HOW AI DATA CENTERS ARE DRAINING POWER, WATER, AND PATIENCE: THE HIDDEN COSTS OF THE CLOUD

E75 HOW AI DATA CENTERS ARE DRAINING POWER, WATER, AND PATIENCE_ THE HIDDEN COSTS OF THE CLOUD

Transcript

Dalton Anderson (00:00.814) Welcome to VGSTEP Podcasts, where we discuss entrepreneurship, industry trends, and the occasional book review. Today, we're going to be discussing a touchy topic to some and to others fills them with optimism. We're going to talk about AI data centers and data centers in general and the compute load that it puts on both the world and especially its local populace.

Currently AI data centers or data centers in general use up to about 1.5 % of the global energy consumption. And by 2030, that energy consumption is predicted to increase to 950 terawatt hours. To put that in perspective, a terawatt hour, and once again, these numbers are just mind boggling.

one terawatt hour can power the whole state of California for one and a half weeks. And so if you just say easy math, 1000 terawatt hours, and you can power the state of California for 1.5 weeks, you get a certain amount of weeks. And then you get to, you just do that to years because you just got to make sense. And it gets you, it gets you to 28.8 years.

So one year of consumption in 2030 is going to be able to power, assuming that California energy consumption doesn't increase here, like that's the control in this theoretical conversation and this analogy or I'm getting lost in my own story here. Basically, if California doesn't increase their energy consumption in the current standard is one terawatt hours, California for 1.5 weeks and then

Energy usage for AI data centers in 2030 reaches to a thousand. You could power.

Dalton Anderson (02:04.3) you could power.

the state of California for almost 30 years, which.

Dalton Anderson (02:15.776) I don't know. It's mind boggling. It doesn't make sense. Like the state of California, that's a big state, big population, lots of homes, the whole state, everything for 30 years, for one year of usage. And then I was curious, like how, how much, how much

how much energy are we predicted to use in 2030? And it was close to 30,000 terawatt hours. So we put it at, you know, we were at a thousand and we are closing into, you know, the thousand mark. So that puts us at double the energy usage of what it was previously. So it was at 1.5 % and now it's, it's coming out to 3.3.

I wonder how much higher it's going to go in 2050. I'm not sure, but I do find it very fascinating how much energy is being used. I knew it was a lot, but when you put the numbers and you write them down or you, you write them out on your screen, you're like, wait, did I, did I add a zero here or, or what's going on? Like, are these numbers correct? I had some head scratchers and I was like,

Okay, let me double check this before I go on the podcast. Like this doesn't make sense. Am I sure this is right? And then I looked at sources and dual verification on different things. I was like, okay, yep, these are correct. what are we gonna do about that? I don't know. So in this episode, I'm gonna talk about AI data centers, their rapid development, their power usage and grid strain, some of the politic, not political.

the, I said political, the, now I lost it, the pollution, the pollution contribution of these data centers. And then also both the construction costs and the water usage costs to the local communities, all of which,

Dalton Anderson (04:30.05) I want to talk about in this episode and then the next episode I want to break down. Okay. Here are the problems. What are we doing about it? And are we making any progress? And are there States or state bills like Senate bills that are coming out that are promising? Are they well received? And if not, why not? And getting a little deeper into the technology piece of, okay, what technologies are being developed?

to make the water usage less and the burn off less or the evaporation. I'll talk about it more when I get later in the episode. So the first thing that we'll talk about is data centers.

planned build out, they want to get to 5,000 data centers. I think there's 3000 and some change planned to be built. One of the most populous places is called data center alley, which is in North Virginia's area. And that area has just seen just a massive growth in data centers and the cloud. And originally before this whole AI workflow was about, was

came about really there is a couple of major cloud providers that I'm sure you're aware of. There's Microsoft, there's Google and Amazon AWS. So there's AWS, there's Google cloud services and there's Microsoft Azure. So basically powers the internet. mean, there's a couple other folks, but those are the three main people, especially in the U S and when you're looking for a data center, you want somewhere that has, I mean, ideally has access to water.

or ease of access to water, and then also good power grid stability, because you need a lot of power. And then one of the other things that you want to avoid is a place that has catastrophic risk or risk for catastrophic events like earthquake, fires, hurricanes, things of that nature you kind of want to avoid. Like you don't want to spend billions of dollars on a build out.

Dalton Anderson (06:40.566) near Miami and then a hurricane comes through floods your data center and you're you're out of, I don't know, $900 million in GPUs. Something insane. So those are the kind of the key parameters to create a data center. A lot of times these places that have data centers built are

of their nature since they don't have catastrophic risk for the most part. They are in areas that have less availability of water or their water scarce areas. Majority of the data center concentration is in areas that don't have that much water available. So that's an important note there. Keep that in your back of mind when we get further into the episode and talk about

Okay, what's the water usage look like? And the water usage is in fact insane. So one thing that is crazy as these cities that I was very surprised about this when I was doing research is that these cities don't have separate policies for the data centers. And there was just this big rush to get the data center people in there like, you know, we'll give you, we'll give you tax write-offs. We'll give you benefits.

We'll give you land, whatever you need, just get in here. so there is some meaningful benefits when it comes to taxes, right? Like Virginia, every year stated that they're getting $1 billion in taxes from these data centers. And so it produces a lot of tax revenue, but there also is different costs that are socialized among the population.

And one of the costs is energy usage. So in 2023, energy usage for Virginia, it's electricity grid. 25 % of Virginia's energy usage.

Dalton Anderson (08:53.838) is from data centers. And by 2030, it's predicted to be 46 % of the total areas energy usage, which I think is quite a sight there.

And so there is a push from local municipalities that they've seen the strain that it's been putting on to their citizens and the citizens have seen the videos and they're like, hold on, we're not, we're not doing this. Like we need, we need some kind of guarantee that they're going to make their own energy. They can be on the grid, but they can't, they can't just take all our stuff and not pay.

which I think is fair. there's a push by certain places like that. If you want to do a data center, then you need to build out your own power. And you've seen articles about Google and Microsoft and Amazon acquiring nuclear startups or nuclear licenses to produce nuclear energy. another issue is that the majority of these data centers, like one third of the data centers in the data center alley in Virginia,

they are within 200 feet of a home. And so 33 % of the data centers in that area are within 200 feet of a home, which is very close.

I was going to convert it for like Europeans, but then I'm like, you know what? Look it up. I don't, I don't know. But,

Dalton Anderson (10:38.446) Yeah. So anyways, so there's a close proximity of these data centers to residential areas. And I think originally that they're just, they're just not zoned as like a large industrial, whereas they're, they're zoned. I don't really get the zoning, right? Like there's this, you're talking about like a 200, no, I gonna say 200.

I mean, you're talking about like a multi-million square footage facility that is requiring like the Colossus data center. They're requiring 300 megawatts of power.

That is the power of a city. It's just really hard to...

Dalton Anderson (11:44.76) example.

example.

All right, so.

Dalton Anderson (11:57.806) Okay, so 400,000 would require a 400,000 megawatts would power 400,000 homes. So in this example, call it 350 megawatts would power 350,000 homes.

Dalton Anderson (12:17.024) So this is not just like some kind of project that they're doing on the side. Like these are massive, massive facilities with large capital outlays, very big construction projects. And I just don't understand the thought process of putting them so close to residents.

Dalton Anderson (12:39.106) And it's not like people live everywhere. mean, there's gotta be some people that aren't happy about it, but I was just surprised that 200 feet is just very close. So, okay. So.

Dalton Anderson (12:55.886) there's this large, large requirement of energy, right? Per data center. And then another thing that's happened recently is that we've got these AI workflows, right? So we've got ChatChipT, Gemini, Anthropic. If anyone even knows Anthropic anymore, I think that it's kind of like slid under the rug except for law firms. These AI...

agents or chats use 10 to 30 times the energy of a Google search. And these data centers were originally used for like storing files, photos, some video stuff, analytics that was high, high compute and cost a lot of energy. But also there's a

lower amount of usage because there's not that many people that are doing those things. There is companies, yes, but it's not like every geosmo is running a massive analytics workflow through the cloud, which is not the case.

And so a large data center, it just puts a massive, massive strain on the utilities. And then another issue that the strain creates both a bottleneck for the data centers to progress and expand with new GPU clusters. And then it puts a strain on the local populace.

So you have two different people that are unhappy. And so what I've observed with Elon's XAI Colossus data center is that they needed a lot more compute to compete with their competitors, but they only were promised 150 megawatts from the city. So they needed more. And as I mentioned earlier, they're at 300 megawatts. And so they purchased these.

Dalton Anderson (15:06.784) or least these like gas turbines, but they're the size of like a bus. They don't look like a temporary structure. They're very big, but they're supposed to be for emergencies and to power. But each of these things are like multi-megawatt turbines. And so they're massive. I'll show the video when we get there.

basically that is creating massive pollution in the local area and people are getting respiratory diseases and or going to the hospital for breathing issues because of the pollution that these gas turbines are creating and the carcinogens that they're putting in the air. So there's that. And then also

There's not a different policy for data centers within their area. They're using up a lot of power. As I mentioned earlier, Virginia is stating that in 2025 data centers used up 25 % of the total consumption for the state.

That's a lot, right? And it's only predicted to grow. And one thing that was being observed in Georgia when I was watching the video about I live 400 yards away from Mark Zuckerberg's data center, which I'll share a clip of, and we'll also watch together the first couple of minutes. It talks about how the power company has the biggest power company has had

six rate increases and also they're not treating data centers differently than they do other corporations like a small business or these different classes of, I was gonna say these different business classes. And then it's being passed on to these different groups like small business owners or residents.

Dalton Anderson (17:21.587) And they're talking about how in peak season, their energy bill has doubled, nearly doubled. They went to 250 to 450.

Dalton Anderson (17:31.586) And that's all because of the usage of the power is just so much, it puts a strain on the utility plants. Okay, so that's the background of energy. The next thing that we're gonna come to is water. And this was something that I was very surprised about. I mean, I knew that it was a lot of water, right? But I had no idea that

a large to mid-sized data center is using 5 million gallons of water a day. 5 million gallons. And 5 million gallons is enough for a town of 30 to 50,000 people. So we're talking about the power of multiple towns. So we're talking about powering 300,000 homes and enough water per day to

be spent on 30 to 50,000 people.

Dalton Anderson (18:35.278) I mean, the scale of these things is just, it's hard to understand. You know, I'd love to see one in person and like see how big it is. mean, Mark was talking about how it's so funny. I refer to him as like his first name. Like I know the guy like Mark, my boy Mark, but Mark was talking about how one of the data centers they have planned or the expand expansion, I think for Georgia is going to roughly be the size of Manhattan. And I was looking that up today.

And I was like, how big is Manhattan? Like that must be pretty big. And it's like 13 miles long and three miles wide.

which like that's massive. Like that's supposed to be a building. I've never seen a building multiple miles long. I don't know. I just, it's just hard to understand. Like I run half marathons. And so you're telling me that I'm supposed to run a half marathon and that's the building.

Dalton Anderson (19:44.463) It's hard to understand. Okay, that's far. That's a long ways. Maybe I should start investing in golf cart companies. Just go all out in golf cart companies, the most efficient golf cart companies to allow your 40 to 50 permanent workers at these data centers to be able to get around. I mean, they probably just drive a car at that point, I don't know.

So five million gallons a day. then Google and Microsoft, Google is at 5.6 billion gallons of water, up by think 30%. And then Microsoft's up 40%, give or take at 1.7 billion gallons of water in 2022.

Dalton Anderson (20:42.072) So.

Dalton Anderson (20:45.944) That's a lot of water. And as I mentioned earlier in the episode is that majority of these data centers are in states that are water strained. 66 % of the data centers are in areas where water is scarce.

Dalton Anderson (21:05.064) And one of the things that I found most surprising about this whole thing, the water usage, I was like, wow, that's a lot. I didn't really know that it was five million gallons a day. Another thing that I thought was insane was, okay, so the whole, let me backtrack. The whole process of the cooling, the...

servers or the GPUs or the equipment is that they take currently they're taking purified water like drinking water. It's not wastewater. It's drinking water. They're taking that drinking water. They're putting some kind of solution in it, like sterilizing the water. So growth doesn't allow for growth like fungus or bacteria, the whole nine yards. So they have this solution. They put it in the water.

and then they run it through.

the equipment and the equipment is so hot that it evaporates the water and superheats it. So 80 % of the water just turns into vapor and just escapes in the atmosphere. And then 20 % of the water is dirty water that is very hot and it's got this chemical solution in it. And so when it goes back to the wastewater plant, it

puts a lot of strain on the utility, the water utility purification company because they've got this massive volume of water they're dealing with and they have to clean it and it's got to get ready for consumption. And before you know it, they've already siphoned off another 5 million gallons for the next day. And then with this 80 % of water that is evaporated,

Dalton Anderson (23:01.034) and these water scarce areas, goes into the atmosphere and the winds take the water vapor and it goes somewhere else. It doesn't stay local. so 80 % of the water that they're using is going to a place that isn't beneficial to the local group. And so I'm going to transition. I'm going to share my screen for a couple minutes about this family who

Dalton Anderson (23:32.532) live 400 yards away from Elon Musk, not Elon Musk, Mark Zuckerberg's data center. And so I'm just gonna play like the first four minutes, I think are pretty important. And then we'll get talk about it. But I think this is a really well done video by more perfect union. And

I appreciated their perspective and thought it was pretty good. So I would like to share it with everyone. And also I'll put this link in the show notes. So if you want to watch the full video, you can, but all right, I'll get started.

Dalton Anderson (28:24.95) Okay, so just watch the first four minutes of that video and it was reminiscent of a video that I'm sure a lot of people saw back in the day. Don't necessarily know what year it was, but when fracking was the big topic where fracking was supposed to be the new thing and it was supposed to be cleaner and it will have as many issues with

environmental polluting or local pollution of the area. I could have said that way better, but you get what I mean. The local effects were minimized by fracking. And then you had these small communities complaining about like my water lights on fire and all these different crazy scenarios. They didn't have any water pressure because the fracking

companies were using so much of the water. And then it was just public outcry and that kind of just died, died down and they changed their, changed their tune, right? And so in that video where in Mansfield, Georgia, they talk about how their water pressure is all messed up and how there's sediment in the water. And so in this scenario,

they're using a well. So I'm not necessarily totally clear on if the data center in Atlanta or near Atlanta in Mansfield, Georgia, that is data center. If that's also tapping into the well water or if it's just strictly from the pollution with the construction that disrupted the well and the local environment. Not necessarily clear. mean, from what I've read, some data centers like to use the aquifers, but

that is typically pretty frowned upon because they use so much water. But who knows? all goes at this point. then another thing that wasn't noted in the video, which is talked about later on, is the cost of their electric bill has doubled. I mentioned it earlier in the episode where normally in peak season it was $250, and now they're saying peak season is $450.

Dalton Anderson (30:49.086) And it has put a financial strain on them because they want to retire. One's retired, one is working. And they're afraid that if he does retire that they're just not going to be in a good spot. So that's something to note. think that was pretty important to hear somebody's perspective, how it's affected them. And once again, this episode is really about the effects.

and not the benefits. And so the next episode will be about the benefits. So I just, I don't want to harp on, I just want to allow people to express themselves or if they've already expressed themselves for me to share that and give people perspective on something that they might not have thought about too much. And then next episode, we could talk about the benefits to the local communities at least. So that was the situation with Metta.

But the situation with XAI is a little different because in this situation, it was more, hey, you know, we had construction, it messed up local environment and our well is not working. And then, you know, it's, it's super loud outside and

Dalton Anderson (32:05.674) at night it's bright, like it's bright all the time. I don't need to have the lights on at night because I can see, because it's massively bright, like crazy bright. The Elon Musk XAI, they're not getting enough power. So for them to move faster, they had to.

use these gas turbines and these gas turbines emit carcinogens, low ozone pollution that leads to respiratory diseases and all this nasty stuff and the pollution can spread up to 20 miles. it's a kind of a tough situation where they want to move fast.

And it's like move fast, break stuff. But in this scenario, you're moving fast and you're hurting people. So I'll share my screen and we will also be listening to the more perfect union. There are two videos are great. Since this person is Elon Musk, it's a little bit more politically charged than

The last video, the last video was pretty neutral. This one is, has a political stance to it. Whereas the other one was more perspective and in voicing, giving someone an audience for allowing them to voice their, their opinion. But also I get that this one's more political because of Elon's, Elon's station with Doge.

which I think he's since disbanded, but I think Doge is still functioning. I don't know what his exact role is, but I know he's stepped away and just focused on Tesla XAI and SpaceX.

Dalton Anderson (34:12.888) So I get where they're coming from and then people are also getting hurt. So I understand that there's people are going to be pretty upset, but just wanted to call that out. If a video started and now I'm to start the video.

Dalton Anderson (34:25.11) Also, I skipped the first five minutes of this video because it just provides a lot of background. But since you're already 30 minutes into this episode, you don't need all the background. I just want to show you the gas turbines.

Dalton Anderson (39:55.424) Okay, so the other half of the video is very, very political, but that section from like the first five minutes to

Dalton Anderson (40:10.67) the 10 minutes of the first five to 10 minutes, like five minutes to 10 minutes, five minutes total. It's a pretty good section where it very informative, talks about the core issue where the lawyer that was speaking in the episode in the more perfect union talked about how he had requested permits from, or they got complaints. And then he's like, that's weird. I don't remember hearing anything about this.

requested permits from the city. The city's like, hey, I don't got anything. And then he's like, So then he goes to the environmental protection agency and he's like, hey, do you have anything? And they're like, no, man, I don't got anything. So then he's like, wait, what the heck? Like that doesn't make any sense. Like you can't have these and not have anything for air pollution. And then the other issue was,

or you can't have these without permits. And then you also can't have these without any protection against the pollution that it emits. It was to put stuff on it to limit the actual pollution that is created. None of that's being done. So no permit, no way to mitigate the pollution. It's just, it's just shoot now pollution and

Dalton Anderson (41:35.17) I mean, the stories from the local groups that are fired up, and rightfully so, they should be, it's pretty devastating to have somewhere that you've lived your whole life.

just become somewhere where you legitimately can't live there anymore. Or if you continue to live there, it's going to adversely affect your health. And then you have folks that are just really old where they're just, Hey, like I've lived here 50 years. I've got five years left. If the air takes me, the polluted air takes me, it takes me, but I'm not leaving. And then you have younger folks like,

They didn't show any kids on there, but I'm sure that there's kids that are, that are affected and yeah, I mean, it's just not, it's just not a good situation.

And there's really no other way to put it. mean, it's not good. And the local government has been pretty silent about the whole thing, given that.

XAI is spending billions of dollars. I think, you know, they're, they're spending just an insane amount of money on GPUs and they're expanding the Colossus. And so the city is just going to, they're going to reap the rewards of the taxes and the benefits of that. And we'll talk about those kinds of things in the next episode. But there's also a lot of people that are being left behind and more extreme in this episode where like people are

Dalton Anderson (43:12.994) legitimately dying from these breathing issues that are developing.

And I think the biggest thing from what I've read from the Colossus is that they didn't tell the local population that this thing was being planned. And one of the main ways that people found out about it was Elon was talking about he's meeting with people in Memphis and doing different things. And so then there was a lot of speculation that he was gonna take on a old building to produce

a data center to stand up a data center because you need a relatively large building to start with. And then you also need a lot of power or original power. I don't think it had enough power. I think it had 12 megawatts, which is a lot of power for a building, but they needed 150 megawatts. So they did the turbine stuff to help them get more power. But then their ambitions kept increasing on how much power they needed. And they only had an agreement with the city.

to be at 150 megawatts. So the only thing that they had to do was just turn on more turbines. And they never got permission to use the turbines. The city never said anything. Nor did they tell the locals that, we're thinking about doing this data center. Here's the parameters. Here's what it's going to do for the community. Here's the cost to the people living in close proximity.

None of that was done. The deal was announced when everything was approved and ready to go and construction already started and people in that area found out about it like on X when Elon was talking about it and like how rapidly they're moving. So it was a real, like a real like rug pull on folks where they weren't aware of it. They didn't know. They didn't have me say in it. And then the next thing you know,

Dalton Anderson (45:09.218) They've got these massive turbines polluting their area and then people are getting sick and dying.

So overall, I wouldn't be too thrilled either. And I'm not sure how well those homes will sell for. mean, they were already in a very polluted area in the first place, like the garbage dump, the power plant, the natural gas processing plant, and one other place, and then XAI. So it's already in a place where there's quite a bit of industrial pollution.

But at the same time, this is just like blown out of proportion and it's way more than the other massive polluters and already a very high.

included area like the, I mean, could, man, that's a rough one. I don't know. I'm getting tired, but.

In a roundabout way, what's going on in Memphis is disappointing and I don't think represents the group. Like doesn't doesn't really align with what it may align with Elon's long term mission, which is to build stuff on Mars and to be energy efficient and clean energy and battery storage and.

Dalton Anderson (46:39.862) living on Mars and becoming a multi planetary civilization.

And maybe this in his mind is a short term cost for long term gains, but the cost is high for other folks. So I think I saw a comment earlier and I thought it was nice is.

These data centers are prioritizing profits and socializing costs. And I that was well put where the local population is paying for these things with pollution, noise pollution, light pollution, and disrupting their daily life or the environment. And then there is

this reaping of rewards for these AI companies where the AI company is getting all the revenue, but they're not paying, they're not paying their, their pound of flesh, the local population with either providing remedies to the pollution that they're creating. They're not paying for the power increases of the power rate increases for everyone in the state. And they're also

paying politicians and making sure that they're well fed so that things are not, you know, when I say paying politicians, maybe I should refuse that and say like power companies or the AI data center companies are paying politicians. And so there's not these add these, these bills where they're potentially malicious towards these different groups.

Dalton Anderson (48:21.718) And so just the local citizens in activism groups, they just don't have a voice. They have a voice, but no one's listening. And so they're just screaming into the void.

Dalton Anderson (48:40.854) In the next episode, we're going to be discussing the benefits of these data centers. we'll talk about benefits and increased technology and different approaches to cooling and being more energy efficient. So we'll talk about those things. But appreciate you listening in today. And of course, have a great day, good afternoon, or good evening. Thank you for listening and listen in next week. Thank you. Goodbye.