Local implications of data center growth

[00:00:27]

SARA WHITLOCK: Hello, everyone, and welcome to SciLine and EPI Center’s media briefing on the local implications of data center growth. Data center development is accelerating across the United States, and this briefing will cover data center basics, why their growth is accelerating, and how data centers demands for electricity and cooling are impacting local power grids, water resources, and air quality.

My name is Sarah Whitlock, and I’m SciLine’s Scientific Outreach Manager. SciLine is an editorially independent nonprofit based at the American Association for the Advancement of Science and funded by Philanthropies. So all of our services for journalists are free. Our mission is to make scientific evidence and expertise easy for journalists to use as they cover all kinds of topics from immigration, to business, to personal health. And today we’re hosting this briefing together with the AAA Center for Scientific Evidence in public issues or EPI Center. EPI Center delivers clear, concise, and actionable scientific evidence to policymakers and other decision makers, mostly those at the state and local level. And notably, EPI Center is nonpartisan and non advocacy. You can access SciLine’s resources at sciline.org, which include interview opportunities, reporting resources, and our weekly newsletter. And you can see EPI Centers’ latest toolkits and workshops for policymakers on EPI Center’s page within the AAA website.

A couple of notes before we begin. I’m joined here by three experts who have studied data centers development and impacts. So I’ll go ahead and let each of them introduce themselves and their areas of research. Dr. Liang, would you go ahead?

[00:01:51]

XIAOFAN LIANG: Hello. My name is Xiaofan Liang. I’m assistant professor of urban and regional planning at University of Michigan. My research mainly focus on network infrastructure in cities, including digital infrastructure like data center and also the application of AI in urban contexts.

[00:02:10]

SARA WHITLOCK: Thank you. Dr. Coskun, would you introduce yourself next?

[00:02:13]

AYSE COSKUN: Sure. I’m Ayse Coskun. I’m a professor at the electrical and computer engineering department at Boston University. I’m also the chief scientist at a startup called Emerald AI. My research has focused on data center power management and data center power grid interactions.

[00:02:29]

SARA WHITLOCK: Thank you. And Dr. Ren?

[00:02:31]

SHAOLEI REN: Hi, I’m Shaolei Ren. I’m associate professor of electrical and computer engineering from the University of California at Riverside. My research has been primarily focused on data center energy management, power systems, and also the community impacts of data centers, including water, public health, and electricity price.

[00:02:52]

SARA WHITLOCK: Thank you. Before we begin taking audience questions, I’m going to ask each of our panelists a few questions myself. So attendees who are on the line, you can submit your questions at any time during the briefing. You’re just going to click the Q&A icon at the bottom of your Zoom screen, and do please let us know if you’d like your question directed to a specific panelist. We will plan to post a recording of this briefing on our website later today and then a transcript will be added within the next few days. Finally, we’re going to be sharing resources in the chat right before we open up to some questions, so keep an eye out for those. With that, let’s begin. Dr. Liang, can you tell us what a data center is?

[00:03:29]

XIAOFAN LIANG: Data center is basically a digital infrastructure facility that house servers, computer equipment that can be used to process, store, and transmit digital information for users. So this is a very crucial infrastructure for digital life, and it has been around since digital information is present on the web, even though the debate of data center has recently emerged, data center infrastructure has existed for decades.

[00:03:56]

SARA WHITLOCK: Thank you. And what can you tell us about, are there different kinds of data centers and why are they built?

[00:04:04]

XIAOFAN LIANG: Yeah, so there’s many different types of data centers, and there are different ways to classify them as well. So I think the most useful mental models to think of them are three ways. So one is through the size of the data center. So the largest data centers are called hyperscale data centers. They often occupy whole buildings or what they call campus, so that has multiple server halls that can have hosted thousands of servers, more than 100,000 square feet and maybe using powers that’s exceeding 100 megawatts. So these are the cloud factories of big tech companies like Amazon and Googles and OpenAI.

The second type is called enterprise data centers. These are owned and operated by a single organizations. So for instance, many banks have their own enterprise data centers for data securities. So these facilities can range 1-5 megawatts for smaller facilities can up to 100 megawatts with bigger facilities. So some non commercial research data centers may also function similarly to these data center as well. The third type of data center is called edge data centers. These are smaller facilities intentionally placed closer to the end user to reduce the latency in information transmission. The fourth type is called colocation data center. So you can consider these as co working space for different data centers tenants. So basically, they can rent out a space in the colocation data centers by different clients. So they serve multiple clients rather than one client oftentimes. So these are types of data center classified by scale.

So the second mental model to think about them is to differentiate them by purposes. So data center used for AI trainings and inference often demand much higher levels of electricity and cooling capacity, require larger sites cluster more intensively and thus have more interconnections with regional urban systems than general purpose data center. So the general purpose data centers are those that’s used to do typical data storage. So basically, these are the traditional data center that has existed for decades.

So the third mental models to differentiate the data centers are by levels of redundancy. So redundancy is a term that means data center needs to provide extra independent capacity in critical systems. So these critical system could be power, cooling, or networking, so the facility can continue operating in events of natural disaster, power outage, or even human actions. So it’s very important for data center to not have any downtime to make sure that any facility they have to have a backup. So higher level of redundancy means they need to have multiple sets of excessive infrastructure that’s needed, even though most of the time they’re sitting just there as a backup but need to be there, so that this is also one of the ways that developers tend to explain why they need so much infrastructure and they’re unavoidable to local communities.

[00:06:56]

SARA WHITLOCK: Thank you. And can you tell us where there are hotspots for data center development in the U.S. and why those places might be being picked for that development?

[00:07:06]

XIAOFAN LIANG: So first of all, the hotspot has been changing. But I think one of the canonical capital of data center has been Virginia. The state of Virginia and especially in Loudoun County. So this was more of a historical contingency reason. So they have the best fiber optic network infrastructure decades back, and the region started promoting for data center industry very early on a decade back. So they have basically accumulated the most agglomeration impact for data center industry for that particular region. So this is like the quintessential capital for data center, even in the world.

Other state, I would say, more recently, Texas, California, and cities like Atlanta, particularly, are attracting a lot more data centers. I think the factor is actually quite mixed for sighting these days. So for a lot of these places, they used to have better text incentive or they’re more accessible for power and water resource. So I think in general, power, land, water, fiber connectivities, proximity to user demand, text incentive, and nowadays, community oppositions are some of the main factor for developer to talk between what are some appropriate sites.

[00:08:23]

SARA WHITLOCK: That’s make sense. And can you tell us why data center growth is accelerating in recent years and some of the driving forces behind that growth?

[00:08:32]

XIAOFAN LIANG: The most obvious driving forces is the boom of AI and also the intensified international competitions on AI growth. So that’s really the background for that. So, in fact, in the past decades, as I mentioned, data center has been around for a long time, but it’s kind of a secretive industry. No one really care about them. They haven’t been on the news,  but it’s really because now more data center are built for AI purposes. So AI training, AI inference that they have been brought to the front of the public, especially because of their increasing energy and water capacity.

[00:09:05]

SARA WHITLOCK: Thank you. Let’s move on to you, Dr. Coskun. Can you answer why data centers use so much electricity? How much of that energy is used for cooling?

[00:09:18]

AYSE COSKUN: Sure. AI is particularly energy intensive because of scale and specialization. Unlike traditional cloud facilities, traditional data centers, AI data centers rely heavily on these specialized chips called GPUs, which perform massive numbers of calculations simultaneously to train and run AI models. So training or building a large AI model, such as ChatGPT, Gemini, and others that people can query can require thousands of GPUs operating simultaneously, continuously for weeks or even months at a time. So that concentration of computing power translates directly into high electricity demand.

Power density has also changed dramatically in former generation traditional data centers or cloud data centers, a server rack, essentially a cabinet of servers uses about as much electricity as a home roughly. In an AI data center, a single rack can now use roughly ten times that power. So the issue is not just having more chips and more servers, it’s much more power packed into a smaller physical footprint. So cooling, electricity in a data center broadly goes into two main places. One is running the computer equipment, and the other is removing the heat that the computer generates. So historically, cooling accounted for 30, 40% of the total electricity used with new or more efficient designs, especially with the liquid cooling infrastructures, the amount of power going into cooling is now closer to 10, 20% depending on the climate of the location of the data center and facility design.

You may hear the term power usage effectiveness or PUE, this is a popular term. So a perfect score would be one, which means all the electricity going into the data center would be spent or consumed by computing. So any value higher than that would mean that some amount of power is going into cooling. Modern hyperscale facilities report operating around 1.1, 1.2, much better than former generation of data centers, meaning most of the electricity is used more productively and there are some new data centers reporting even going below 1.1. So why is this important? AI data centers are essentially becoming some of the largest single electricity customers in the region. So their demand is continuous and concentrated, as opposed to a city’s demand being spread out, typically. So this changes how we plan and invest in grid infrastructure.

[00:12:15]

SARA WHITLOCK: Thank you. And I know you’ve already talked a little bit about how this power usage is intensifying, but can you say a little more about the impacts on local power grids and energy prices when one of these hyperscale data centers is built?

[00:12:27]

AYSE COSKUN: Sure. As it was mentioned before, a hyperscale data center is typically a 100 megawatts or more. To put that in context, that’s roughly equivalent to the total electricity demand of a small city. So when a hyperscale data center connects to the grid, utilities may need new substations, new infrastructure, transmission lines that deliver the power may require upgrades. Interconnection studies of these data centers could take years potentially, and our electrical grid originally was not designed for rapid additions of single customers at this scale. So typically, if 100 megawatt, 200 megawatts of power is coming up, it comes at various locations in a city, not all in one location. So that’s the big difference.

On electricity prices, the impact of a data center being built depends on several factors. Who pays for the infrastructure upgrades? How constrained the local power grid already is and whether the data center operates in the traditional more rigid form or is it adopting any flexibility measures by shifting or reducing its power when needed. If upgrade costs are shared broadly across all the ratepayers, then residential customers may see higher bills and we saw some examples of this in the recent years. If developers cover most of the infrastructure costs and facility also operates flexibly, meaning, again, it can change its power during high demand periods and the like, then the pricing impact can be more limited. There are also reliabilty implications potentially. Large facilities can increase the peak demand stress on the grid and it can increase congestion in an already constrained area of the power grid. So this is one option, but they can also become part of a better and more resilient solution.

For instance, in data centers, there’s a lot of discussion in various power grid regions in the U.S. If data centers are designed as flexible loads, meaning they can temporarily reduce or shift their electric use during grid stress times, then they can support more stability. They can help avoid rapid infrastructure upgrades, and these upgrades can be planned and done over a longer period of time, they can also help with more renewable energy integration, which also takes time. So all of this affects overall household electricity bills, potentially, grid reliability during extreme weather and how quickly renewable energy can be added. So data centers are no longer just technology infrastructure. They are actually major energy infrastructure actors right now.

[00:15:23]

SARA WHITLOCK: Thank you. And related to that statement you just made, what are the implications for greenhouse gas emissions and climate goals with this increased power consumption by data centers?

[00:15:34]

AYSE COSKUN: Sure. So emissions of a data center or emissions broadly depend on the local energy mix. So electricity itself is not inherently clean or dirty. It depends on how it is generated. If a data center operates in a region heavily dependent on coal or natural gas, then higher electricity demand can increase fossil fuel generation and emissions as a consequence. If the data center is in a region with abundant wind, solar, hydro, nuclear power, then the emissions impact might be much lower. Timing also matters. It’s not the generation, but also the timing of the generation and the usage of electricity.

So many companies, many tech companies, for instance, purchase renewable energy credits or sign long term power purchase agreements with wind or solar firms, and that helps on an annual accounting basis of that company. But the specific climate impact depends on when the electricity is consumed. For example, if AI workloads at a data center are running at times when fossil generation dominates, such as in the evenings, even if that company buys renewables elsewhere, emissions of that region can still be high. This is why there’s growing discussion around doing computing in a carbon aware way or matching electricity use to renewable availability for data centers.

Renewables absolutely have some place in this conversation, but integration is key. Renewable energy is variable, solar, wind, they change on a daily or seasonal basis. So AI data centers can actually help by shifting workloads to times of high renewable generation by using energy storage and coordinating with grid operators as opposed to just being a passive rigid consumer. So without improvements in efficiency, flexibility, and clean energy integration, then rapid AI growth could increase regional emissions. But with the right policies and design choices, data centers can also accelerate renewable deployment and help modernize the grid. AI growth and climate policy are now in a way interconnected. How we design these facilities today will influence emission trajectories for possibly four decades.

[00:17:54]

SARA WHITLOCK: Thank you. All right. Now turning to Dr. Ren, why and how do data centers consume water?

[00:18:01]

SHAOLEI REN: Sure. Data centers use water directly and indirectly. The direct water usage is for cooling the systems, cooling the data center facilities. Indirect water usage for generating electricity and they are of different type of water because direct water is typically the municipal water. In most cases just potable water, and indirect water is a non potable water. And we’re seeing a lot of discussions around the water number. I mean, some people say this data center water is completely fake and others say data center is taking away all of our water. But I would say the real story is a lot more complicated than that because water itself is just complicated. Actually, there’s no national reservoir that data center can tap into because water is a highly localized, highly regional and seasonal resource.

So I think for the communities or individual towns, the most important metrics is the direct water usage. And if you look at some data center systems, they may not use a lot of water. Their total annual water usage number could be pretty low. The most important metric that is not disclosed in those annual number is the peak water demand. Just like when we design the power grid, we have to make sure at anytime during the summer or the winter storm, the power grid is able to supply the electricity to all the demand, but water is the same. Water system has to be designed to meet all the water demand during the summertime and also at all pressure zones.

And in many communities, the public water system do not have the available capacity to accommodate the new industry scale demand from data centers. So some data center, they could be using very low annual amount of water, but their water usage is concentrated around a couple of weeks a year. And that means during the summertime, they’re using really high amount of water and in the order of like million gallons per day, and they don’t use water for the rest of the year. So that’s why you see a really low number, which is really misleading and misinforming the public discussion.

So these millions of gallons of water capacity requests from the local water infrastructure is really straining the system. And in many cases, the communities don’t have that much water available to support the demand. According to the Susquehanna River Basin Commission, most communities in the United States don’t have the available surplus capacity to support this type of demand from data centers. And that’s why or I would say that’s partially why we’re seeing some data centers are using so called waterless dry cooling system to cool their data centers. I mean, it may not be a voluntary choice. It’s actually a forced decision because the town simply doesn’t have that water capacity to meet the request from the data center. And using water is a good thing.

According to some large tech companies, water is the most efficient means of cooling in many places. And yesterday, a leading technology company just announced up to $400 million to upgrade the local water infrastructure to support their needs because this cooling technique can reduce the peak summer electricity demand by 25-35%. So that means if they are using water responsibly and there’s water infrastructure, they can reduce the stress on the grid. And this benefits water is often ignored in the public discussion. People often think, oh, using water is just bad. That’s not true because we use water to wash our hands. That’s a good thing. And similarly for data center, they use water to reduce the peak cooling energy demand. That’s a significant benefits for the whole power system, for the whole society. And it’s just that the US doesn’t have enough water for this data center in many places during the summertime.

[00:22:26]

SARA WHITLOCK: That makes a lot of sense. So to follow up on a little bit of that, is there any ways that you haven’t yet mentioned that data centers can be more sustainable or reduce their water usage, especially in those regions that might be experiencing water scarcity already, which you’ve mentioned is maybe a lot more than we would think of?

[00:22:42]

SHAOLEI REN: So in places like Arizona, yeah, there is water availability challenge. But even in places like Oregon in Virginia, they don’t have the public water infrastructure to support their needs. In Loudoun Water service area, they are serving more than 300,000 people plus many tens of large data centers. You see, one third or I don’t have the exact number, but one third to half the data center using dry cooling during the summer. That’s because the Loudoun Water couldn’t support the water demand from all those data centers. And that’s I think it’s actually pushing up the electricity demand during the summer, which further push up the electricity price potentially.

So, how do we reduce this water demand? I mean, if there’s water available, let’s say in Virginia, in Oregon, we need to expand the water infrastructure. Like the $400 million investment by some tech company yesterday. So we need to expand this infrastructure to supply water to meet the demand and responsibly, and this is very important because we need to plan the water and power infrastructure in a coordinated way. Instead of just blindly, say, if there’s no water, then we’re going to be using more electricity. That’s not a good strategy in many cases, because we’re just sidestepping the challenges, water capacity constraint to deferring the challenge to the future instead of addressing them right now.

I think, because of this new industry skill demand during the summertime, we pick water demand from the data centers. This is also an opportunity for many of the public water system to become modernized. So we did some study recently. We haven’t published that results yet, but in 2030, in the next few years, a new water capacity request by data centers in the U.S. could be reaching up to one billion gallons of water per day. And this amount of water capacity, if you put them together, hypothetically this could be enough to support the New York City’s water supply on typical days. So this is a huge infrastructure need, and similar to the power, it’s a binding constraint. We need to address it so to support this data center growth and also reduce the peak demand from the grid.

[00:25:19]

SARA WHITLOCK: Thank you. So in a little bit of a different direction, what are the local air impacts of data centers and what are the sources of unhealthy emissions?

[00:25:27]

SHAOLEI REN: So these data centers are installing a lot of backup generators, and these generators are called backup, but they are not only operated during an actual outage. They are also operated for regular maintenance, and also when there’s a grid capacity shortage, for example, during the winter storm a few weeks ago, a lot of backup generators just activated in Northern Virginia. And these generators are mostly powered by diesel, and this can pollute the air releasing large amount of nitrogen oxides, PM 2.5, those are fine particles that can penetrate deeply into people’s lungs and create immediate health outcomes like asthma, a heart attack, even premature death.

So we did some study using the EPA’s modeling tool, showing that as of the end of 2024, if these generators release just 10% of the permitted amount, the total annual health impact is roughly 200 to $300 per year. And nowadays we’re seeing a lot more diesel generators installed. So the health outcome could be much higher than our earlier estimate. And this is just a direct impact. And data center are also using electricity, and some of the electricity grid is not that clean. They’re not releasing carbon, but they are also releasing air pollutants like those PM 2.5 or nitrogen oxides.

In the U.S., despite years of progress, the power plants is still a leading source of air pollutants, and this can harm people’s health, especially for the communities living next to those power plants. And the elevated health risk of living next to some power plants is very well documented. And also these air polluters are not just limited by the county line. They can travel hundreds of miles along with the wind. So overall, we had a study to show that in 2028, the US data center health impact in total, including the backup generators and also the electricity usage, it could be roughly $10-20 billion. At the national level, it’s probably modest. It’s not that high, but most of the impact is concentrated around those communities living next to data centers or living next to power plants. Nowadays, we’re seeing a lot of data centers are building their own power plants outside. So essentially, this is turning the data center campus into a power plant. In most cases just fossil fuel power plants.

[00:28:05]

SARA WHITLOCK: Thank you. We’re now going to start asking questions from the audience to our experts. And I want to note quickly that we already have quite a few questions, so please bear with us. We’re going to do our best to get to as many of those as we can. So first off, and also I will note we are about to post a bunch of resources from our experts into the chat, so you can take a look at those, and hopefully they will help you in your reporting and policy endeavors. But our first question, and this is from a reporter at the Tampa Bay Times, they say, in the face of pushback, the data center industry has said that data center’s huge electricity consumption can actually help lower power bills for the public if regulated correctly. Is this actually true? Where has this happened and is it rare? So maybe Dr. Coskun, you can start us off with that one.

[00:28:49]

AYSE COSKUN: Sure, certainly. So what I’m working on is power flexibility, like I mentioned briefly. What that means is when the grid power has high demand in a hot summer day when everybody is using their air condition or during an extreme weather event when there’s congestion in the grid or a power plant failure, data center can actually reduce its power temporarily or shift its load elsewhere. Where does this come from? Data centers run compute workloads and compute workloads have a lot more flexibility than other consumers of electricity. Like if I want to turn on my lights in my home or use air conditioning, it might have some flexibility, but I can’t move it elsewhere, but compute loads are a lot more flexible.

Now, if this is embedded into the new data center designs and also in existing data center operation, then it’s possible to gain some time in massive expensive infrastructure upgrades which are driving the cost essentially. So rather than going for immediately expanding the grid as quickly as possible, usually using fossil fuel because that’s the fastest essentially to integrate in some scenarios or avoiding retiring older plants, etc, and spending a lot of cost infrastructure, we can plan better, make it more resilient, more affordable, and in the meantime, data centers can integrate, while using the available power in the grid.

So power grid is not 100% utilized most of the time, overwhelming majority of the time. It’s a small percent of the times where we hit the peak, and those are the problem periods. So can flexibility in this way help with prices? Yes, by gaining time, providing better planning opportunities for the grid upgrades, and has this happened yet? So two answers. One is, there’s a lot of discussion in the works, especially over the last year. I would encourage looking into PJM and ERCOT particularly, there’s a new program coming out of ERCOT and there’s a lot of flexibility interconnection discussions in PJM, that’s ongoing. The question said, has this happened yet? Not in the specific way yet that I’m describing. However, there’s been analysis showing that energy price increases are not always correlated with data center projects in regions, which says that there are other factors essentially that one needs to look at in terms of understanding pricing.

So maybe that’s somewhat ambiguous answer, but data centers are not the only factor obviously affecting prices in a region. Flexibility is heavily in discussion, regulatory processes being built, data centers are looking into building capabilities of various sorts of how to provide this power of flexibility, how to get faster interconnect. At the same time, pricing analysis shows that data centers are not necessarily the only factor affecting prices, so it’s a combination of factors. Together, when I look into all of these things, I believe data centers can be designed in a way that doesn’t hike up the prices. And in some case, that can help with better control of energy prices by helping with this peak management.

[00:32:35]

SARA WHITLOCK: Thank you. And now moving on to a question that’s in the same ballpark from a reporter at the Kansas City Beacon. And that is, what sort of protections or conditions should reporters look for when trying to determine if data centers will have an effect on utility bills? Maybe we can spat with you and then, Dr. Liang, feel free to weigh in as well.

[00:32:53]

AYSE COSKUN: Thank you. Generally speaking, I’ve looked into this myself and it’s not easy to suddenly look at data and understand whether a single data center project has impacted costs in a particular region. We try to correlate, has there been data center projects in that region in the last couple of years and what the ratepayers have been experiencing. And when we look at this, we’ve seen news that there are places where data centers have been built and ratepayers have been complaining for seeing substantially higher bills.

Now, to understand whether this is due to data center, what I would recommend looking into would be whether that particular area has done infrastructure upgrades. Have they added new generation? Did they have to upgrade or build new transmission lines? Did they have to do new analysis and prevention mechanisms for congestion in the grid. So if there has been grid upgrades following data center projects or before being able to integrate data centers, then that could constitute more evidence whether this pricing increase has been happening due to data centers projects. Because like I mentioned, there are other factors in driving prices and data centers are not only variable, only parameters in this equation.

[00:34:27]

XIAOFAN LIANG: I think Dr. Coskun answer this comprehensively. I don’t have much to add.

[00:34:31]

SARA WHITLOCK: Thank you. And now a question from a reporter at WUNC Public Radio in Chapel Hill. And that is, how does climate change factor into water usage? For example, these data centers need all this water, but at the same time, drought may be more common because of climate change. So, Dr. Ren, you want to start us off there

[00:34:49]

SHAOLEI REN: Yeah, so this climate change will be further compounding the water risk because during the summertime, if there’s an extended drought, then the water availability could become more frequent issue. It’s actually even in states like New Jersey, which is typically not a drought prone state. We were seeing some drought emergencies last year, and at a certain time, I think, a couple of years back, nearly all the 50 states were in a drought situation. So this drought is really becoming a norm, I would say. It’s highly localized and it could be, you know, even though there’s plenty of water elsewhere, but, you know, this town could have a drought problem. And certain like the Dow Oregon, the city could be facing water shortage due to the water capacity constraint. And the water reservoir is located in a national forest, and expanding that water reservoir is really complicated process because of the federal level approval needed, and now the city is seeking some land transfer from the federal government to the city, and then there has been some heated discussion around that issue recently. And also water is probably the first thing we notice when there’s a climate change.

[00:36:19]

XIAOFAN LIANG: And I want to add a few comments on the lack of transparency issues based on my conversation with water resource planners. So a lot of times the developer will actually want to have a transparent conversation with water resource manager and be transparent about how much water they want to use. So there’s actually two kind of models to handle the work from the perspective water resource manager. So they either have an honest conversation and figure out what’s the capacity they need and determine whether they will provide to them. And the second model is more like proactively, they will do a lot of systematic research into their own water capacity to understand what capacity they can supply and what’s their water planning look like in the next 5-10 years. So I think it’s a two way street there for making this transparent. And, yeah, I’ll conclude it there.

[00:37:10]

SARA WHITLOCK: We actually have a related question from a reporter at CBS 7 in Pittsburgh, and their question is, there’s a data center proposed locally, but an NDA was signed with the land owner. How can a news agency cover this when we don’t know the classification of the center? Why is there an early need to cloak what kind of data center it is? And the lack of disclosure causes a huge variety of concern about local resources from electric grid to water use from local streams and rivers. Do you think you might want to add there, Dr. Liang?

[00:37:37]

XIAOFAN LIANG: Yeah, I’ll continue on this line. So data center is a private business, most of the time. So there’s lots of commercial interests in there that they don’t want their competitor to know where they open up a data center or what functions it locates. So even though you know that a data center coming up doesn’t mean you know exactly who’s operating it. So, for instance, the OpenAI data center in Saline, Michigan, recently, no one knows that behind it is OpenAI until very late. Once they set all the deals, then OpenAI announced that that’s one of their start gate location. So there’s commercial interest in there in terms of how to break that in transparency.

So from the planning perspective, when a data center come in, zoning is actually one of the mechanism where you can enforce some public review. So there’s two ways that comes in. So if your local zoning ordinance says, data centers are permitted by right, which means, let’s say you hold an industrial piece of land, and your zoning said, as long as you check the boxes, all the requirement on zoning ordinance, then you can locate. So that’s by right. Then the public will have very minimum ability to demand some public disclosure of the information. But if your local zoning ordinances, data center use specifically has to go through discretion. So that’s not by right approval anymore. So you need to go through discretion. Then local authority usually have a right to look into, not meaning the public will have a right, but at least the local authority will have a right to go in and look into details of the data centers. So in that process, understanding better their energy and water consumptions.

There’s another process called rezoning in the planning process, which means, let’s say, if you have agricultural land that needs to be rezoned into industrial for data center usage, which is increasingly common because especially hyperscale data center require lots of land. Rezoning in almost every places will require a public process. So they have to actually open up information to elected officials and the public and go through public meeting for the community to decide if the rezoning should be approved or not. That is a very standard process where a lot of the discussion or information will open up for data center. But other than that, I think there hasn’t really been enforcing mechanism that I have seen that will ask the data center to contribute information.

[00:39:59]

SARA WHITLOCK: Thank you. And kind of continuing along this line of transparency, we have a question from a reporter at State Broadcast News based in New Jersey, and their question is, what gives you any confidence that tech companies are interested in cooperating with local utilities and governments to manage their power requirements flexibly? They don’t seem to be showing much flexibility in their dealings with New Jersey utilities and regulators, especially when it comes to shouldering increased power infrastructure costs they would impose. Dr. Coskun, do you want to start?

[00:40:27]

AYSE COSKUN: Yes. I mean, I understand where the question is coming from. And traditionally, as was mentioned before, these sectors, data centers and tech companies versus the energy sector, utilities ISO haven’t been interacting very closely.And now what gives me confidence that the tech site might be changing its view and be willing to coordinate with the utilities a lot more. We are talking about multi-year wait times for data centers to be interconnected to the power grid in many regions. We mentioned Virginia before as one of the areas with data center growth, essentially one of the hotspots. Virginia in dominion utility. We’ve seen five to seven year interconnect wait times. That means even if the data center is built in a year or two, which is the typical time, that would have to wait for several more years to get power.

Obviously, this is a big business concern if you are building a data center but can’t connect it to power for many years. In some other regions, by the way, Virginia is not only one giving these multi-year wait times in other zones, we’ve seen three to five years, et cetera, in the US, as well as globally in some other countries as well. This changes the dynamic, essentially, and this is not a quick problem. There’s no silver bullet solution

So we’ve seen interesting anecdotes of tech companies considering building their own generation, taking over some power plants, et cetera, and we’ll see some of that. But overall, the growth is limited by how fast data centers can be connected. This changes the equation, meaning that tech companies are sitting at the table with ISOs, independent service operators, utilities, with regulatory process, taking into account all the stakeholders and the like. We’ve seen more evidence of this happening. For instance, in the PJM case over the last year in 2025, there’s been many proposals going back and forth between PJM and tech companies and some of their energy company partners. So basically the proposals we’re discussing how to build flexibility should be a requirement should be voluntary, What benefits should provide to the data center that’s offering flexibility and the like. This is the part that gives me confidence that tech is and will continue to be looking into this because there’s a roadblock essentially without a quick shortcut to get over.

[00:43:18]

SARA WHITLOCK: And related to how long it can take to get connected to power, we have a question from a reporter at WSHU Public Radio in Connecticut, who’s asking what challenges arise if a data center seeks to generate their own power rather than relying on the public grid? And is there anything you can say about how well that goes and if it is successful in integrating?

[00:43:40]

AYSE COSKUN: There is definitely more discussion of this ongoing data centers bringing their own power. Maybe one thing to highlight is that this could happen in terms of data centers actually being collocated with a new power plant that’s built by that company. There’s also a broader concept, bring your own capacity, essentially. That capacity could mean different things. Could be a power plant, could be more storage, could be power flexibility. Essentially, we can go from the data centers having to have a physical power plant connected to them to a broader concept.

I think the broader concept is useful in understanding how this might shape up depending on different regions specific. Now, what are some issues with this? Uh, generally speaking, there’s a reason why power grid is a public infrastructure serving many others that brings a lot of resilience benefits. When there are interesting cases like this single consumer consuming so much electricity, requiring power grid upgrades, et cetera, then this equation, or this practice, essentially, of how we build things changes, and maybe some of these big companies will bring their own power.

But overall, it’s not a scalable solution, in my opinion, for every data center to come with its own power plant. So we are talking about tens of gigawatts of data center growth, small ones, large ones across the U.S. to support the growing digital infrastructure with AI. And I don’t expect it’s a scalable solution that all of these data centers will bring their power. Uh, similarly, there we could see some microgrid solutions where data center and its power generation is content. But again, microgrids have been around for a long time. Microgrid meaning basically, you have a small grid that’s not connected to the main power grid.

And we have examples of this in the US already, not in the context of data centers necessarily. But again, this becoming the go to solution for all the data centers is unlikely. While we’ll see some of these data centers bringing their own power, having their own little grid around them, et cetera, broadly, we need a more scalable, sustainable, broadly adaptable solution. That’s why I think at the end of the day, data centers have to work with the grid to some extent and become more grid aware or grid friendly citizens to the power grid.

[00:46:27]

SHAOLEI REN: So I want to follow up on this bring your own power point. So data centers have been voluntarily in the past few years bringing their own water to their communities. For example, they have been using called this initiative called water positive, but this is like replenishing more water than they actually consume. Consume water is the evaporative water, which is considered as lost. So this is, I mean, getting more water is a lot more challenging because water is through the global cycle, and getting water is not always as easy as we think. So in many cases, these companies can only replenish, like, currently about half of the water consumption.

And even though they have been spending a lot of money on this initiative, and more recently, I was actually just more recent weeks, we are seeing these companies are doing something addition to those water positive, they’re bringing more water infrastructure to the community because the communities simply don’t have the water capacity to support the needs to reduce the peak power demand. And one of the Newton water representative from the Newton County Water Authority in Georgia, we interviewed about the six million gallons of water, request from a per day request from a data center operator, the representative said, they are taking away the community wealth, and we simply we just don’t have the water. And so that’s why yesterday the big tech company just announced $400 million to upgrade water infrastructure. This is a way to bring their own water capacity to the community. So besides this power or grid integrated data center, I think this community integrated data center is also important.

[00:48:29]

SARA WHITLOCK: Thank you. And following up a little bit on that idea of not having enough water, we have two related questions, one from a reporter at KMSU Public Radio in Minnesota and another from a Michigan-based freelancer. And they’re asking, can data centers water usage impact aquifer levels and dry up an area causing future issues? And what does the research say about the long-term effects of this volume of water usage?

[00:48:53]

SHAOLEI REN: So at this point, at this point, the total volume is relatively modest, if you look at the national level. But if you look at the consumptive rate, because data centers evaporate about 75% of their water usage compared to other regular uses like residential customers, we only according to the USGS statistics, the public water consumption rate for regular users is just on average 12%. So this means, the water loss due to the consumption by the data center is higher impact, and I mean, it could potentially affect the aquifer level availability.

But at this point, the more significant impact is on the water pipe, water infrastructure treatment plant, waste water processing plant capacity. Similar to the data center energy impact on the grid, it’s mostly on the peak season, let’s say, during the summer. And because during the summer, data center cooling energy increases a lot, and that’s also when the other users need energy most. Similar to the power, the water is more strained during the summertime, and that’s also the time a lot of data center are using evaporative system to cool the servers. If they have the water capacity available. If they don’t, of course, they will say, we’re using zero water or very small amount of water. But it’s often a forced decision. It’s not like they prefer to use zero water because that’s drive up their stress on the grid.

[00:50:37]

SARA WHITLOCK: Thank you. Turning to a question from a reporter at USA Today, they ask, it seems like these facilities cannot be built fast enough in Pennsylvania to meet the AI demand. At what point would data centers be fully built out and can enough of them be built fast enough to support current demand?

[00:50:54]

AYSE COSKUN: Maybe I can quickly mention, and others can chip in too. They can’t be built fast enough because of various things. Power is a primary bottleneck. Like I was mentioning before, data center building typically takes a year or two, but right now we are looking at five years, seven years, three years type of power times, interconnect times essentially for the data center.

So I mean, power is not the only factor. There are other supply chain issues. With power grid upgrades, potentially, for instance, there might be a shortage of transformers on the power grid. There could be chip shortages. Building the data center the building is one thing and physical infrastructure, the electrical infrastructure, bringing the service and chips is another element. So there are multiple sources of potential delays, but I would say right now, power timelines are much longer than the other shortages, meaning that power data centers can be built potentially if there was a shorter path to power. I don’t know if others would agree.

[00:52:10]

XIAOFAN LIANG: But yeah, I’ll add in one consideration, which is, I think for planners, they’re not only concerning about whether we’re fast enough for building, but also what happened if it’s too fast, right? So, if at some point, the A bubble bust and we have too many data centers and what do we do with the land use. So I think, in fact, this has to be a consideration that put into the deal early on. So I think increasingly their recommended practice to ask the data center developer to put in a decommissioning plan, which means they set aside a certain amount of money to deal with, you know, tearing down their infrastructure or converting it to other uses, um, if they are not in use anymore.This actually would be a good practice to put in early while you’re negotiating a deal with a data center developer rather than a post hoc effort.

[00:53:00]

SARA WHITLOCK: That’s an interesting point. Another bigger scale question is from a reporter at MLive in Michigan. The question is, what are the potential concerns for immediate neighbors of large scale data centers? What kind of noise do they emit? Does this depend on the kind of cooling or backup power systems that those data centers are employing?

[00:53:23]

SHAOLEI REN: Yeah, maybe I can jump in. So yeah, if they are using so called dry cooling, their fans could be generating a lot of noise, which could be annoying for people living next to the data center. Um, there has been some documented cases where people are complaining about this noise and, it’s in Northern Virginia due to the data center cooling system and also the humming noise. And when they are using their backup their diesel generators, this black smoke will be coming out and then affecting people’s health immediately, especially for people with asthma with lung diseases. So this is clearly not something pleasant. Yeah. So I think these are some of the immediate, visible, tangible impacts that we can see.

[00:54:23]

AYSE COSKUN: It’s maybe. Go ahead.

[00:54:25]

XIAOFAN LIANG: Okay, so maybe to add to that, if I want to rank, you know, the proximity of the effect by spatial distance, then actually, I would say light is the most immediate one or the shorter distance one. So many people don’t realize that data center need to have lights on 2/47. This could be an issue if you know, you don’t want curtain on your window 24 hours on your building. It could affect some people’s sleep. So that could be one consideration. Then I would say noise and air pollution come next as something that can travel further, and Professor already explained it. Then I think slightly more on the regional scale, I think, in fact, traffic is something else that people don’t usually consider. So there are lots of few trucks that might be coming in help with the backup diesel generators that can cause congestion on local roads and also just a pure increase of traffic for both construction, much more severe during construction, but less so during operation is another consideration for the nearby communities as well.

[00:55:25]

AYSE COSKUN: So I wanted to add, I mean, Dr. Ren provides a comprehensive view on the health effects. There are also potential more economical or socioeconomical effects if a data center is local to a community. These are essentially big buildings and large areas. So they might when they come too close to urban areas like in the case of some counties in Virginia, they might change, for instance, the overall type of businesses that are located over there in a long time, Vulnerability of that neighborhood. It might change how real estate is priced. There are these type of also impacts.

Sometimes these impacts could be problematic for the community, like their real estate got too expensive, some of the businesses are being drawn out, and sometimes it could be beneficial because sometimes the data center company builds and contributes into the local community in terms of, we’ve seen examples of companies building parks, supporting public infrastructure, and, it depends on the type of agreement the community or the town or the county over there has with the data center. But there could be socio-economic effects as well in addition to potential health effects.

[00:56:53]

SARA WHITLOCK: Thank you. I know we’ve just touched on this a little bit in this previous question, but we have a summary question from two different reporters, one at the Kansas City Beacon and one from Little Rock Public Radio. That is, is there any consensus on whether data center projects pay off for municipalities in general? And what, if any, are the economic benefits of a data center being constructed within a community? I’ll open that up to whoever would like to jump in.

[00:57:15]

XIAOFAN LIANG: Maybe I’ll jump into that question really quickly because I interviewed a few regional planners before about this question, do you regret having data centers in your communities? I think overall the answer, both from the regional planner side and the developer side is no one regrets. And I think the tax benefit is significant and sometimes maybe underrated, that I think all the stakeholders acknowledge that the tax revenue is a lot, especially if it’s located in rural communities, that means a significant lump sum to the local governments.

And I think job creation is more of a controversial topic because data centers do tend to produce less jobs, especially compared with construction phase, the operation phase, the jobs can drop significantly, like the traffic. But I think that argument could also be argued against it because developer will say, it’s much higher tax revenue per capita, right? So essentially require less training, less service or maintenance required per person for the kind of productivity that produce.

[00:58:22]

SARA WHITLOCK: Thank you. All right. We’re now going to transition to some questions that are coming in from attendees who are based at local state houses and some other community organizations. We’ll start with a question from a state affiliate in Colorado, that is, are data center is more likely to be cited in rural areas versus urban areas. So Dr. Liang, perhaps, that’s a question for you as well.

[00:58:45]

XIAOFAN LIANG: Yeah. I think I don’t have a evidence back answer to this to say how many data center are located in rural area versus urban core. I would say, if you classify them by purposes, there are data center that want to locate closer to urban area, which are those that use for AI inference or edge data center that basically are sensitive to latency of the response. There are some if they are closer to the end user or want to provide faster service, they want to be in the urban space.

Obviously, there’s trade off there, right? So, the closer you are to population, the more uh, you know, complaints you’re going to deal with because you’re closer to people, right? So, in fact, for developer, if they can be far away from people, they will try every single bet they can to be far away from people. So there’s that kind of level of factor to pull them away from city center. But also something in the middle is data center also requires significant infrastructure, right? So if you’re too far into the rural area, then the water infrastructure, energy infrastructure or fiber is not there, then it’s not even possible to have a data center. So all these factor will pull them, you know, kind of in between, most likely not too far from the urban center, but also not completely in a rural area.

[00:59:57]

AYSE COSKUN: Perhaps to add on that, most of the growth we’ve seen in the last five years has been on the hyperscalar front 100 megawatts and above, these are the super large data centers and by definition, because they are super large, they are not located in the middle of urban area, typically. However, we are also seeing more growth in smaller data centers or edge data centers in some cases, and some studies suggest substantial market growth at 20% or more in edge data center market in the next five years.

We have yet to see whether that happens, but like Dr. Liang was mentioning, there are benefits uh to locating sometimes close to the users. We typically need a mix of larger data centers that are bigger, therefore, potentially cheaper in some ways to build in terms of the overall capacity they have, um, but then have some local, closer to urban infrastructure data centers to serve, um, you know, users of various kinds, businesses, government, others. So likely we’ll continue see this trend of hyperscalers growing more, but also substantial growth in smaller data centers.

[01:01:14]

SHAOLEI REN: Um, and I want to add that historically, a lot of data centers are just built in city centers, even in Manhattan because those are serving the financial industry. So those are co location data centers. They are not necessarily small. Some of them are actually pretty big, digital realty, I should mention that anyway, some big names. Um, in L.A. downtown, there are tens of such data centers. And for the new ones, we are seeing some coming to the city center as well, like in Monterey Park, which is Monterey Park, California, it’s a crowded area, and in San Jose, there are I think around ten or more data centers being planned in San Jose. So it’s I would call them city centers.

[01:02:07]

SARA WHITLOCK: Thank you. And now a question from a state affiliate based in New Jersey, and this one is for Dr. Ren. Can you speak about the use of combined cooling heat and power versus the traditional electrical grid?

[01:02:21]

SHAOLEI REN: Well, I guess that’s one way to reuse the waste heat from data centers. So a lot of AI servers are use liquid cooled system for the internal cooling, and so they can tolerate higher temperature set points. Essentially, that can increase the quality of the heat from the data centers that can be used for so called combined generation. So I think it’s one way to improve the efficiency, reduce the stress on the grid, reduce water usage, compared to the traditional grid. It’s, but there are some technical challenges due to that even though the temperature has increased, it’s still considered low grade heat from the data center servers. So it’s a lot of research progress being made in that space.

[01:03:11]

SARA WHITLOCK: Thank you. And a question that’s probably top of mind for many from a local affiliate in Williamson County, Texas. That is, how does local government hold data centers responsible for water and power improvements that are necessary to support them rather than those costs being passed on to the community via fees added to monthly bills. So I will open this up to whoever would like to start.

[01:03:33]

AYSE COSKUN: I think it starts in the planning process. So the types of, I mean, because I work in power, power management, power flexibility, I’ve been following more of the power side of things rather than water. But a data center can’t just connect itself, it needs to at the end plan with the utility or in some cases directly with the utility, some cases with independent service operators at that particular region. This is why we’ve been seeing this regulatory process. So negotiations, proposals, new bills coming out. To basically provide a standardized mechanism or a structure for the data center to integrate. And that would include the accountability aspect as well. So this would be here is how data centers can integrate into the power grid, and these would be the requirements it would have. This is agreement, essentially, that would need to be signed. So it starts with the planning process, and part of it is this, electrical utility and independent service operator side.Of course, part of it is the counties that are approving the data center projects. I think there’s more development over there and others can speak more to that to understand data centers and evaluate things more holistically because data centers don’t really resemble the other typical businesses town or a county administration might be approving or rejecting.

[01:05:13]

SARA WHITLOCK: Does anyone else have things to add?

[01:05:14]

SHAOLEI REN: Yeah, the power access is, in many cases, controlled by the utility or the ISOs. The water is more controlled by the local municipalities because the water is, you know, the city manages the water system in many cases. So they have more more power in terms of deciding how much water to allocate and these things. But yeah, just in many cases, I think they need to protect the remaining water capacity to deal with the future growth because once the data center gets connected to the water service, I’m not sure in other states, but in California, it’s really hard to kick out an existing water customer to accommodate the new needs because, you know, this water rights allocation permission is a really complicated process.

[01:06:07]

XIAOFAN LIANG: Yeah, and I’ll add that the first layer of this is to add some regulatory gates for all these facets. For some state, for instance, like Georgia, they will say, if you use water exceeding a certain capacity, you have to get approval or get some sort of discretion from local authority. Some state don’t have that, getting some sort of regulatory gate for water, energy, all these important components or even traffic, right? So a lot of times some local authority got to know that a data center coming because they cause traffic impact, and that triggered the attention from planners.

The second thing I would say is really to get all stakeholders in one table. So I think some city already starting working group or some sort of mechanism to say, you know, once we have a data center proposal, we pull in people from utility from water, from traffic, from all these related domain that might be impacted to have that conversation to have that mechanism for conversation early on. And I think for planner, we also have kind of a unique tools, which is zoning, which as I mentioned before, you can encode or rewrite your zoning code to require more discretion and local approval for data center proposal. And specifically, for instance, we have seen in DeKalb County, Atlanta, so they actually amend their zoning ordinance to require closed loop water system for the data centers, right? So there are ways that, you know, communities can enforce their preference by writing them into their zoning code as well.

[01:07:35]

SARA WHITLOCK: This is probably a related question from a local affiliate in Illinois and they’re asking what level of government should regulate data centers to best assure the resilience of multi-jurisdictional resources, things like power and water. Anything anyone wants to add there?

[01:07:57]

AYSE COSKUN: I mean, it’s not my area, who should govern it. But I think the key part is to provide more holistic assessment of data center projects. Because like I was mentioning, given the size they reached in terms of how much power they consume, how much water they consume, how much they might impact the specific area. It needs to be determined looking into this more holistic angle of data centers, power, water, health effects, business effects, socioeconomic effects, and then be approved or redesigned or moved elsewhere, et cetera. I think some of this is happening with the awareness and like I said, I don’t have a specific answer to who should regulate this, but evaluation should have multiple aspects to it.

[01:08:53]

SHAOLEI REN: Yeah, I’m not a governance or policymaker, policy research as well, but I want to echo that, we need to have more holistic benefit cost analysis. So using power, using water is in many cases, a good thing, right? We measure the development index based on how much energy use, how much infrastructure we have, right? So I mean, to support this data center growth using more power, using more water, sounds like a bad thing, but if you look at the benefits versus the cost, and then based on that, we can have a more informed decision instead of just looking at things from a negative side.

[01:09:40]

SARA WHITLOCK: Thank you. And a question from a local affiliate in Commerce City, Colorado. And that is, what do you think about complaints related to water pressure after a data center has been constructed? And what can we do ahead of time to warrant against this? So perhaps, Dr. Ren.

[01:09:54]

SHAOLEI REN: So I guess if that happens, then that means the water utility is not doing a good planning job. So because when they do the approval, they need to make sure the press the water can be supplied at at the same pressure as before without affecting the existing customers. It’s then if the water pressure jobs, then they need more pump station. They need more storage, bigger pipes or something like that, infrastructure upgrades. And I hope I mean, in this case, it probably the infrastructure needs to be paid by the developer.

[01:10:34]

SARA WHITLOCK: Thank you. And a question probably for Dr. Liang, which is from a local affiliate in Bloomington, Illinois. And that is, do you recommend that data centers always be special use, or are there zoning districts where you think they should be permitted by right?

[01:10:47]

XIAOFAN LIANG: So this is more of a specific zoning question. So means you don’t go through approval. So there are pros and cons for that. So district if they set up a data center special district, then it means they already put in a lot of deliberation into streamlining what should be permitted, what shouldn’t be what consideration go into it. Then having by right in those district particularly might be okay because approval takes energy. If you’re an area that have lots and lots of data centers and if you have to go through each proposal with that level of detail of unique discretion, it does take a lot of administrative time. Approval or discretion style is more common if you are area that expect one or two or some data center, but not a huge cluster of them.

So, you want to test it out and better understand what the implication would be, but without letting a slide the by right approval, um, there are locations that intentionally don’t set up any specifications on data centers. I asked the planners for their reasoning. I think most of the time they said they couldn’t really deliberate why this has to be different from other industrial use, even though electricity and water demand is higher, but from a regulation perspective, they’re usually carrying more so about the impact, right? So you can have a beer factory that also use a lot of water. So if you put, very specific regulation against water usage alone, then it’s hard to say whether that will affect other industry or opportunity for other industry to come in as well. So I think there are area that basically just couldn’t make up their mind to say, you know, if I set this regulation specific to data centers, how’s that going to affect other industry?

[01:12:32]

SHAOLEI REN: Yeah. And also, if let’s say, stop data center from using water evaporate cooling, then you’re actually pushing up the our grid stress. So that’s why I would say we need a coordinated planning for power and water. These two are highly connected.

[01:12:53]

SARA WHITLOCK: That makes sense. And a question from a local affiliate in Loudoun County, Virginia. What is the scope of authority that local government has when it comes to utility extensions under the purview of the SEC and similar federal agencies, if anyone has contexts to offer there?

[01:13:14]

SHAOLEI REN: I’m not sure about the power side. I think power connection is mostly controlled by the PJM or the meaning those utilities side, not on the local government side. But water.

[01:13:30]

AYSE COSKUN: Sorry, go ahead.

[01:13:33]

SHAOLEI REN: For example, Loudoun Water, yeah, it is managing the services to many data centers.

[01:13:44]

AYSE COSKUN: I was going to also mention there’s also a Federal Energy Regulatory Commission, FERC, I mean, Dr. Ren is right that typically power connections are also handled at the local level by the utility or independent service operator in the U.S., but FERC also participates in the regulatory process depending on the specific topic. Data centers recently is in the agenda, essentially.

[01:14:16]

SARA WHITLOCK: Thank you. And we’ve reached the end of our time here together. So we have one more question, which will give our experts a chance to offer all of you on the line some brief takeaway messages. But I want to emphasize that for those of you who are reporters, you’re going to receive a quick email survey from SciLine when you sign off from this briefing, if you can give us any feedback, it would help us plan our services for reporters. If you’re attending, but you are not a journalist, you can skip that survey, and instead, we have a separate survey for you from the EPI Center, which you will see up here in the chat here. Please take a moment to fill that out. For our final question, in about 30 seconds, what is one take home message from the briefing that you have for reporters and policymakers? I’ll start with Dr. Liang.

[01:14:53]

XIAOFAN LIANG: Since I am a planner, so I’ll advocate for the message that looking beyond techs, job, water, and energy, right, because I think this is typically what the data center discourse has been about. But I think there are impact on housing, transportation, land use, even social vibrancy, considering a lot of them located in business district, quality of life concerns that hasn’t been well discussed in public realm, so I will encourage people to pay attention to these costs.

[01:15:22]

SARA WHITLOCK: Thank you. And Dr. Coskun.

[01:15:27]

AYSE COSKUN: Sorry. The key takeaway is this. AI data centers are no longer just digital infrastructure. They are major energy infrastructure as well. AI will grow. The real question is, will these new facilities operate as rigid, high-demand loads that strain the grid or will they operate as flexible grid supporting resources that help resilient growth and that help integrate clean energy. The decisions we make now in interconnection policy, rate design, carbon accounting, grid coordination of data centers will determine whether AI growth aligns with affordability, reliability, and climate goals.

[01:16:18]

SARA WHITLOCK: Thank you. And Dr. Ren.

[01:16:21]

SHAOLEI REN: Data centers’ power needs water has air pollution problem. So these are the things that we need to address. And but using these resources like water and power is not necessarily bad. Like, water is in many places, the most efficient means of cooling. And data center, I mean, we want to use water for cooling if we can, but the U.S. just doesn’t have it in many places. We need a comprehensive community-integrated approach to address these challenges altogether and to support the data center growth.

[01:17:00]

SARA WHITLOCK: Wonderful. So many thanks to all of these experts here for sharing your insights and the research around data centers impacts on their local communities. And from all of us at both SciLine and EPI Center, thanks to everyone who logged on to gather context for your coverage and policy endeavors. I hope we will see you at our next briefing. Thank you.