What is the Climate Cost of AI?

In the last two years, artificial intelligence has rapidly changed the world that we live in.  If you spend time online, you’ve seen it incorporated into almost every aspect of our digital lives. Search engines, social media, and online tools are all in a race to incorporate AI.

Powering AI are data centers – buildings that house endless rows of servers that bring LLMs, cloud computing, and high-speed internet to our fingertips. But all this power comes at a cost.

Data centers use energy and water – and plenty of it. Until now those demands have been predictable and easy for utilities to incorporate into their forecasts. Now, the new rapid growth and change of AI threatens to break that mold.

WRA has been investing a lot of time and energy into understanding what that cost for the climate could be and solutions to address those costs.

To better understand, we asked some of WRAs experts to explain what’s happening with data centers and what we can do about it. In this podcast, we will explore why data centers are different in the age of AI, how they pose a threat, and what our states can do about it.

On this Episode of Two Degrees Out West – What is the climate cost of AI, and what can we do about it? 

2 Degrees Out West is a podcast from Western Resource Advocates, an environmental conservation organization that's focused on the Interior West. WRA works across seven states to protect our climate, land, air, and water. WRA protects and advocates for Arizona, Colorado, Utah, New Mexico, Nevada, Montana, and Wyoming.

2 Degrees out West is a podcast for advocates and decision makers who want to fight climate change and its impacts across the West.

On 2 Degrees Out West we talk with climate experts and advocates to bring you stories, experiences, and insights from their work in the places we call home.

It is Hosted by Dave Papineau

Want to be part of 2 Degrees Out West?

• Email us with thoughts, suggestions, and interviewees at dave.papineau@westernresources.org.
• Follow us @westernresourceadvocates on Instagram.
• Create a clip sharing "our Bright Climate Future for the West" and send it to us at dave.papineau@westernresources.org. to be included in one of our next episodes.
• Join our email list to stay updated on when new episodes are released and calls to action on how you can help the West.
• ...

Chapters

• 0:00 Intro
• 1:50 What is a Data Center?
• 7:00 The Energy Cost of Data Centers
• 15:30 The Water Cost of Data Centers
• 22:20 The Risk to Consumers
• 25:53 What We Can Do

Transcript

Data Centers Podcast

[00:00:00] Dave Papineau (Host): In the last two years, artificial intelligence has rapidly changed the world that we live in. If you're spending any time online, it's easy to see how it's been incorporated into almost every aspect of artificial lives. Search engines, social media and online tools are all in erase to add in ai.

[00:00:22] Powering AI are data centers. Buildings that house endless rows of servers that bring ai, cloud computing, and high speed internet to our fingertips. All this computing power, however, comes at a cost. Data centers use power and water and plenty of it until now, those demands have been predictable and easy for utilities to incorporate into their forecasts.

[00:00:45] But the rapid growth of AI threatens to break that mold. WRA has been investing a lot of its time and energy into understanding what that cost for the climate could be and how we can address it to better understand that. I asked some of W's experts to explain what's going on with data centers and what we can do about it.

[00:01:04] And this podcast, we'll explore why data centers are different in the age of ai, how they pose a threat, and what our states can do about it.

[00:01:17] Welcome to Two Degrees Out West, a podcast for advocates and decision makers seeking solutions to climate change and its impacts around the west. On two degrees out West, we talk with our climate experts and advocates to bring you the stories, experiences, and insights from their work and the places we call home.

[00:01:33] On this episode of Two Degrees Out West, what is the climate cost of AI and what can we do about it?

[00:01:43] Deb Capal and Stacey telling Tellen are two of W's experts on energy and how we use it. 

[00:01:48] Deborah Kapiloff: I'm Deborah Kapiloff. I'm a policy advisor on the Clean Energy Team at Western Resource Advocates. 

[00:01:55] Stacy Tellinghuisen: And my name is Stacy Tellinghuisen. I'm deputy director of policy development in WRA's Clean Energy team, and 

[00:02:04] Deborah Kapiloff: My work focuses on decarbonization of the power and transportation sectors in the Intermountain West.

[00:02:11] I mostly work with the investor owned utilities in our region. To ensure they're meeting science-based emissions reductions, targets for power sector decarbonization. And I also focus on emerging issues in the clean energy space, including large loads and data centers. 

[00:02:28] Stacy Tellinghuisen: And I oversee our work on building electrification and transportation, electrification and other Emergen emerging issues like data center loads.

[00:02:37] Our broader long-term goal is to ensure that WRA's energy sector across our region is on a path to meet science-based climate goals. 

[00:02:47] Dave Papineau (Host): Deb says that data centers aren't new technology. In fact, if you're listening to this podcast, you're using data centers right now. 

[00:02:54] Deborah Kapiloff: So a data center is the physical infrastructure that supports a lot of the things that we do online, like online banking or streaming or most recently artificial intelligence use cases.

[00:03:07] So what a data center actually is is a structure that houses. Servers for computing to complete all of those tasks in order to complete those computing tasks. There are thousands and thousands of servers that are housed in those data centers, and as they're running these computations, they're using a lot of energy to run those and they're also getting very hot.

[00:03:31] And so there is this cooling component that either consumes water or electricity. 

[00:03:36] Dave Papineau (Host): According to Stacy, there's three kinds of data centers to think about. 

[00:03:40] Stacy Tellinghuisen: So the first are the traditional data centers that run our cloud computing. That is all the processes that Deb described, like online banking and the things that we've been doing for many years online.

[00:03:52] The second is big data centers that are run by hyper scale hyperscalers, google or Facebook. And then the third and the sort of biggest new emerging type of data center are these ones that that advance or artificial intelligence. 

[00:04:09] Dave Papineau (Host): Basically, if it's happening on the internet, it involves data centers.

[00:04:13] We don't often think of our online lives as having a climate cost. It's good to remember the computers that power our internet all actually exist somewhere and they've got to be powered somehow. But we've had the internet and cloud computing for a long time. So why are data centers just now coming up as a concern?

[00:04:31] Stacy Tellinghuisen: So I think over the last couple decades we've seen cloud computing and general demand for computing resources growing. So that growth has. Been slow and steady, and it really hasn't garnered much attention because energy and water utilities can account for that growth in their traditional resource planning.

[00:04:50] But the emergence of AI has really upended that AI is just much more energy intensive than traditional computing. So one report estimated that a chat GPT query which uses AI requires roughly 10 times as much energy as a traditional Google search. 

[00:05:09] Dave Papineau (Host): What is a hyperscaler? You might ask generally what we mean by that is one of the major developers of data centers or AI tech, think Google Meta, Amazon OpenAI, the large companies who own and build the infrastructure that is primarily training and running that AI model.

[00:05:24] Deborah Kapiloff: It's categorically used to describe the really large data centers. So not necessarily like a five megawatt data center, which is actually a huge energy user still, but relatively small in the world of data centers. And what I've seen thrown around for a hyperscaler minimum is around like 20 megawatts in terms of minimum size.

[00:05:43] Stacy Tellinghuisen: So the hyperscalers are all developing these really large new data centers to run their AI computing modules. And because this AI technology has emerged virtually overnight, we're just seeing these astronomical new energy demands in our energy utilities resource plans. And that's really what's new and different about these AI loads compared to the traditional computing loads that we've seen over the last couple decades.

[00:06:10] Dave Papineau (Host): If you wanna learn more about integrated resource planning, you can listen to our podcast all about IRPs or integrated resource plans, which I will link to in the show notes. But the thing to know is that a utility needs to predict with a good degree of accuracy, how much energy they're going to need.

[00:06:25] These plants spell out things like where we're going to get our energy and what the climate impacts might be. They're also made years in advance as AI has exploded over the past years. Utilities across the west are saying they're going to need a lot more energy to meet these demands. 

[00:06:41] Stacy Tellinghuisen: If all of the projected data center loads come to fruition, we could see really extensive impacts in our region.

[00:06:49] And I wanna caveat that I think that's a big if, and maybe we can come back to that later in the podcast, but. We're seeing data center growth in each of our states. So there's huge projected new demands in northern Nevada, near Reno, in central Arizona in southern New Mexico and in Colorado and Utah as well.

[00:07:08] And so at w we're focused on really three key impacts of data centers and those. Three areas of impacts are on first on energy demands and greenhouse gas emissions. The second key impact that we see is on water resources in the region. And then the third key impact that we're concerned about is customer costs.

[00:07:28] And that's primarily the impact of these data center loads on electricity customers. And I wanna also note that these are the areas that WRA is focused on, but there's other impacts of data centers that we shouldn't ignore. So data centers have noise and land use impacts, and artificial intelligence can exacerbate existing biases or discrimination with impacts on equity.

[00:07:53] So I think those are really important considerations as we broadly look to manage and regulate artificial intelligence and data centers. But WRA is really focused on those energy, water, and customer cost impacts.

[00:08:19] Dave Papineau (Host): So AI uses energy. How much power are we talking? 

[00:08:22] Stacy Tellinghuisen: We looked at the largest eight utilities in WRA's region to try and understand like what are they forecasting in terms of energy growth. And what we found is that the collective annual energy demands of the utilities are projected to be 32% higher in 2030 and over 55% higher in 2035.

[00:08:42] Relative to today's levels of energy demand. And this is really significantly higher than the forecast that utilities were making just a few years ago. I think it's important to note that these data centers are what we would describe as high load factor customers. So they use a energy around the clock and that means that they have really high energy demands, but they also contribute to peak demands.

[00:09:07] Dave Papineau (Host): And what is peak demand? 

[00:09:10] Stacy Tellinghuisen: Peak demand is when the energy utility is working to meet the highest level of demand across their system. So it's when it, for most of our western utilities, it's when every household and every business has turned on their air conditioner. At, mid to late afternoon.

[00:09:26] It's when people go home and they turn on the lights and they start to run a load of laundry. So all of that used as electricity and it's. What contributes to a peak demand is when all of those demands are stacked on top of each other, so they're on net high at their highest. 

[00:09:44] Dave Papineau (Host): Our electric grid is really quite a logistical accomplishment.

[00:09:47] I think it's helpful to understand how complicated it all is. Utilities have the challenge of figuring out how much energy you're going to need and then meeting that. When we need less energy during the day and nobody's home, they have to produce less. And when we need more energy, like when we're running air conditioning or when we're all at home in the evening, they have to produce more.

[00:10:06] Utilities are pretty good at keeping up with us. The utility has certain resources that they use to meet based demand, the lowest amount of energy we need. And on top of that, they have additional energy sources to ramp up or down with the energy needs of the grid. Peak demand, like Stacy is talking about, is met by power sources that ramp up or down with energy needs of the grid.

[00:10:26] Stacy says that peak demand is particularly important to consider with the data center conversation. 

[00:10:32] Stacy Tellinghuisen: And so when we talk about peak demands, again, this is what contributes to in large part to the overall system costs, and those are the costs that all customers pay. So when we looked across these eight key utilities, what we found is that they're forecasting their peak demand will be.

[00:10:50] Almost 10,000 megawatts higher in 2030, and almost 17,000 megawatts higher in 2035. That's a lot of new energy or capacity that's needed to meet it. Demands 

[00:11:03] Dave Papineau (Host): and just how much is a megawatt, for example, could your residential house ever use a megawatt? Deb says, absolutely not. 

[00:11:10] Deborah Kapiloff: So your house could really never use a megawatt of power.

[00:11:15] It would be multiple households. It's around 500 to I think like 750 ish households, depending on their energy use would be a megawatt. 

[00:11:25] Dave Papineau (Host): So when we talk about megawatts, think about the amount of power that an entire neighborhood might use. 

[00:11:29] Deborah Kapiloff: When I go to an ev fast charger, the ev fast charger I use that can fill my car from 20% to 80% in 15 minutes is 150 kilowatts.

[00:11:40] And you would need a thousand kilowatts to equal a megawatt. So if you think about seven or eight EV fast chargers, that's a megawatt. 

[00:11:49] Dave Papineau (Host): 17,000 megawatts is nothing to laugh at. And it comes with a climate cost 

[00:11:53] Stacy Tellinghuisen: and it directly translates into higher emissions. So there's some states in our region like Colorado that have good policies in place that limit the impact of new loads on emissions.

[00:12:04] But others don't. And just as one example in Northern Nevada, NV Energy filed an integrated resource plan in 2024, and a resource plan is just the utilities effort to say, here's what we think are all the sources of new loads, and here's what's gonna happen to our demand over the next 10 to 20 years, and here's all the ways and the strategies we're gonna use to meet that demand.

[00:12:29] So NV Energy, the biggest utility in Nevada filed an integrated resource plan in 2024 and for their northern utility, they forecast a staggering 53% jump in emissions in 2030 relative to the plan they had filed just a couple years before that. That's over 3 million tons of additional greenhouse gas pollution in 2030.

[00:12:53] That's a lot and I think that's really underscores. And of why we see data centers as such an existential threat to meeting our region's climate goals. 

[00:13:04] Dave Papineau (Host): Stacey also says that because data centers influence peak demand, but also base demand, they represent a unique challenge when considering load growth.

[00:13:13] Stacy Tellinghuisen: We're seeing load growth in many different areas, but I think it's really important to clarify how data center loads are really different from other types of loads like beneficial electrification that we're seeing. And so WRA is really working to promote electrification of buildings and transportation because it's a key component of meeting our science base and our long-term climate goals.

[00:13:37] Dave Papineau (Host): Deb explained just what beneficial electrification means. 

[00:13:40] Deborah Kapiloff: Beneficial electrification is taking end uses that would otherwise be served by fossil fuels. So things like driving your car or heating your home and transitioning them to electric beneficial electrification is pretty flexible in terms of load is you do have this flexibility of when you're charging an electric vehicle with some of the electric heating technologies.

[00:14:06] If you're able to do demand response actually with your heat, it's an asset to the grid in some ways because that load can be moved around to avoid peak times and to avoid being a cost causer during peak demands. 

[00:14:20] Stacy Tellinghuisen: Electric vehicles can be charged in the middle of the day to help us integrate more solar loads.

[00:14:24] Electric heat pumps typically run at their highest level in the wintertime when electric utilities grids are underutilized. 

[00:14:31] Dave Papineau (Host): Stacey says one big issue with data centers is that they're just not flexible, 

[00:14:38] Stacy Tellinghuisen: but those new loads are really different from data centers. So data centers, again, typically run around the clock.

[00:14:44] And so they, they are not necessarily nimble in a way that they can help us better integrate solar or wind resources or respond to peak demand periods. So those two types of load can help reduce customer costs 'cause it helps us use the grid more efficiently. And so those are, that's how they're really different from these emerging data center loads.

[00:15:12] Dave Papineau (Host): The energy cost of data centers makes sense. Computers use energy. Data centers are in some ways just big computers. They use a lot of energy, but if you've ever run an old pc, you might know that when computers work hard, they get hot. All that energy going into data centers creates a lot of heat.

[00:15:34] Lindsay Rogers: My name is Lindsay Rogers. I'm the policy manager for Municipal Water Conservation at Western Resource Advocates. I sit on our Healthy Rivers team, and my work is really focused on advancing urban water efficiency and the integration of water and land use planning. 

[00:15:51] Lindsay is one of WRA's experts on water.

[00:15:54] Dave Papineau (Host): She says, alongside energy demands, data centers also threaten our limited water in the west. She says there's two major ways that data centers use water. 

[00:16:03] Lindsay Rogers: Data centers use water in two primary ways. They consume significant amounts of water on site or directly to keep their servers cool. And then they also use an even greater amount of water offsite for power generation.

[00:16:18] So our report estimates that by 2035, the total consumptive water footprint of data centers across our region. Could equal almost 90,000 acre feet per year. And that's the equivalent of about 270,000 households, roughly the size of Las Vegas. So these are big numbers that we're looking at. Around 20% of that water would be used on site again for cooling, but the other 80% would be used for power generation.

[00:16:49] And that's why the collaboration between our two teams, healthy rivers and clean energy is so important and we're working so closely 'cause it presents some really complex water and energy nexus questions. I think that's particularly well illustrated when it comes to cooling technologies for data centers.

[00:17:08] Dave Papineau (Host): We're going to talk about how data centers use water for electric generation in a minute. One thing you might be wondering is how do data centers actually use up water consumptively in cooling their computers? As I was learning about this, I was envisioning something like a liquid cooled gaming pc, which works more similarly to the way an engine vehicle is cooled liquid runs through a closed loop system to keep mechanisms cold.

[00:17:31] This wouldn't necessarily result in a lot of watery loss because it's a closed loop system. Deb says Data centers usually work a little bit different. 

[00:17:39] Deborah Kapiloff: It's equivalent to a residential swamp heater where in order to cool that air, you are using water to cool it. And so some of that water is evaporated.

[00:17:49] What we're seeing in the majority of data centers is that the air that is around the computing equipment is cooled as opposed to having liquid that is cooling the computing equipment itself. 

[00:18:04] Dave Papineau (Host): If they use large evaporative coolers like Deb described, that's a consumptive use because the water is evaporating to cool the air in the data center.

[00:18:12] Funny enough, Lindsay says that non-consumptive cooling you might be familiar with from your car or gaming PC is something WRA is excited about. 

[00:18:20] Lindsay Rogers: It's more of an emerging technology, but we're starting to see it more commonly in data centers where the servers are basically directly cooled through a liquid, which could be water or a cool in 

[00:18:31] Deborah Kapiloff: there are.

[00:18:32] Innovations in that and that some data centers are moving towards liquid cooling which is really exciting. 

[00:18:38] Lindsay Rogers: But typically the systems that use the, these technologies that use less water are gonna use more energy and then the cooling systems that use more energy will use less water. So it's a pretty clear water versus energy trade off there.

[00:18:53] Dave Papineau (Host): Now we get back to how electric generation uses water. There's a contemptive use involved in electric generation two that needs to be addressed. 

[00:19:00] Stacy Tellinghuisen: Water's used in power plants in a couple of different ways, but the biggest way is actually is a cooling system. So a traditional coal plant will burn coal.

[00:19:12] That coal, the heat from that is used to generate steam. That steam runs through the turbine and the turbine generates electricity. Then the power plant needs to cool and condense that steam so that it can be reused. And power plants typically reuse steam five or six times before it's, it becomes too salty or mineral laden for them to keep using.

[00:19:35] So in the process of cooling that steam, most power plants use water. They use like essentially a wet sponge to cool off the steam. And so some of that water evaporates and that's where most of the consumptive use of western power plants 

[00:19:51] Lindsay Rogers: is. But even if we hypothetically decided that water conservation was just more important overall than energy con conservation, which I'm not saying it is it really wouldn't be.

[00:20:02] As cut and dry as saying, we're only going to recommend dry cooled water efficient systems because those dry cooled data centers could actually end up having more overall water demands because they require more electric generation. Which in turn requires more water. A 

[00:20:19] Stacy Tellinghuisen: traditional coal plant might use five or 600 gallons of water for every megawatt hour of electricity it produces.

[00:20:26] A combined cycle gas plant is more efficient and may only use a couple hundred gallons of water to generate electricity. And then solar, wind, typically use no water at all. 

[00:20:38] Lindsay Rogers: So it gets a little complicated here, but it makes the solutions trickier.

[00:20:46] Dave Papineau (Host): In addition to the complicated relationship between energy and water saving. Lindsay says The places where hyperscalers, like meta Google or OpenAI are building these data centers further complicates things. 

[00:20:58] Lindsay Rogers: And the other threat we're seeing on the water side is that for a variety of reasons, data center development tends to be clustered.

[00:21:06] So we mentioned, there's lots of projected new data centers in the Phoenix Metro area and in northern Nevada. And that means that the impact of that additional 90,000 acre feet of water isn't gonna get spread out evenly across states and water providers. The demands are gonna be higher in those specific regions, and then because of that, we're likely going to see greater risk to water security in regions that experience much of that growth.

[00:21:35] Dave Papineau (Host): We covered the complications with increased load growth from data centers, and we're starting to understand some of the complicated landscape around water use. A third area of concern for WRA is the risk to you and the increases you may see in your electricity bill. 

[00:21:49] Deborah Kapiloff: Right now the utilities in our region are seeing these huge projected demands, and all of that is happening in a process that's called load forecasting, which is essentially when the utility submits their projected electricity demands based on a variety of factors.

[00:22:06] In order to get approval that's the agreed upon number for the amount of electricity that the utility is going to need to provide in the coming years. And then based on that approved number, they figure out how to ensure that there's the generation resources that would be needed to provide that amount of electricity.

[00:22:27] Dave Papineau (Host): Any new power sources of course costs something. If utilities build a bunch of new electric generation, somebody's got to pay for it. 

[00:22:34] Deborah Kapiloff: Data centers really throw a curve ball into this because they're such huge customers with such large amounts of uncertainty tied to whether or not they're actually going to materialize.

[00:22:46] And so when you think about these really large customers, if one of them. Ends up not coming online, not building a data center somewhere, but generation resources to produce that power have already been procured. And that has been approved, then that's something that existing customers are going to have to pay for.

[00:23:06] And so there's this element of financial risk that is being assumed as these loads are being forecasted and approved, and resources are being procured to serve them because. Whether or not existing customers have to pay for resources that aren't utilized, isn't something that's necessarily addressed in the earlier part of that process.

[00:23:30] Dave Papineau (Host): New data centers, lots of new demand, a need for new generation, but it's not guaranteed that demand will actually be there in the future. Where does that leave consumers? 

[00:23:40] Deborah Kapiloff: There's a process called cost allocation, which essentially looks at the total costs of running the system and then divvies them up into kind of slices in a pie for different types of customers.

[00:23:52] And the principle behind that is really that if you're a certain type of customer, like a residential customer or a commercial customer, you should be paying a rate that reflects how on average that type of customer contributes to costs on the grid. And we're really concerned that the cost allocation frameworks across the west are not set up to account for data centers and how they are causing costs on the electric grid because they're requiring huge investments, not only in additional generation resources.

[00:24:24] But additional transmission build out as well as power, quality and additional resources that are needed to maintain grid stability. And so if those costs aren't divvied up in a way that really reflects all of the costs that data centers are causing. Everyday folks could be on the line and could see their bills go up.

[00:24:50] Dave Papineau (Host): Data centers are really resource intensive. They use a lot of energy and they use a lot of water. We've worked hard to reduce greenhouse gas emissions and pollution in the west. With our help states like Colorado and New Mexico have set aggressive climate goals that help put them on a track to a cleaner.

[00:25:03] West data centers represent an enormous threat to that progress. Both our energy and water systems aren't set up to easily absorb this much change. The West has gotten hotter and drier. It's hard enough to figure out how to manage our limited water supply without adding in any extra demand. Key rivers in the West, like the Colorado, have reached their breaking points.

[00:25:22] As states struggle to find solutions to managing the dwindling water supply, we need to figure out how to reduce demands, not increase them, because our energy grid is a precise, finely tuned machine. Large power users like data centers are disruptive. Rate payers like you and me could end up on the hook for expensive new additions to the grid if data center demands fail to materialize.

[00:25:44] But we wouldn't be sharing this with you if we didn't have some ideas on how to deal with it. Lindsay, Deb, and Stacy say that across the board there's policies we could advocate for to help address and mitigate disruptions of data centers. 

[00:25:59] Lindsay Rogers: On the water side we see three opportunities or, and the first is really starting with the basics, and that's having more transparent and consistent data across the board about the direct and indirect water demands of these facilities.

[00:26:17] That would include metrics like water use effectiveness. So a measure of the number of liters needed per kilowatt hour of electricity. It would include the type of source water. So is this is this facility gonna be pumping groundwater or are they using municipal potable supplies? Do they have access to recycled water?

[00:26:41] What are their peak water demands? So something of an equivalent on the water side. We wanna know the impact of the system on the hottest day in August and how that will affect a utilities infrastructure needs. And then we wanna know, what are these facilities. Implementing when it comes to water efficiency.

[00:27:03] What cooling technology are they using? Are there other design features that they're using to reduce demands? The second recommendation that we'd like to see is really more innovation in the industry. We'd really like to see we there's a lot of re financial resources and expertise in, in AI and the industries behind these data centers and would really like to see them come to the forefront and and put resources behind figuring out how to design data centers to be as efficient as possible with both their energy and their water demands. And lastly, we'd really like to support cities and counties and water providers too.

[00:27:49] Better understand the suite of policy tools available to them to help inform where and how these data centers are developed. 

[00:28:00] Dave Papineau (Host): Stacy says there's also ways to incorporate clean energy and emissions reducing technologies into our future at data centers in the west. 

[00:28:06] Stacy Tellinghuisen: At a high level, we think either data centers or the utilities that serve them should deploy clean energy and maximize energy efficiency.

[00:28:14] So there's a couple of interesting models out there. One is a clean transition tariff, which is a program through which a, an electric utility could develop a new clean resource for, specifically for a data center. And we've seen. That deployed in Nevada where NV Energy is developing a geothermal resource alongside Google.

[00:28:35] So that's a really interesting model. We think it's most advantageous if data centers are used to, or these clean transition tariffs are used to advance new or emerging clean energy technologies. There's other methods of developing clean energy on behalf of data centers.

[00:28:51] So you can allow data centers to dev, develop behind the meter, clean resources behind the meter. Means an energy generating resource that is essentially owned by the customer. So in this case, by the data center. So often it's a solar system solar panel system. 

[00:29:09] Dave Papineau (Host): But large customers like data centers don't always use clean energy for behind the meter generation.

[00:29:14] Stacy Tellinghuisen: I think one thing that's important is to ensure that those behind the meter resources are clean and it's not just diesel generators or gas turbines because those will further exacerbate any existing and contribute to local air pollution and greenhouse gas pollution. So I think that's really important.

[00:29:31] And then of course we want data centers to maximize their energy efficiency as they're being developed. And. Ideally participate in demand response or load shifting programs so they don't contribute to electric utilities peak demands. 

[00:29:48] Dave Papineau (Host): Stacy also says, in addition to working with demand response data centers could use their behind the meter generation to help diversify the grid.

[00:29:57] Stacy Tellinghuisen: One, one important consideration I think as we develop these policies is that in certain states, in our region and certain utility systems, we're starting to see that the systems are getting saturated with. With solar, for example. And so we wanna make sure that if a data center develops a new solar resource, it doesn't end up just curtailing solar that's already on the system and serving other customers.

[00:30:22] I think ideally a data center would develop a solar resource paired with battery storage or a wind resource paired with battery storage so that it is providing energy to that data center that's. That's closer to matching the data center's actual load profile. 

[00:30:38] Dave Papineau (Host): And then how are we going to protect rate payers from floating the bill?

[00:30:41] Deb had some strong suggestions there too

[00:30:47] Deborah Kapiloff: for consumer protection recommendations that W Ray has. One of 'em is pretty simple. It's just having really strong contract provisions for data center customers. When you have customers of such a large size that have a elevated risk profile in terms of being a pretty emerging industry and being such a large customer having those provisions in place can be really important.

[00:31:15] And some of those provisions can be just minimum contract links. The customer agrees that when they come online with the utility that they will remain a utility customer for X number of years. Another really important contract provision can be minimum billing provisions, so ensuring that even if in a certain month.

[00:31:33] If the data center doesn't necessarily have what would otherwise be billed at a certain level, they are paying that level to ensure that the utility is able to recoup the costs that it invested in bringing that customer online and having a system that's capable of serving that customer. Another really important contract provision can be having financial collateral requirements and exit penalties.

[00:31:59] So that if there is a situation in which the data center company goes bankrupt or anything like that, the utility is not the one that's eating the costs of. Investments that it made to be able to serve that customer. And similarly, if the customer decides that they want to downsize and use less capacity, but the utility has already invested in giving them a certain amount of capacity, it's again, really important that those costs are born by the customer that is producing that risk versus the other customers of the utility.

[00:32:32] That were existing customers. 

[00:32:34] Dave Papineau (Host): Deb also says that we can revisit economic development policy that's often outdated in the context of data centers. 

[00:32:41] Deborah Kapiloff: Additionally, something that can be done is looking at economic development rates. So what an economic development rate is essentially a discounted electricity rate.

[00:32:52] These were generally premised on the idea that if you had a really large electricity customer coming into town it was maybe going to promote really significant job growth. When you think about like a steel mill or a manufacturing facility, those were things that policymakers thought were worth giving a discounted electricity rate too, because they had co-benefits for the community.

[00:33:15] But the way that those rates are structured is often that if you have a certain amount of electricity demand, you can qualify for that rate. And so we're seeing data centers qualify for and take service on those rates, despite the fact that they don't have the same type of economic benefits to the community in terms of employment figures.

[00:33:34] And also they're driving a huge amount of. Costs in terms of the investment needed to serve them. So having policymakers really revisit the statutes that created economic development rates and decide whether they want to amend them so that these data centers do not necessarily automatically qualify just by virtue of being really large energy users.

[00:34:02] Stacy Tellinghuisen: I think I just wanna underscore that. All of these policies that WA is recommending, I think are imminently doable and they're really essential for ensuring, again, that we achieve our climate and clean energy and our water goals. 

[00:34:18] Dave Papineau (Host): Stacy pointed out that data center developers have the ability and the responsibility to think about taking care of the communities they build in, 

[00:34:25] Stacy Tellinghuisen: and I think it's important to remember that.

[00:34:29] Particularly these large AI data centers are really being developed by some of the world's. Wealthiest corporations, so Google and Amazon and Facebook and others. And I think this is an opportunity for good policy to, to help leverage these big corporations to advance our clean energy and our Clean and healthy Rivers initiatives.

[00:34:52] So I think, we need to think about how we can leverage the capital that these corporations bring to, again, to advance clean energy and advance healthy rivers. 

[00:35:03] Deborah Kapiloff: On the flip side, we are extremely concerned that data centers potentially causing cost increases in electricity could be detrimental to our goals for transportation and building electrification because that essentially is making the price of those fuels go up.

[00:35:21] And making the economics of switching to an electric vehicle or to a heat pump less attractive.

[00:35:32] Stacy Tellinghuisen: Yeah. I think that's one of the key challenges in, in how, in wrestling with these data center loads. They're really uncertain loads. And so what we have seen is that a hyperscaler may put in a, an interconnection request or request for electricity service in a couple different utility jurisdictions because they, their goal is to connect to the grid as fastest as possible, and so they may shop around for the best deal or the fastest interconnection.

[00:36:04] What we're finding is that not all of these interconnection requests actually materialize, but because this is such a new industry, it's really difficult, I think, for utilities and decision makers and stakeholders to understand, what's real in, in these load forecasts, what are we actually going to see?

[00:36:23] And that's critical because as Deb. Talked about earlier. If we overbuild the electric system and these data centers don't materialize, then it's the remaining residential and commercial customers that are likely to be stuck footing the bill. And so I think that's a key piece that many of our recommendations try to address is how do we get more certainty around what this load growth actually is.

[00:36:48] Dave Papineau (Host): One of the most important things WRA is trying to address is the uncertainty around data centers. They're certainly disruptive, but Stacy says they also come with significant unpredictability. 

[00:36:58] Stacy Tellinghuisen: The challenge here is that these data center loads have emerged overnight essentially in the, particularly when you look at the timeframe that is typically involved in electric utility and water utility planning.

[00:37:11] We really need to act quickly, and that's the challenge here. Policy makers at the state and local level really have the tools to manage these new impacts. And so I think that's the critical piece, is that, we need to take opportunities where we have them. So what we can't do is 

[00:37:28] Lindsay Rogers: Wait for maybe our, what we would typically prefer in the energy and water industry and take our time to really, spend five or 10 years.

[00:37:38] Identifying and formulating these strategies like the policy makers need to move fast if they're gonna keep up with the growth in this industry, 

[00:37:46] Dave Papineau (Host): data centers ultimately are largely going to be regulated and addressed on the state level. It's state regulators and legislatures who are best positioned to address these threats.

[00:37:55] Deborah Kapiloff: State public utility commissions are the regulators of investor-owned utilities, and so they are uniquely positioned to lead on this. And then additionally. State legislatures can di give direction to those state public utility commissions on actions they'd like them to take, or considerations they'd like to bear in mind in this new paradigm of AI load growth.

[00:38:18] And so this really is an issue where we are going to see action on the state level.

[00:38:36] Dave Papineau (Host): Data centers pose a unique and fast-paced threat to climate progress in the West. If you've listened so far, thank you. WRA has been leading state level climate action for over 35 years in the West in 2025. Our commitment and resolve has only strengthened despite the challenging landscape for advocating for climate policy.

[00:38:56] We are still seeing winds in the west on the state level, in the places that we live and love working to make sure that data centers don't disrupt our progress is a key part of our plan and vision for the future. If you wanna support that plan or work to protect, the West is supported by listeners like you, people who care about the places they live, people who are smart and educated, and know that states have the power to take climate action into their own hands.

[00:39:22] You can support our work at the link in the show notes. Even something small $5 a month matters. Think of it as a cup of coffee for the climate.

[00:39:37] If you wanna stay up to date on our work data centers in the future, sign up for our email list and follow us on social media for the latest news and progress in our region. Lastly, we'd like to thank our sponsors who are enable our work in the West. Our Champion sponsor is First Bank. Our signature sponsors are Denver Water Torch, clean Energy, Scarpa, goco, and Southwest Energy Efficiency Project.

[00:40:00] Our supporting partners are BSW Wealth Partners. Meridian Public Affairs Group 14, engineering and Kind Design. Thank you for listening.