Data center planning: municipal, utility, and community considerations

[00:00:33]

ELENA RENKEN: Hello everyone, and welcome to SciLine’s media briefing on data center planning, including municipal, utility, and community considerations. I’m Elena Renken, SciLine’s journalism projects editor. With so many data centers under consideration or under construction, today we’ll cover how localities are planning to handle the energy demands from data centers, infrastructure projects that tend to come with data centers, and how data centers affect local sustainability goals.

A little context about SciLine before we begin. We’re a philanthropically funded and editorially independent non-profit based at the American Association for the Advancement of Science. All our services for journalists are free, and our mission is to make scientific evidence and expertise accessible for journalists in the course of your reporting on all beats. That might be reporting explicitly related to science, like news about disease outbreaks, but it might also be stories about education or voting access, where the scientific angle is less obvious but just as crucial. There’s scientific research on all topics that can ground your reporting and evidence. More of our resources are available on sciline.org. You can also click the blue, “I Need an Expert”, button on that website anytime you need to speak with a scientific expert for your story, and we will look for a source with the right background to answer your questions before your deadline.

And a couple notes before we begin. I’m joined here by three experts on data center planning. I’ll let each of them introduce themselves and their topics of research. Dr. Kate Stoll, would you go ahead?

[00:02:15]

KATE STOLL: Great. Thanks, Elena. Thanks for having me today. I’m Kate Stoll, project director for the Center for Scientific Evidence and Public Issues or the EPI Center at AAAS. Our mission is to connect scientists and scientific evidence with decision makers at all levels of government, and we are non-partisan and non-advocacy. Thanks for having me here.

[00:02:36]

ELENA RENKEN: Thank you. And Dr. Kerri Hickenbottom, would you introduce yourself next?

[00:02:41]

KERRI HICKENBOTTOM: Yes. My name is Dr. Kerri Hickenbottom, and I’m an associate professor at the University of Arizona, in the department of chemical and environmental engineering. My research expertise is in data center water and energy use and looking at on site water use strategies for data centers and other large utilities.

[00:03:01]

ELENA RENKEN: Thanks. And Dr. Lauren Keeler?

[00:03:05]

LAUREN WITHYCOMBE KEELER: Thanks so much for having me. I’m Lauren Keeler. I’m an associate professor and center director in the School for the Future of Innovation and Society at Arizona State University. I direct our Just Energy Transition Center and we support communities where new energy infrastructure is being located and old energy infrastructure is being taken offline in advocating for community needs. Thanks for having me.

[00:03:32]

ELENA RENKEN: Thank you all for being here. Before we begin taking audience questions, I’ll ask each of our panelists here a few questions myself. Journalists, you can submit your questions at any time during the briefing. Just click the Q&A icon at the bottom of your Zoom screen, and please do let us know if you’d like your question directed to a specific panelist. We’ll be posting a recording of this briefing on our website later today and a transcript will be added in the next couple of days. With that, we can get started. Dr. Stoll, my first question for you, how are decisions about data center development typically made at the state and local level?

[00:04:11]

KATE STOLL: The push and the urgency to build data centers are coming from a national level, but much of the decision making on data centers take place locally, and the impacts are also felt locally. So decisions about where to build data centers and regulations around data centers are largely driven by zoning rules, permitting, ordinances, and to some extent community benefits agreements, which are often set at the county and local level.

Decision makers are considering many factors, including land, energy, water use, infrastructure impacts, such as power grid stability, proximity to people, sensitive populations, air quality from on site power generation, such as backup diesel generators that can be found at data centers, noise levels, and more. Local leaders are also considering the economic impacts of data centers, whether that be on tax revenue, job creation, or other economic development. So state and local leaders are balancing the pros and the cons of data center growth, which as we know is moving really fast these days.

[00:05:28]

ELENA RENKEN: Definitely. What role do city councilors, utilities, and state agencies play in approving or shaping those projects?

[00:05:38]

KATE STOLL: There’s a lot of players in decisions around data centers. I’ll try to touch on just a few top level things and try to keep it short. City planners and planning departments or boards approve or can deny development permits according to zoning rules or other criteria, which sometimes include community feedback. City councils can also negotiate or approve or deny data center projects if they’re not administratively approved by city planning departments, for example. City councils, of course, also set relevant ordinances and we’re seeing more and more recently that there are ordinances specifically written about data centers.

We’re also seeing some trends of cities and counties and actually state even setting pauses or moratoria on data centers while they shore up their data center specific policies in this time of rapid growth. Other cities are rezoning for figuring out where data centers does fit into the rezoning rules because they don’t always fit into traditional or older zoning classifications. City managers also play an important role at the intersection of lots of different municipal departments and agencies, whether it be land management, transportation, power, water, broadband connections. There’s lots of infrastructure that come into play for data centers.

And of course, state legislatures are crafting policies on data centers, whether it’s tax incentives or requirements to receive those incentives. Sometimes those are tied to job creation or resource efficiency. We’re seeing more bills on that front. They also have oversight of public utilities to some extent and how they’re treating large load customers like data centers. So utilities and public utility commissions are also a very important player here.

Power utilities receive requests from data center developers to connect to the grid. These are called interconnection requests. Sometimes data centers will make more requests than they ultimately use. So utilities have to be weighing those factors when they’re forecasting future power needs and how much they need to build up the power infrastructure in the region. Water utilities, of course, are also receiving requests for water usage for data centers because a lot of the water used to cool data centers comes from municipal taps.

Lastly, I’ll talk about, today there’s many more players, but state agencies such as environmental departments are responsible for air quality permits. Often the backup generation on data center sites, like I said, you’ll hear more about this later, could be diesel generators and so there are emissions associated with that. There’s also environmental assessments required in advance of construction. That’s a brief snapshot. It’s not the whole world of data center policy, but there’s a lot of players across the state, county, and local levels that have decision making roles in both siting and development of data centers. Happy to share more resources on those.

[00:08:56]

ELENA RENKEN: Thank you. Those all sound like good starting points for reporters covering a data center locally. To get some more specific guidance for reporters, are there specific questions that reporters could ask local officials or developers to understand how these decisions are being made? Or based on your experience working with policymakers, what should communities and journalists pay attention to early in the decision making process?

[00:09:20]

KATE STOLL: That’s a great question. We’ve been working with technical experts as well as our partners at the National League of Cities and the GovAI Coalition to create a short list of key questions to ask when considering data centers in your community. This resource could be used by journalists, policymakers, community members. It’s free and open for everyone. The goal with these questions is to reduce the information asymmetry that often appears between data center developers who are building very technically complex facilities, and local decision makers who are time strapped and they have to make decisions about 1,000 different issues in their communities, sometimes with limited information. We’ll make sure that people have access to that resource. It’s free and open for anyone to use and I think it’s a great starting point to get a handle on the major factors that go into weighing the pros and cons of developing a new data center. It covers themes such as energy use, water use, air quality, economic development among others. So I encourage folks to check it out.

[00:10:28]

ELENA RENKEN: Thank you. Let’s move on to you, Dr. Hickenbottom. To start off, how do different cooling approaches change the balance between water use and electricity demand?

[00:10:40]

KERRI HICKENBOTTOM: That’s a great question because there is a direct trade off. When we talk about water use, we can talk about it in two different types of forms in data centers. One is on-site water use. This is generally used for humidification of IT room space, any building personnel needs, but mainly it’s used for cooling. There’s a lot of different types of cooling systems out there and it’s not just a one size fits all type of cooling system approach for a data center. It’s very much based on geographical settings and environmental conditions.

So there is this direct trade off between water and energy. For example, in the Phoenix area, which is where we’re located near, if you’re looking at air cooled chillers, because these are much less water intensive, there’s a trade off. They’re about two-and-a-half times more energy intensive than, say, water cooled chillers. But if you’re thinking about using a water cooled chiller, although these have lower energy demands, they use about 20 times more on site water. So this is very significant. The thing is at this point, water is severely undervalued. And so as a data center, if you are trying to minimize your costs, it would make more sense to go for a more water based solution than an air based solution.

However, what we see is that communities in particular are realizing this trade off between water and energy usage on data centers and pushing back and saying hey, these are our resources and there is value at them that’s not in the form of a dollar sign. And so most data centers that we see now in the Phoenix area or coming to the Phoenix area are air cold chillers or some hybrid that really do try and minimize the water use. I do want to also say that even though these air cold chillers don’t necessarily use water on site except for maybe IT room humidification, there’s a lot of water that’s embedded in energy generation, especially if we’re thinking about thermoelectric energy generation or other energy generation such as hydropower. These are very water intensive ways to produce energy. Even though they might not be using water on-site, there is that whole portion of the usage of water off-site for energy generation.

[00:13:37]

ELENA RENKEN: Thank you. That’s really good to remember the whole use of water throughout that process, even if it’s not necessarily at the end. How might those trade offs you mentioned affect communities in different climates or regions?

[00:13:50]

KERRI HICKENBOTTOM: Right. Yes. So in the Phoenix area, air cool chillers aren’t necessarily going to be as efficient as using an air cool chiller or even using outdoor air to essentially economize the chilling process, say in a cooler climate. There is this shift of data centers looking at cooler climates where they can use free cooling. Using that outdoor air or some percentage of outdoor air to actually cool the IT room space. So we have started to see a little bit of this shift in where these data centers are placed, and it’s very climate location based.

[00:14:39]

ELENA RENKEN: Thank you. How often do data centers upgrade their computing equipment and what happens to the hardware that’s replaced?

[00:14:47]

KERRI HICKENBOTTOM: That’s an interesting question because it really depends on several factors. It depends on, for example, what the server is being used for. If it’s used for really high density computing, say for AI or if it’s just used for storing data. It really depends on how the servers are used, but also how that IT room space is operated. If those servers are operated at very high temperatures, they tend to have a shorter lifespan. When we talk about servers and server racks, data centers is basically a massive warehouse that houses these server racks. These racks are very, very heavy. They’re about one to six tons. Again, that depends on the density of the servers and the type of data center.

So there is this whole supply chain and economy that data centers in particular are really starting to think about, especially in this global economy. So for example, the servers, there’s this trade off between operating at these higher temperatures in the IT room space. However, you have more server failure. So you’re saving on operational energy, but you may have to invest in replacing your servers more often. It really varies widely, again, under these different conditions, but anywhere between two to maybe six years is the general lifespan. Oftentimes, what they do with these servers after they’ve reached their lifespan is they might have some extension via improved maintenance or try to say maybe wipe and refurbish the servers for other use. They could donate them to schools or non-profits or things like that. So there’s lots of different levels where these servers can be reused.

However, inevitably, they do end up, as e-waste, and there is this whole chain of informal recycling and emerging economies and really a need for better regulated recycling practices. And so it’s a really large picture and data centers are looking at this, or looking at how to reuse these servers, recover some of the raw minerals, these rare earth metals that are very expensive and challenging to get. And then also all the metal that’s used for these servers. So it’s a really big challenge and a lot more research and design and engineering and policy is needed for thinking about not only the water and energy needs for this operational space, but when we start to decommission these spaces, what is the end of life for these systems?

[00:18:04]

ELENA RENKEN: Thank you. That’s a good example of ongoing reporting that journalists can do even after a data center is up and running. I also wanted to ask about the many facilities that rely on diesel backup generators for reliability. What are some community concerns with these generators? And are data centers pursuing new technologies for energy backup?

[00:18:26]

KERRI HICKENBOTTOM: Yes, on-site diesel generators is really a critical part of data center resiliency. These data centers house really critical important information anywhere from our personal medical files, to government documents that are highly classified. So having some breach or grid backout is a huge problem. These data centers, they have on-site backup diesel generators to help in case there’s any power outage, so these data centers can keep running. The thing is, oftentimes we think of our grid being quite stable and there are occasional power outages and not all the time, do these data centers need to use these backup diesel generators. They do need to be checked and maintained. So usually on a monthly basis, data centers do some check for these diesel generators, and as Kate had mentioned before, there’s a lot of noise and pollution concerns with local communities.

There is a lot of interest and ongoing research in a couple of alternative technologies that could replace diesel generators. For example, looking at different battery technologies like flow cell batteries is one thing that data centers are starting to look at. Also, hydrogen fuel cells is another good example of a backup energy source that is cleaner and less disruptive to local communities. And then finally, I would say there’s also a lot of interest in looking into microgrids. They tend to be more resilient, more nimble and can operate independently in case there are these breaks in energy supply from the main grid.

[00:20:29]

ELENA RENKEN: Thank you very much. And over to you, Dr. Keeler. My first question for you, how do large infrastructure projects like data centers fit into broader community planning and sustainability energy goals?

[00:20:44]

LAUREN WITHYCOMBE KEELER: Well, as you heard just now, data centers require a lot of energy and water resources. Then on the operational side, provide minimal long term employment opportunities. Most of the data center jobs that you see are going to be associated with the construction. Deciding to cite a data center in your community requires consideration of trade offs. Where is that water and power not going, because you have cited a data center there. The power required for data centers often requires the construction of new energy resources. It outstrips our existing grid. However, some data centers are building on-site generation. This is usually natural gas. But most of these data centers are going to require that local utilities build new power plants in order to meet that demand.

Here in Arizona, that power is increasingly coming from natural gas. Now that’s cleaner than coal, but there still are air quality impacts including CO2 emissions. Consideration of how to achieve sustainability goals like net zero is definitely impacted by this rapid expansion of energy demand. Renewables can power data centers, but with things like wind and solar, you also need battery storage because although power demand is variable at data centers, they operate 24/7. So you’re going to see round the clock demand.

The cost of new plant construction whether it’s renewable or gas, unless special circumstances have been dictated is going to be passed on to utility customers, and so it has the potential to raise utility rates. And so I guess when thinking about broader community planning, I want to emphasize data centers don’t need to be in your community in order for people in your community to access the internet or even to use AI to perform specific tasks. However, when communities are looking at a higher tech economy, having proximate data centers becomes important and so it may shift the trade off considerations and having those nearby.

[00:23:20]

ELENA RENKEN: Good to know. Thank you. What kinds of infrastructure upgrades like transmission lines, substations or roads tend to accompany new data centers?

[00:23:30]

LAUREN WITHYCOMBE KEELER: Yes, it’s a really important question because the conversation about data centers is necessarily a conversation about our electricity system. Just for reference, in Arizona, the demand queue, which is the list of future customers that are requesting power from our utilities, they are asking for as much power to be added to our grid in the next five years as has been added to the grid in the last 50 years. So there is no meeting that with our existing energy infrastructure we have to build. So the biggest infrastructure upgrades are going to be to the energy system. Yes, new power generation, new plants, new transmission. If it’s natural gas, that also means new gas pipelines. A data center can use as much power as 100,000 homes. So they’re incredibly energy intensive and our utilities really are struggling to keep pace with this new demand and that’s another challenge of these infrastructure upgrades.

You mentioned substations, the lead time right now for a 500 kilovolt transformer, which takes power down from those high voltage transmission lines is 5-7 years. The lead time for a generator for a gas plant is another five years. So there is a lot of competition right now for a limited set of resources to build the energy infrastructure necessary to power those data centers. While there’s a great deal of attention, I think, right now because of the legislature in Maine passing moratorium on data centers on the impact of community opposition to where data centers are located. I think the biggest throttle on the data center expansion continues to be our ability to expand the electricity system at pace.

[00:25:40]

ELENA RENKEN: Very useful to know, thank you, that infrastructure capacity is a key part of the story. And what types of benefits or protections can communities negotiate when a project is proposed?

[00:25:54]

LAUREN WITHYCOMBE KEELER: Well, I think first and foremost, communities can and need to demand transparency from data center developers as a part of the siting and permitting process and on an ongoing basis when data centers are operational. Water use, energy use, quality of water that’s discharged from data centers, all of that should be monitored and provision should be made for continual improvement, some of those updates that Kerri mentioned are possible. In addition, communities can negotiate revenue sharing where a portion of the revenue that’s generated by data centers could be paid into something like a community fund that could go to specified purposes, like improving education or community amenities.

I think a really important piece back to the expansion of electricity infrastructure is making provisions that safeguard utility customers from the impacts of the energy system expansion on their electricity bills. There’s a number of models that are being proposed that would pass the cost of new infrastructure construction onto those large customers like data centers, rather than onto the average utility customer. I think that is strongly encouraged, particularly given that there are some fears that maybe all of this demand for data centers could be wrapped up in an AI bubble that doesn’t completely manifest. What you really don’t want is an overbuilt energy system with your average household having to foot the bill for too much energy infrastructure over the next 25 years. There’s an opportunity, as Kate mentioned through things like community benefits agreements to channel some of the wealth that’s being generated by this creation of data centers and the expansion of AI to communities to get funds to do projects that communities want.

[00:27:58]

ELENA RENKEN: Excellent. Thank you all so much. We’ll now begin asking questions to our experts, and I want to remind reporters on the line to please submit your questions using the Q&A box found at the bottom of your Zoom screen. But first, I want to ask all three of you, what is one misconception you often encounter when talking with reporters about data centers? Dr. Stoll, would you start?

[00:28:23]

KATE STOLL: Sure. I think one I’ve heard relates back to what Dr. Hickenbottom was talking about, which is that this misconception about data center water use on-site versus in the entire system. So there’s a misconception that if you have a closed loop cooling system that you no longer need to think about water consumption. But that doesn’t account for the water consumption that might be increased at the power source, for example, if you’re having a higher energy draw at the data center. So really taking a system’s view because as Kerri said, energy and water are linked. And so thinking about the whole system is an important factor.

[00:29:07]

ELENA RENKEN: And Dr. Hickenbottom?

[00:29:10]

KERRI HICKENBOTTOM: I’ll just build a little bit off of what Kate said. Yes, there is a misconception that some data centers are completely closed loop and don’t use any water. When they talk about closed loop and data centers, it can really mean one of two things. One of the closed loop aspects is closed loop in the IT space, meaning that there’s more direct cooling of that IT equipment. That is related to cooling inside the IT space. It doesn’t relate to the actual cooling system. Closed loop, for a cooling system could just mean that they’re circulating that water inside a closed loop heat exchanger, but that water or that working fluid could be cooled by evaporative cooling or cooling tower. It’s very deceptive this idea of closed loop, and we’re talking about it in data centers, it can either be in the IT room space or the cooling system. The last thing I would say too is the misconception of water. I don’t think it’s necessarily, maybe it is more of a misconception, but we have one water source. That water gets reused over and over and over again. So they might not be using potable tap water, but data centers could be using another water source that say, could be used for irrigation or aquifer recharge. Water is water.

[00:30:51]

ELENA RENKEN: Thank you. Dr. Keeler?

[00:30:53]

LAUREN WITHYCOMBE KEELER: I don’t know if this is a misconception or just what I see as the dominant narrative in the media as a yes or no to data centers. Those are the choices. I really want to emphasize that certainly no to data centers and the opposition around citing them to communities. That is an option and there’s been quite a bit of success among communities in stopping data center projects. But the yes side of the equation, there’s a lot of nuance and there are a lot of things communities can do in permitting a data center or allowing a data center to be in their community that can really direct benefit to those communities. So on the yes side of the equation for hosting a data center, there’s a wide range of benefit potential that communities can pursue as opposed to simply no. We’ve seen a bit of success on that side too.

[00:32:02]

ELENA RENKEN: Thank you very much. Now we can move on to questions from our audience. First off, from Illinois Public Media. Are you seeing people’s opinions on data centers changing? For example, are areas that may have been excited about the economic development a year ago now more hesitant? Is anyone able to comment on any perceptions there?

[00:32:26]

LAUREN WITHYCOMBE KEELER: Let me say, I haven’t seen a reversal in the economic development community among folks that are courting data centers. But I think the Arizona Governor, Katie Hobbs, is an example of a shift to a more nuanced approach to data centers. In her 2026-2027 budget, she’s proposed removing the tax incentives for data centers. So it isn’t to say, “We don’t want data centers here in Arizona,” but instead to say, “Hey, there’s an opportunity if you’re going to site here to share as well.” So I think the opposition to data centers has grown, it has empowered communities and decision makers to ask more of data center operators in exchange for their siting.

[00:33:21]

KERRI HICKENBOTTOM: I’ll add a little bit onto that. Data centers in a large community, a large metropolitan community, for example, people can have very opposing views, or other cities within that larger metropolitan domain can have very opposing views. One can look at the data center as an asset, as Lauren said, they do bring in some tax monies. Also, as was mentioned, there are some levers that communities can twist. They could say, “Okay, you want a water source. Well, you’re going to have to take this waste water, and you’re going to have to clean it up and use that as your water source.” Or, “For this space that you’re developing, we want the X percent or amount of that space to be developed as park or recreational areas.” Or, “Okay, you need this energy to be built out. We want you to invest in a microgrid.” Or even data centers will say, “We want to invest, we want to have only renewable energy. So when we build out the energy or make a contract with you, we only want renewable energy.”

So that money is going to these types of infrastructure projects, which can be looked at as a positive thing, but then there’s also cases where data centers aren’t incredibly transparent in the resources that they are going to need, and there’s also this long-term sustainability. There’s a threat, a risk that if the data center pulls out and goes to a different area because something has changed, maybe tax incentives or XYZ, that now this community has this huge infrastructure that’s been built, and now the cost will be imposed on them. There’s still a lot of room there for policy, transparency, and the communities do have an active voice. We, as people, have representatives, and if we have questions about these types of utilities, these big projects coming in, there are hearings where the public can voice their concerns and their interests. It’s a very dynamic situation, and it just changes from place to place.

[00:35:43]

ELENA RENKEN: Thank you. A question here from Spectrum News 13 Orlando: “Regarding energy and water usage, who usually pays for the costs of the use of all the water and the energy for these data centers? Is this something that residents in the host cities will have to pay for?” Anyone able to give some context there?

[00:36:07]

LAUREN WITHYCOMBE KEELER: I can start us off. As Dr. Hickenbottom mentioned in her opening remarks, data centers are operating as a municipal customer, so they’re taking the water that you would take from your municipal utility. They’re buying it just like any other large water customer, and I’ll let Kerri expand on the water side. As I mentioned on the energy side, when utilities look to expand energy infrastructure, the cost for that expansion gets amortized over a number of years and paid back through increased rates. Those rates, unless special provisions are made, are paid by all ratepayers.

Now, we have seen proposals come from our utilities here in Arizona for a special class of energy user, that is, a large energy user, and imposes a tariff and requires that those new large customers pay for the infrastructure that they require for expansion. I think that’s a really good model. As I mentioned, I think what you really don’t want to see is that the energy system gets overbuilt and the rapidly expanding demand for AI doesn’t last, and those costs end up getting borne by your average household paying their electricity bill.

[00:37:45]

KATE STOLL: I think we’re seeing more of some of the largest tech companies saying upfront that they will pay for infrastructure buildout. But it’s also important to know that there’s lots of data center developers that aren’t necessarily the names everyone knows. There’s a lot of people in this space, so it does seem like policy is an important factor in order to get those types of agreements with all data center developers.

[00:38:13]

KERRI HICKENBOTTOM: I’ll just add on briefly that water utilities are doing similar things as energy utilities, and that even though there’s this large upfront cost for build up, there’s also a very large cost for maintaining these facilities. And these facilities, it’s not like you can just turn one part off; they’re designed at a certain capacity, and that capacity is at a certain efficiency. If they operate outside of that threshold, the efficiency reduces, then you have inflated cost. The thing is, generally, these data centers are in clusters. We see them in Virginia, we see them in Phoenix, we see them in Central Oregon and Washington areas.

They’re in these clusters, and anytime you have a cluster of things, there’s good and bad. But when we’re talking about energy and water and large infrastructure projects, it can really put a stress and strain, and I think that’s why it’s really important to shift our mindsets to looking at decentralized systems, on-site water treatment and reuse, microgrids, building out more resilient energy systems and having a larger portfolio of different types of energy and water resources that these systems draw from. So it’s just not one huge giant infrastructure project; there’s a lot of different flexibility and how we manage these resources.

[00:39:46]

ELENA RENKEN: I have a question here from KSHB 41 News in Kansas City that gets at some of those longer-term concerns: “Since hyperscale data centers haven’t been around for very long, are there any studies or reports we can look to for answers on how these centers could affect a community five to ten years from now? What would you tell neighbors who are skeptical about companies saying there won’t be any adverse long-term effects?”

[00:40:15]

KATE STOLL: I haven’t seen studies that look to the long-term impacts of hyperscalers because they are newer than other types of data centers, which have been around for decades, and the capacity or just the resource draw is so much bigger with hyperscalers than with other data centers. I think it comes back to what Kerri and Lauren were talking about, about infrastructure buildout, that it is one of the risks if you either overbuild or you underbuild. If you overbuild, then the costs get passed on to rate payers unfairly. If you underbuild, you have instability in the grid or the water infrastructure systems that result in brownouts or other problems. I think that’s both the near-term and the long-term risks that we’re seeing with the really big hyperscalers.

[00:41:06]

KERRI HICKENBOTTOM: That’s a great point, Kate’s made. This is a new frontier that we’re dealing with, and I think the only thing we can really do is turn to other examples of larger industrial users that have once been really predominant in communities and then now have located elsewhere or overseas. The thing is, for the long-term impact for this huge infrastructure that’s built to sustain and support these industries, and if they do flee or go to another area, how do you decommission that infrastructure? Utilities are often, at least water is very public, energy is usually kind of a mix of things. It’s a lot of consideration for decommissioning, a lot of funding, a lot of monies. So I think that’s probably more an example of a long-term effect.

[00:42:12]

LAUREN WITHYCOMBE KEELER: I just want to echo this notion that a lot can be gleaned from looking at other large-scale industries developed over relatively short periods of time, and there’s plenty of cautionary tales. I think an overarching one is that reducing community and environmental impact requires policy, it requires monitoring, and it requires consequences that matter to the companies that are operating these facilities. I think having a model of self-monitoring without external guardrails has not proven to work when we look at things like the fossil fuel industry, the chemical industry, for example. One brief consideration that has come from studying the development of energy projects is the impact of a rapid rise in construction jobs in communities that don’t have the workforce for them.

That’s not something that I’ve seen in a lot of the reporting, but if you’re a small community and you’re going to have a data center built and it’s going to require 1,000 people to build it over a couple of years, there’s temporary housing that goes in. There’s often social issues that are associated with a large collection of temporary workers coming into a space. It can put strain on local healthcare resources. So while we don’t know exactly what five to ten years looks like for data centers, the history of industry and the energy industry in particular, I think, offers a lot of cautionary tales.

[00:44:01]

ELENA RENKEN: Here I have a combined question from KOPN Public Radio in Columbia and from WNDU in South Bend: “Who benefits most from data centers, and what inequities are at play in terms of higher rates, land use, displacement, etc?”

[00:44:23]

KATE STOLL: Some of the benefits do depend on the tax structures that apply to data centers in each location. You can have tax structures that apply either just to the property of the data center, or can also apply to the capital, or the sales tax for the resources inside the data center, all these servers that Dr. Hickenbottom was talking about. That’s the expensive part of the data center, not the warehouse it’s put in necessarily, and so communities that have tax structures that include some of those additional, either sales or capital expenditure taxes get a lot more revenue than those who are limited to just property tax or have really high tax abatements for data centers, for example. So revenue can be a huge benefit to communities with respect to data centers, but it depends on how it’s set up.

[00:45:33]

LAUREN WITHYCOMBE KEELER: I agree. The question of who benefits, the answer is, at least in part, it depends. A community can negotiate substantial benefit from the siting of a data center. In the absence of those kinds of negotiations, the benefits can be minimal for a community. As I said, the direct jobs associated with data centers are generally relatively low. I also just wanted to point out that the part of the expansion of data centers, which predates AI, is the expansion of connected devices in all of our lives. The average American household has 21 connected devices in it, and the processing of data for those devices goes on in data centers.

Increasingly, critical services that we rely on are using AI, and have been using data centers to do processing that includes our healthcare system, for example. Data centers do provide an infrastructure that increasingly underpins the essential functions of our lives. Figuring out a way to allow communities that are impacted by the presence of data centers directly, to benefit from their presence and to minimize harm, I think, is really important, particularly if we think data centers really are a critical infrastructure to the US.

[00:47:18]

ELENA RENKEN: I have a question here from the East Oregonian: “How do you recommend journalists who are new to this issue start looking into the impacts their local data centers have had on the region, especially given that data center representatives are typically pretty reluctant to give details? Who should we be talking to and what questions should we ask?”

[00:47:38]

KERRI HICKENBOTTOM: I would say utilities are really great places to start because they’re the ones that are the interface between the data center and the community in a way. When a data center comes in, they need to talk to the utilities and see if there’s capacity for their buildout. I would say that’s a good non-partisan type of place to start. I think there’s a lot of questions that even utilities are struggling asking data centers, and are not equipped to deal with this new type of infrastructure in their communities. There’s also information that data centers should be more forthcoming when they’re talking to communities as well. So I would say utilities would probably be a good place to start.

[00:48:48]

KATE STOLL: I would agree. I think transparency brings more trust for communities, and it is sometimes hard to get answers about some information that could be proprietary to the business model of the data centers. But remember that communities do have bargaining power. They do have leverage to get the information they need so that they can make decisions that make sense for their communities.

So I think asking the questions more than once if you need to is perfectly valid. Some of those questions are particularly important when asking about peak energy use, peak water use, not just average energy and water use over the course of a year. That’s when you get infrastructure strain, it’s usually in those hottest days of the summer. I would start with those two questions: “What would be the peak energy and peak water use of this particular facility or campus?”

[00:49:47]

KERRI HICKENBOTTOM: I’ll also add city managers because they often are the ones that are negotiating with the data centers. The data centers come in, and they’ll start talking to the city manager. I would also say that’s another good start, and they can tell you a little bit more about policy and ordinances. Every city is very different.

[00:50:12]

ELENA RENKEN: Here we have a question from WNDU NBC 16 in South Bend, Indiana: “What are the most important questions a planning board should ask before approving a project? How do communities verify compliance after approval? What kind of monitoring and penalties actually work?”

[00:50:38]

KATE STOLL: I think a lot of the questions can be around the balancing of resource use and the trade-offs when it comes to economic development. So again, I’ll point to our list, I think it’s like 20 or 25 questions we think are essential foundation that you can ask data center developers. We’ll make sure everyone has that resource, but it’s around energy use, water use, air quality, job and revenue generation. I think it’s a valid point to talk about enforcement of existing environmental quality laws, for example. Enforcement of air permits and monitoring air quality is just as important as having those policies there in the first place.

[00:51:28]

LAUREN WITHYCOMBE KEELER: I would just add on the monitoring side that it’s a good idea to negotiate for third-party monitoring, rather than self-monitoring, and have punitive financial punishments for non-compliance. I’ve seen a community ask for $50,000 a day for willful non-compliance with environmental provisions of a benefits agreement.

[00:52:02]

KERRI HICKENBOTTOM: Yes, and then there’s a lot of nuanced questions to ask: “What type of data center is this? What type of water requirements do you have? What types of discharges are there going to be? What’s your backup energy system going to look like? What’s your buildout plan? What are some community benefits? How do you plan to regulate and monitor these things?” I think the more questions you have and the more data you have, the better equipped you are, and I think it’s not just one set of questions; things come up. So I think really being engaged in the planning process and being honest and working with other utilities, too. In cities as well, because some cities and utilities have developed city ordinances for water use.

This is how much water use we can give per parcel of land; this is how we did this permitting process. That’s something that collectively, I think, can be done as well. That would be nice for cities to have a data repository and sharing of information.

[00:53:46]

ELENA RENKEN: A question from KJZZ News in Phoenix: “As utilities face unprecedented demand growth from large users like data centers, could they end up extending the lifetimes of coal plants? Are there any signs that coal could end up being used longer than utilities had projected as they shore up energy generation sources?”

[00:54:10]

LAUREN WITHYCOMBE KEELER: It certainly could be. I recognize you’re from KJZZ which is my local station here in Arizona. I want to note that we haven’t gotten an indication from our utilities that they had planned to extend the life of the coal plants that already have scheduled closure dates. The plan, or what is being proposed, is conversion to gas. But certainly, and then I’ll speak generally and not about the Arizona energy system and as the director of the Just Energy Transition Center, where we’re working with communities that are impacted by coal plant closures, one of my concerns is a buildout of natural gas infrastructure that locks us in to this carbon-based energy source for the next 25 years rather than just extending the life of a coal plant for five years while we figure out how to meet energy demand with renewables and battery. I don’t necessarily think we get a better deal by shutting down our coal plants and converting to gas.

[00:55:22]

ELENA RENKEN: Another combination question here, I’m wondering if anyone can weigh in on. This is from the Arkansas Democrat-Gazette and the WYSO Public Radio station in Dayton: “It seems like a lot of secrecy goes into the application and rate agreements for the data centers. Are you finding that data centers require or ask for more secrecy than other types of industrial projects, and can it be difficult to report on data centers due to disclosure agreements? Do you have any insights on what documents could be publicly accessible on these proposals?” Any advice on these less transparent aspects?

[00:56:02]

KERRI HICKENBOTTOM: I think that’s really a cause for concern. Anecdotally, I’ve heard of some communities where there’s a permit for a large warehouse and it hasn’t been listed or they haven’t put on the on the permitting process that this is going to be used for a data center, but the way that the city knows that it’s a data center is because they have a huge backup generator that’s part of their warehouse plan. Generally, a warehouse is used to store things. That gives light to that there’s probably going to be a data center there.  I think it’s a fine line because you do want to have some disclosure and transparency, but you also can’t jump too far ahead because for a company to say, “Relocate somewhere,” they don’t want to say that they’re relocating until it’s a done deal.

So it is a really fine line for data centers, but that is not to say that there should be more transparency. I think, too, often times as utilities are learning about these things, utilities and cities that haven’t dealt with data centers before. Again, they’re having a hard time figuring out how they can plan out for these data centers. What type of cooling infrastructure they’re going to use and how that’ll impact their water supply. One of the things that we’ve done with working with SRP, which is a large energy and water utility in Arizona, is to develop a multi-criteria decision support tool that weighs a lot of factors.

It weighs different cooling technologies with resource availability in the area. Also, reliability and maintenance of these different cooling systems. So just to give utilities an idea about: if a large data center comes to us and says that they’re going to use this type of cooling, this is an idea of how much water they’re going to be using, or if they have this IT capacity. This is how much energy they’re going to be using. This is how much water they’re going to be using. This is an opportunity that they could probably use for on-site water reuse. This is the type of technology they could use. We’ve been working with utilities and cities on developing these types of multi-criteria decision support tools. Also, they’re a little more educated and have a better idea about these types of numbers and considerations.

[00:58:57]

ELENA RENKEN: Now, we have one last question, which will give our experts a chance to offer some brief takeaway messages. But quickly, I wanted to flag for reporters on the line that you’ll get a brief email survey when you sign off from this briefing. If you have even 30 seconds, your feedback would really help us design our services so that they’re helpful to you. For our last question, in about 30 seconds: What is one take-home message you have for reporters? Dr. Stoll, would you be able to go first?

[00:59:24]

KATE STOLL: Sure, thanks. I’ll reiterate a point I made before, which is that the information asymmetry that can sometimes exist between data center developers and local leaders makes it hard to make informed decisions, and we all have a role to play to close that information gap as the scientific community. I think that starts by knowing what questions to ask when you’re presented with a new data center proposal, for example. Dr. Hickenbottom’s decision tool would be even better; another step closer to getting the information you need for those decisions. But I think our key questions is a good starting point, and it should be paired with local technical expertise and community engagement, to get all the information needed to make useful decisions for communities.

[01:00:18]

ELENA RENKEN: Dr. Hickenbottom?

[01:00:20]

KERRI HICKENBOTTOM: I would say it’s obvious that the demand for data centers and Cloud infrastructure is outpacing the supply, and that utilities and communities are really scrambling to figure out how to integrate these into their network, and that there’s really no one-size-fits-all, magic data center to every type of environment or location. These types of decision support tools and collective actions amongst community members, utilities, and data centers are needed to really come up with a solution that’s really place-based. And I do just want to also say that we need to think at multiple levels, from better chip development to server performance to building infrastructure, and then to community integration. But also, we as individuals have a role and a responsibility in this grand challenge. And we also need to be really responsible of how we’re using data, just similar to how we use our water and energy resources. Cloud isn’t some magical evaporative mystical thing that’s in the air in space. It’s a very resource-intensive product, and so we need to think wisely about how we’re using this resource.

[01:01:47]

ELENA RENKEN: Dr. Keeler?

[01:01:49]

LAUREN WITHYCOMBE KEELER: I’ll go back to my earlier point and just reiterate that it doesn’t have to just be, ‘Yes’ to hosting a data center, or ‘No’. ‘No’ is certainly an option, but negotiations with data center operators also present an opportunity to distribute some of the wealth of the AI economy, and well-developed community benefits agreements that have monitoring and have teeth can make housing a data center worthwhile to a community. They can fund schools, they can fund public works. They can also provide the monitoring and safeguards necessary to make sure that things like water resources are properly stewarded.

[01:02:35]

ELENA RENKEN: Thank you so much, all three of you, for contributing your expertise here, especially on a topic that is affecting more and more areas across the country. And to the audience, I hope we’ll see you all at SciLine’s next briefing. Thank you so much.