywhere in the world, such protests are pretty much anticipated, whether it is like a big mining factory coming up in the middle of nowhere, or it could be like a chemical factory coming up anywhere when you're, when you are actually talking about building, obviously there is, there is actually associated, uh, associated challenges, whether it could be pollution and whether it could be like the lack of water, the lack of resources. So it is, it is, it is pretty much expected we are, we are actually making strides towards a, like a better future through data centers and AI and all this technology development. So I would say that like, we have to, um, take the communities along, um, in the data center construction and activities and, uh, make them a part of that, uh, that transformation and the development rather than isolating them. And like, like, like letting them suffer all the, all the negatives of data centers. [00:09:06] Speaker 1: Well, Manoj, a few weeks ago, I met with Phil from Edge Connects in, in France, and at the time, one of the things that he raised is that there's these cowboy data center builders who are basically making a bad name for the industry. And one of the things that we were talking about, I'm paraphrasing here, one of the things we talked about is how most data center operators do want to work with the community. But from what we've seen in the area around the Memphis data center where Colossus one and two are, the noise is crazy. Is there anything that a data center operator can do to reduce the noise? I mean, I guess the fundamental question here is, do data centers have to be this close to the communities? Could they have been built somewhere else that is more remote? [00:09:52] Manoj: Yeah, definitely. So, um, when you have a long-term plan, definitely you can, you can actually build maybe, um, uh, outskirts. And if you look at most of the data centers, even right now, it's actually on the edge of the metros or like on the outskirts. None of the operators are like building data centers inside the town or cities because the land is expensive, right? These are constructed in like cheap lands, which is mostly outside of the town and cities. Um, and, but obviously there are some like communities affected in, in that area. And it is not just with, um, uh, the XAI Colossus, but also with like other data centers. I have a few friends in like, uh, Virginia and, uh, they complain about the noise coming from the data centers close by. So, yeah, it's, it's actually there with every, every data center and it's not just data center. Think about a factory, which is running 24 by seven. They also create, because those factories could also run like chillers, coolers and all that for, for the equipment, which is, uh, which is inside those factories. So with, with, with the ramp of industrialization, these are things which will definitely, uh, come up. And obviously we need to, we need to make the communities comfortable as much as possible. [00:11:13] Speaker 1: Uh, I guess the problem is the scale. So when I think about the, uh, XAI data center, I don't have the biggest concerns around the emissions. And the only reasons why not is that the U S grid, 56% of the energy that was generated this year has been from fossil fuels, 40% of that from natural gas, same as what is generating the XAI data centers. Um, but I think my concern is the scale at XAI scale. The amount of gas turbines and engines they've installed in Memphis is places them amongst the top 30 gas plants in the country. And I wonder if to build a gas plant, there is much more planning. There is many more permits while to build the data center with mobile gas turbines, there was no permits whatsoever. And for me, that's a part of the problem. And that's really more or less what Phil, to me, it felt like that's what he's highlighting when we were talking in Cannes. [00:12:21] Manoj: So on the one side, we have, we are talking about the lack of availability of compute, right? The compute is, is not enough in the world to train better models and to make, uh, to make further strides in, in, in, in, in, in, with AI. And also to, uh, widely, uh, make the inference capabilities available to a larger audience, right? Like larger, uh, segment of the population. So we are compute, we are living in a compute constraint world and why we have to do all this right now to build gas turbines and like to, uh, deploy mobile gas turbines and power data centers, because there is a sense of urgency. Everyone wants to have the, uh, data center capacity, the compute capacity. And, uh, uh, uh, it's actually a fast way to deploy or like to, uh, power data centers. And, uh, the, uh, the grid, if you, the U S grid, if you look at it for the, means, uh, for the best part of the last decade, it was flat. The investment on the grid was almost flat. Uh, and, uh, it, it got a momentum when, when the last data center operators start like ramped up their investments because of AI. And then they started like agreeing to large, uh, long-term PPAs, uh, power purchase agreements. And on the back of those, uh, PPAs, the investments have actually started, but expanding the grid capacity and like building another, like, like thermal or like nuclear power plant. It's like a seven, eight year, uh, uh, project. Whether do we have the patients to wait seven years to have the next frontier model, uh, coming out of any, any of these labs? We don't. So there is actually a sense of urgency, which is creating a lot of these problems, but it's like, uh, uh, whether it is like, uh, SAI or Oracle or, um, the, the Stargate project. Everyone is planning for, uh, the behind the meter power, uh, generation, primarily for the rapid expansion they would like to have. They can't really wait for the grid to be ready. And the, like power companies to ramp up, ramp up their scale and capacity, the time that they're actually traditional companies. They're pretty slow to respond. So it's not going to come. [00:15:00] Speaker 1: I guess, Manoj, one of the questions is, is it, I, I understand that we use AI, we can't wait for the next best model. We, for us is very beneficial, but for those communities that are affected, they're not the ones waiting for the next model. Next frontier model necessarily. They, some of them might be, but not necessarily all of them. One of the things that really has been very striking when we look at the global data center capacity distribution is the U S was under half. Depending on how you measure it, maybe half now. I think it's about 70% of data center capacities in the U S in is it, is that the reason why local communities here are more. Should we say suffering compared to the rest of the world, because there are regions where there is energy surplus. In the U S as it happens, the places where there's an energy surplus, like the Western Electricity Coordination Council basin, that's where Utah, Nevada, most of Idaho are located. And the problem is they have a water scarcity, in addition to water scarcity, they have transmission bottlenecks, they have interconnection queues. But if we looked more broadly, for example, if we looked at some places in Europe, there is power that is available. [00:16:23] Manoj: Come on, Europe is even more pathetic. It's actually, they will actually take like one and a half hours, two, two years to approve a data center project. So when you have a sense of urgency to build compute, I don't think that nobody's going to Europe and building that capacity just because of the, like the, the, the, the bureaucracy in Europe. So, and, and, and, and I don't, I don't think that like, there is enough power available in Europe widely, but obviously there are like areas like the Nordics and there are a couple of isolated places where there could be power availability, but that, that would not be like, uh, to the scale that the, the cloud providers of today wants to have their data centers. [00:17:11] Speaker 1: So let's check the stats so Norway at the moment, Norway is deemed the new loud and county, um, is, has a fast track program for data centers. They have a surplus of hydropower. They have a cold climate, which is, um, quite favorable for data center cooling, Sweden, Finland, France, Portugal, Spain, Italy, all have dedicated data center programs. So these are frameworks to attract data centers. They want hyperscalers to come in. Um, I told you when, when we met that, even when I was, um, in Bulgaria, I spoke with a few developers who have sites, who have electricity. They, they have the contracts ready for data centers to come, but they're waiting for this to happen. So is it that, is the, am I missing something? Should the local communities, are the local communities here missing something? Are data center operators not looking wide enough or taking advantage of some of these favorite programs? Because here I'm counting, you know, a large number of countries, not one or two. [00:18:15] Speaker 2: I, I think it's totally, uh, not totally up to the data center planners or owners to find the best position from a completely, uh, capitalist, uh, point of view. Money goes where it needs it the most. So if I'm the owner of data center, I definitely want to go to the cheapest, most suitable date place to build my data center to make the most of the money. But in reality, there are a lot of constraints. So first of all, does your country allow you to make such large investment aboard and, uh, transport your users data aboard and train your data board? That is one of the biggest question of nowadays, everybody, every country nowadays are building their own data center infrastructure and want to build, uh, bring their own, uh, money to invest in the AI infrastructures. So that's number one. Number two is that if you go to, uh, for example, I think I get one of the reasons why in U.S. it's so hard to find a data center location. It's because if you think there is one good spot where you have enough power, enough water and far away from, from, uh, from any residential areas, but the land is owned by a farmer or land could be owned by some, uh, uh, family who wants to do something else. And that family, you have to persuade that family to give you permission to build a data center. And again, that's my previous statements. If you, if I build a distance on one of the lands, what benefit do I get? Actually for the local hosts, uh, the benefit that they get from data center is not as much as they do something else on the same lands. The hyperscalers are making crazy amount of money, but none of the money is earned. It's made to the local, it's spent to the local. And also there's no more jobs and a lot of pollution and you don't know what, they will find out some, uh, bypass or tricks to, uh, bypass all the permits. So nowadays people have concerns, even landlords have concerns about whether they should let the data center be built on this place or not. [00:20:16] Manoj: Yeah, but there is one point I would like to, uh, uh, point out here because, uh, it's not just like, uh, by pouring in money into the data center projects. It's just, it's not just like, um, there is a big economic activity happening in these locations and that economic activity definitely will permeate, permeate to the community as well. For example, the wages of the plumbers and, uh, the, the carpenters and the construction workers are all time high in the U S currently. And these are actually, maybe they're not really, uh, uh, uh, direct beneficiary of technology in many ways, but these are actually getting. A chance to participate in that economic activity and, and make good, good, good, good earnings. Right. So I think that is one of the way like, uh, uh, the, the community is benefiting and also these long-term investments that, uh, the hyperscalers or the data center operators are actually putting in. It's actually the, uh, uh, reviving the, the grid. It's actually creating investment in more power, uh, uh, uh, production capacity, which in the long run will benefit the community. Because there'll be a good amount of power, there is good amount of, there is a better grid to distribute the power and, uh, in the long-term, if you look at it, it'll be beneficial to the community in, in, in many ways. [00:21:45] Speaker 1: Oh, so actually on this point, it's worth saying that the equipment that goes into a lot of these data centers is also made in the U S. Absolutely. And Shannon and I were just in, in Buffalo, we saw incredible changes to the city. Um, it wasn't, once upon a time Buffalo was the richest city in America. And then ultimately, you know, had a downfall as manufacturing moved out. Now it very much feels on the rise. There's lots of investment. And one of the reasons why is that there is a bunch of equipment for data centers that's been manufactured there. [00:22:16] Speaker 2: Yeah. I think this is totally the correct direction is to develop data center, build the, uh, bringing your manufacturing, know-how and benefit eventually that's a long-term plan. But for the people in the residential area, in the neighborhood, they don't think that's long. They don't think they have to be very, very long vision enough to think that a very loud, pollutive data center is beneficial to the community. They think about the children. They think about the daily lives, they think about whether they can sleep tight in the night. So that's, I think that's a short-term problem, but for now, if you, uh, remove all the questions on the top, you'll realize that the basic contradiction in this, in this, in this program is a very demanding, uh, futuristic infrastructure versus a very lagging. And, uh, flat energy supply infrastructure, so for sure, the energy infrastructure, the grid power, uh, system is actually getting improved nowadays. It's being accelerated. It's just a matter of time. It's definitely a good thing because this will happen sooner or later. If we don't have AI in five to 10 years later, maybe we'll have another technology wave. And again, requires you to upgrade your power capacity and, uh, and the water piping capacity. [00:23:29] Speaker 1: So when, so what is the solution? So we've talked about a few things, but you know, if these gas turbines are truly mobile and temporary, what is the solution? Is SMR is the solution? Is that what ends up tipping the, the scales is, um, uh, some of the other alternative fuels, like, um, fuel cells, which we're gonna end up seeing this week. Is that a solution? What is the solution? So, so go ahead. [00:23:56] Manoj: So I think like, uh, right now we are actually in a scrambling phase, right? So we are actually trying to build the capacity and make these, uh, like power requirements met with whatever systems that we have available right now. And a lot of the systems that we have available right now has their own shortcomings have higher pollution levels, like higher noise levels and all that. So over a period of time, there'll be optimizations happening in all these levels. For example, I know, um, Amazon working with, uh, the, the cooling tower, basically the water cooling tower, fan, um, uh, fan manufacturer. Uh, they're redesigning the fan so that the noise from these cooling, cooling towers will be reduced. Uh, and I know, um, there are a couple of others researchers, uh, happening to make sure that, uh, the water requirements of the, uh, data centers are actually, uh, lowered. and obviously there are a lot of optimizations happening on the power side with more efficient power consumption of the data centers coming up with new high voltage data center designs, etc. So if you look at every stack of the layer, means every stack of the data center layer, in every stack these optimizations are happening and it is inevitable that these optimizations happen because that's actually contribute to the cost of tokens and reducing the cost of tokens. [00:25:35] Speaker 1: So what you're saying is that most data centers are really trying, most data center operators are really trying, they're good citizens, they care about their community. That's actually what Phil said from EdgeConnects. He said, you should have me, when the camera was not rolling, you should have me and a few of my peers sit down and talk about this because we really are trying and we have ideas and we have programs and perhaps some awareness is lacking. Shen, you were going to say something earlier? [00:26:03] Speaker 2: Yeah, I was trying to comment on the grid problem. I think for sure data center have to figure out this problem, not for the people, but also for themselves because the long running business and healthy business has to be self-sustaining and has to be benefiting the community and the country. But ultimately, I think the solution will not be having an on-site power generation because if you have on-site power generation, you'll have the pollution, you can't process the exhaust gas properly and efficiency will be that high because smaller engines are not as efficient as big engines. So that's why we are using grids at home for a reason, instead of everybody has a small diesel generator. So because the big factories, they can burn on a higher scale and they can have a much higher efficiency and they can process exhaust much properly and they can build far away from the city and have a lot of water to cool down the power generation. So the matter is, we also need to do the green. So SNSMR actually is a very good idea. There are some SMRs that can remove people's concerns about pollutions, leakage, etc. For example, the general, the fourth-gen nuclear reactor. So this kind of technology actually is pretty good. It does not create much of noise. It does not create a pollution and it's very secure. It's a matter of people's psychological concern. If you have a nuclear data center, nuclear reactor next to your house, what would you say? It sounds scary. It's very scary. Of course. Yeah. [00:27:48] Speaker 1: But actually, there's nuclear submarines parked in LA that are not that far from residential places, but it's just we've never really inquired into that or it doesn't really make headlines. [00:28:01] Speaker 2: Yes. Actually, there are a lot of nuclear plants around the world. Let's take France example. More than 80% of the power in France are generated by nuclear power plants. Everywhere in the country, you can see a power plant, which is probably nuclear. So when people are used to it, people don't think about it. People trust that the grid company, the power company, can handle it properly. They have a lot of SOPs, procedures to handle any kind of problem because we learned so much from the previous disasters. But nowadays, with SMR, is that technically, it's a very safe technology. It does not have a meltdown. Even if there's a meltdown, it won't produce the water. It won't emit to the atmosphere. But I think that people need to know about it and people need to accept it. That takes some time. Even for a very knowledgeable community, people still have concerns. So that might take a lot of time. But I think for now, in the transition period from a very lagging local grid system plus some on-site gas generation, which definitely is not the best choice, to a futuristic, purely sustainable grid-based, high-efficiency power system. There's a lot of things we can do in between. We can use SMR. We can use a base system. We can build some microgrid. [00:29:24] Speaker 1: Ben-based systems actually add to what the community feels is a benefit. I think that, you know, Monash mentioned something earlier. To me, that really kind of goes down to the center of the problem. It is people need to feel some sort of a benefit. If there is a BEST that has a bi-directional relationship with the local electricity grid and people feel that their bills are lower, maybe that there is something that they will feel is a benefit. I actually think that potentially another one is revenue sharing. It doesn't have to be that the data center is just something that sets out resources. And in fact, when I was talking to Phil from Aligned and Phil from Edge Connects, yes, they're both called Phil, they were saying that, you know, they're looking at that kind of programs or they're basically giving something back to the community so the community sees like a visible benefit in one way or another. So, let's wrap us up. Your prediction. Will we continue to see data center builds in the U.S.? Or are we going to see a pause in activity or a slowdown in activity because of the different memorandums and concerns? [00:30:40] Manoj: There will be short-term hurdles, but I think that like it will not be a big hindrance to the constructions which is planned. Obviously, in many cases, the government is very supportive. Like every country in the world wants to have compute capacity and to have control of their like data, the models and AI capabilities, right? So, U.S. can't really go back on that and I don't think that like it'll be, all of these concerns and struggles will be created as a big blocker to that growth activity. So, I think that it will be a short-term issue. [00:31:29] Speaker 2: I totally agree with Manoj. I think the data center narrative is a long-term, it's related to the human state. It's a very big story to tell and we have to move towards that direction to further intelligence, to more automation of the society. But for now, we have some problems, but we have dealt with bigger problems before, right? Power problem and cooling problem and water problem are not the largest. I think, but also we talked to so many people in the industry, they all are developing some technologies to solve their own little problems. But in general, if we sum them up, that will give many, many options to the data center industry to build a quite sustainable, healthy and community-friendly data center. [00:32:14] Manoj: Just wanted to add one more point to what Chen said. And this is actually even the investments in data centers and the bringing back manufacturing to US, right? It's actually one of the priority of this government and there is a lot of initiative being done towards that aspect, bring back manufacturing and jobs to US. So, when you're actually bringing back manufacturing and jobs, some of these problems are expected. When you're talking about manufacturing, manufacturing is dirty. It's like it's polluting. It needs energy and it needs labor, etc. So, when you're talking about bringing back manufacturing, these challenges are pretty much expected. And I believe that it's the same with the data centers. The boom in the data centers is pretty much like when you're talking about the fourth industrial revolution or the next industrial revolution. It's actually a big, long-term opportunity for the US. And there should not be any [00:33:30] Speaker 1: blockage to that progress. So, if we were to answer the question why most Americans hate data centers, it really is because they're not necessarily seeing the benefit. But, if everything goes to plan, they really should be. And I think it's important to know that we take it for granted that our email runs on data centers, that our social media runs on data centers. When I spoke to one journalist, I think that we talked about the fact that if a data center was shut down and it impacted websites, if it impacted out the services that we get, because most services are digital, that inconveniences us and we also don't like that. So, I think the reason why people have concerns is because they don't see the benefit. So, what we need to do as an industry is help people to see the benefit and also care enough to provide extra benefits where we can. Thank you for joining us and we will see you next week at the AADC debate. Remember to like and subscribe and join our newsletter. We will start sending episode recaps soon.