Why the Modern Data Center Is Forcing Communities and Policymakers to Rethink Infrastructure

This is This is straight straight straight out of Crumpton with your host, Greg Crumpton. Good afternoon, everybody. Welcome to another episode of straight out of Crumpton. I'm Greg Crumpton coming to you today from a absolutely gorgeous day in the beautiful state of South Carolina where I get to hang out most of the time, unless I'm not here, and then I'm somewhere else. So but today, I'm coming to you with a very special guest. I, as you guys know, love the skilled trades, and I spent a lot of time talking about them, talking about the future of skilled trades, how to recruit the younger generation into the skilled trades. And particular vein is the data center world And how are we building data centers? How are we staffing, maintaining all the good stuff that goes into data center infrastructure? And as you all know, you can't read a story. You can't read an article. You can't watch the news without something being on the topic of AI, on the topic of data centers. You hear about, you know, the next data center in space, underwater. We're building them everywhere, folks. So today, very excited to bring Julia Chang on, and we are going to learn a little bit about her. But she is an associate professor of sociology at the University of Maryland. And if my brain serves me well, those are the terps. Right? That's correct. Yeah. That's correct. Letter. Letter. Letter. Letter. Alright, professor. Give us a little bit of insight. And as my buddy Mitch Jewel of Encanada says, who are you, and what do you do? Thank you. Well, thanks for having me on here, first of all. It's just a great pleasure. So I'm a professor of sociology. Always, like, basically study, like, institutions and groups and group behavior, group dynamics. My previous work, I was really interested in basically land use in China. So I studied factories. I studied the construction and real estate industries in China. I went there. I talked to managers in those industries and workers in those industries. And as you know, like, I mean, today, as you just said, I mean, the big thing today is data centers in the US. I mean, I just saw this crazy statistic that the latter half of twenty twenty five, seventy two percent of our GDP was actually powered by data center adjacent or data center related investments. So, I mean, it's it's AI. It's it touches every industry. So, I mean, naturally, I was fascinated by this. So I started attending data center conferences. I interviewed people in the data center industry, such as yourself. That's how we met. I got in touch with you through another industry person and and collecting Let's call her by name because she'll be mad if we don't because she's the queen of of the Carrie Getz. That's correct. Yeah. She's a such a human connector. So, Carrie, thank you. And if you're not listening to this, shame on you. So go ahead, doc. Yeah. I mean, she's a really long term expert in this now booming field. But, yeah, I mean, it's it's just a fascinating thing because if you just look at the news, I mean, data centers are hugely controversial. If you look to the media, this is very binary picture laid out. You know? Like, the New York Times, most of reports just about, like, super secret hyperscalers and private equity and, you know, like, really shadowy capital at the top that's secretly funding these very power hungry data centers. They're gonna cover Nevada in solar panels, you know, like, very sensational stuff. I mean, it sets up this big tension between residents. Nobody wants a data center in their backyard. Nobody wants their electricity electricity prices to go up, particularly because we already have pretty pretty, you know, rising energy prices today. And on the other hand, you know, there's, of course, like, you know, most companies do use cloud services, services, and the data centers for those cloud services have to go somewhere. Right? So, I mean, I think the reportage the media reportage on this topic, it's kind of it it's binary. It lends itself to this kind of civil society conflict between big business or tech big and the people, basically. But, you know, that's not actually how a lot of infrastructure development in the past was laid out. If you look at, like, our electrical grid, for example, our cell towers, for example, yes, there was there were kind of these incidents of tension very famously, I think in, like, the I don't remember how I think in the Dakotas, the early days of the electrical grid build out, there are farmers who'd go around and burn down, you know, transmission sites and stations. Why? Because, you know, they were afraid of it. They were afraid of displacement, you know, and they're afraid of this new technology that looked scary. There was a lot of media reports about fires and stuff like this. So it's a little similar to that, but but there's a lot of big differences too. Like, one big difference is, you know, electrical and self tower infrastructure. Like, it it's less secretive. It's it benefits locals. Right? You actually need a cell tower near you. Maybe you don't wanna live right under it because you don't want the radiation or whatever people are scared of, but you do wanna be in some kind of proximity to a cell tower so you can use your cell phone. Right? But people don't see data centers the same way. Right? I mean, yeah, people see data centers. It doesn't really matter. I'm just streaming on my so invasive in everything we do. They have to live somewhere. Yeah. They have to live somewhere, and increasingly, they do have to live near us. Like, if we're gonna if Waymo takes off and we get self driving cars everywhere, yeah, you need an edge data center nearby, and it has to be evenly dispersed, especially high density, highly populated cities. So you will need this infrastructure near cities in order for our, you know, everyday top technologies to work. Yeah. You know, I find it really interesting. I was in fact, I was with Carrie, and we were at a data center convention in I wanna say we were in Sterling, Virginia. And there was a protest out out front. Nimbe, you're not in my backyard folks, you know, picketing, essentially. But they were posting on social media from from their their their walking around with their signs. They were posting that, and I thought, how ironic that the thing they're protesting, they're giving voice and giving daylight to it via a data center. Right? Yes. I thought that was super ironic. I mean, it's a it's a tough call. Like, I understand their concerns. I mean, these things do use up a ton of their superpower hungry. They don't generate a ton of jobs once they're up and running. Mean, the construction generates a ton of jobs as you yourself have noted in a lot of your other pod podcasts. But, yeah, I mean, once they're up and running increasingly, you could even have AI powered data centers, which don't require any human labor at all. So I I do understand the concerns. I think that, you know, like okay. So one of the problems with this is that the most highly cited concern of people who oppose data centers is well, there's environmental concerns. They don't want the pollution, etcetera. But then there's also basically, like, the perception that it'll it's gonna drive their energy rates up. And that, you know, that is problem. I do wanna say, though, that and this is less reported in the media. Effects on energy prices vary, right, across regions because of variation in how energy markets work. So if you're looking at, like, Virginia, Virginia has the most data centers in the country, something like this is an old statistic, but something like seventy percent of the world's Internet traffic runs through Loudoun County. I don't know if that's exactly the case anymore. The the the secondary data center markets, new regions like Pennsylvania, Georgia, Texas, they're getting built up pretty fast, but and also internationally. But, anyway, yeah, Virginia, it is true that, you know, energy prices have gone up. And in those places, there's a really opaque pricing strategy by kind of a vertical utility integrated utility. So, basically, it's you know, the utility market is highly regulated. In order to build out new transmission lines, basically, the utilities company pays for all these new transmission lines, and then they put out RFPs, request for proposals for power producers to generate more power, and it's really a highly coordinated process. But, of course, the utilities company is at the center of it. They control all of that. So there's a perception that they can just rake up you know, hike up residential energy rates whenever they need to in order to cover infrastructure. And if that infrastructure is mainly being used by data centers, then, yeah, I mean, there is this problem in a perception of kind of data centers free riding off of residential rate payers. But I have to say that's particular to Virginia or in other places where you have very highly regulated energy markets. In other sites like Texas and Pennsylvania and Georgia, they have less well, it varies, but there's generally less regulated energy markets where, for example, data centers can bring in their own power. They can build out microgrids, their own pipe you know, independent power producers right on-site. And in those states, for various reasons, they're actually allowed to then sell the power they produce back to the public utility. And so there's less of this free rider problem, actually, in these more deregulated markets. So that's something that not really generally seen in the media. I don't have I'm still working on data analysis to figure out what's the actual effect of data center development on residential rate payers or commercial rate payers in those regions. But I have to say, it really is affected. It's it's highly structured by the energy policy in different regions. Yeah. I I've thought about this a fair amount. I think this will lead to more deregulated states is what I mean. Because, my opinion only, this is a way for utilities to take the burden. And and listen. I think a lot of it's perception. Once you I I can't wait till you get your report published where we can really look at numbers, but there's a whole lot of perception because people don't understand all the time between commercial rate, residential rate. I mean, there is a difference when commercial rates are charged, peak power consumption. I mean, there's a lot that goes into that metric. Yeah. But the perception is and and, really, that's what matters in life. Perception is is reality that we, as as residential users, do pay for that. I think that to to restate what I said, it's going to lead to more deregulation, which is going to lead to more entrepreneurial based thinking of how we generate power for these sites. Yeah. And there is a tremendous amount of stuff going on right now in that whole world of power production, whether it's small modular reactors, whether it's Google buying three mile island. I mean, there's so many different ways of trying to, you know, attack the perception and while doing so, a real solution for power consumption because the data centers do suck in a lot of data or I'm sorry, do suck a lot of power. But it it's not like we are going to go backward. We're we we gotta figure it out because nobody's gonna say, you know what? You're right. Let's go back to, you know, the flip phone and no Internet. No. People aren't gonna do that. They're gonna move forward. Yeah. I mean, absolutely. I mean, the old, pricing strategy for electric u utilities didn't really take into account these large scale data centers and users. And, yeah, I mean, you it is I I understand when activists are scared of big tech, big private equity funding in these spaces, but, I mean, you just named a couple of these very highly innovative technologies that are actually being paid for by big tech and big private equity. There's, like, you know, SMRs, nuclear reactors in Susquehanna, Pennsylvania. I think Amazon entered this nice deal with Talon. I mean, there's this lot of different innovations, not to mention geothermal and any number of new battery technologies for storing, renewable energy so it's less intermittent, can be used for data centers. A lot of this takes big money, and it can't just be paid for by residential rate users. So, I mean, there's a second way of viewing data centers, which is that this is a way of getting big money into our infrastructure and paying for these pricey upgrades. I mean, that's kind of the actual explicit, policy line of Alaska exactly actually. Like, the state of Alaska is facing this big energy crisis where I forget the name of their natural gas reserve. They have one natural gas reserve that all of the residents in this state, which are concentrated, their population is mainly in the southern region, and that's a pretty convenient location for them to draw gas from because it's, you know, it's close, but it's gonna run out in twenty twenty twenty eight. So the big push is try to get data centers in the state because these are the people who can these are the corporations that can pay for a new pipeline that's gonna take go up to the northern regions and pull out new natural gas there. So, yeah, I mean I mean, I it's in Alaska, I mean, I can't speak to how successful this strategy is. A lot of data centers can't go to Alaska because of the latency. It's just too far from everything. But at the same time, yeah, this is, like, a a hidden strategy is to try to get big money into electricity markets to update our grid. And, mean, I have to say, this is really different from the previous infrastructural moment. Like, if you look at electrical grid build out, we did have, you know, private sector money getting involved, but private sector money, when we built out our electrical grid, was only interested in powering cities because that's where the money was. Here, we are actually facing a situation where big money wants to go to Nevada, wants to go to like, Iowa is actually a big site for secondary data center markets. You know, these are sites that In rural Iowa. Exactly. Yeah. So, I mean, these if you are a local government and you're looking at your tax revenues base, you're thinking, yeah. I mean, true. It does not generate a lot of jobs. Fine. But you're gonna get a lot of sales taxes. You're gonna get and most of most of the time, the taxes that these corporations pay, these develop developers pay is is basically sales tax on huge IT purchases. And those IT purchases actually far outstrip whatever they're paying for power. I mean, for a hundred megawatt data center, you know, the initial capital expenditure just on initial hardware alone, it's me it's like over a billion dollars. Right? So And the refresh rate. I mean, we're refreshing gear every three years now. Exactly. Yeah. Especially as, like, GPU technology, you know, changes. I mean, the Moore's Law kind of works in favor of local governments. Right? It means that they're gonna have to buy and upgrade all their hardware very frequently. So, I mean, I get it. I think it's a it's actually a pretty good strategy for these rural communities. Well, I mean, it's something that they can basically sell because they don't have anything else. Yeah. Right. They've got they've got land, and if you can so you gotta have land, you gotta have power. So if you got land and you can generate power or bring in power, then you're a win win. You see what prices are in Ashburn, Virginia to buy an acre of land. I mean, jeez, Louise. So you gotta get creative. That's right. So you touched on one point. These data centers do not perpetuate a lot of employment after they're built. That is true. Yeah. But during construction and and if you look at a lot of these campuses that are getting built, these are five eight year builds because They're gonna build, you know, five half million square foot facilities. And they're gonna pipe them, and they're gonna plumb them, and they're going to wire them. And that takes a lot of people, and that is so good for the local economy when you're building this campus. That's right. So couple of things I'm curious about. Does a campus spawn another campus locally? That's a a question for me. I think so if you look at how things are getting built. Totally agree. What does that or maybe a question. Is your study looking at the local economic impact during construction and post construction? Yeah. That's those are great those are great points. So first, yeah, I mean, they do cluster. These are big developments that require some redundancy. So if you are a cloud service provider, it does benefit you to be other near other cloud service providers. Not to mention the infrastructure, there's shared infrastructural costs, that can be minimized if you are locating a ton of, you know, cloud service providers or just any kind of data centers that require interconnection. If you can hook up the fiber to one, you can hook up the fiber to a lot. So they do tend to cluster. In terms of employment, I mean so at first, I I do wanna highlight, like, I think another thing that gets that I've been very, really, pleasantly and optimistically surprised by, actually, is the number of people in the data center industry who are concerned with these kinds of questions. You were the first person I talked to, actually, in the data center industry who's really concerned with how can we make how can we bring jobs to these local communities? How can we make sure that data centers actually provide kind of a social infrastructure in addition to a digital infrastructure? How can they maybe they don't generate jobs in the long term, but how can we make sure that they're actually contributing to communities in a meaningful way? There's an I mean, I talked to Allie Burton who she's a consultant, and she's been going around looking for I don't know who she is. Yeah. Right. I mean, she was fantastic. She was talking about, like, how can we design these deals? I mean, these big hyperscaler deals so that hyperscalers are required to retrain the employees that they're going to be putting out of business through all of this technological advancement? Like, for example, you know, Amazon's building out all these data centers, but can they retrain their warehouse workers so that they can be employed within data centers as well? Like, I think that's have a program to do so. That's really that's great. Yeah. They do. Yeah. They they have a program. So folks working in their distribution and fulfillment centers Can move over to the AWS side of the house if they want to learn how to do that. That's great. Yeah. Yeah. I mean, you've mentioned this too. Right now, there's a huge shortage in people with expertise to build at the data centers. Right? And so, basically, there's a very limited pool of electricians, of HVAC, people who have the expertise necessary for data centers, and that expertise is very specialized, and it's changing every day. Right? So you really need people who are on the bleeding edge of, you know, building out these really huge high density compute data centers. Right? The number of people who are trained in that are very small. And so what happens is companies just bid up prices on these on on these experts. Like, they'll fly out the same expert from one site to another because the necessary, you know, time and investment for training new people is actually prohibitively high. Like, another problem in in training data center experts is that, you know, you might wanna you might be interested in training out an ordinary electrician to become a data center electrician. But how do you have them practice? Right? You can't actually give them it's a huge liability to have an intern working on an actual operating data center. You know? I mean, these data centers can't be shut off for even just a microsecond. Otherwise, disaster ensues. Right? So that's another problem is how do you design the training facilities to get to retrain people in the trades into data center adjacent or data center involved expertise? So that's another thing that people like you and Allie and a a great number of people in the industry are talking about is is how can we create modular or kind of, like, model data centers that we can train interns on so that people can be retrained into this industry? Well, think about a doctor. A doctor that is in residence somewhere is doing the same thing because Yeah. You can't, like, practice on fake sick people. You you have to practice on and practice being the keyword. You that's a real life. And Yeah. Look at what they go to go through for that training to make sure that they don't have an outage and and that being somebody not living. It's an outage for a person is a lot different than an outage for Netflix. But same principle, to your point, it's hard to get that real life training because building a data center, there's no load on it. So you you get familiar with the layout. You get familiar with redundancy and understanding the different levels of redundancy. That's one thing. But once you go live, that's a whole another thing of understanding outage windows and maintenance windows. Excuse me. So that's a real that's a that's a challenge that I think through AI and VR, we can simulate a lot of things to help train people. So just Yeah. One one more thing to think about while you're being a associate professor of sociology. No. Absolutely. I mean, I think that some of I mean, there are some limits to this. Right? I mean, data centers don't actually require that many people on-site once up and running. That is true. On the other hand, they can come contribute to local communities in other ways. Right? We used to I mean, I have to say, they they are they do try to do this. Like, hyperscalers, like Amazon, Google, when they go into local communities, they do have, you know, huge teams of people who are trying to figure out how can we generate good relationships with local community. Some of the ways that they do this, though, I do think suffer from a little bit of, like, tone deafness. Not to be pejorative, but, you know, I saw somewhere that I forget which of these hyperscalers. One of them says, okay. We're gonna we're gonna find a girls in coding camp for for young kids in the area that we're, loading our data center in. But and that's nice, but it's just that seems like a mismatch in what those communities need. And, how many young girls are living in this these rural communities that are actually interested in coding? Like, probably a small number, but these are rural communities. Right? The the the kind of amenities that they're trying to provide don't seem to really match up with the needs and concerns of local people. Skills required. Yeah. That's right. I mean, there are there are more I think it's an evolving field. Right? So there was another hyperscaler, I forget which one, that recently announced that they're going to they've entered into this, very innovative pricing, so they're gonna try to subsidize energy costs for the members of the local community that they're building in. Things like things like that seem to be a little bit more of a direct contribution that it's it's speaking the same language as the people the concerns of the people who live in those communities. Well, nobody knows yet how this is going to what what all is going to work because I think we're so new into the the problem that we have to try a lot of different stuff and see what sticks and then lean into what works. And because we you know, it's like I was talking to my wife and about an ailment I had, and she said, well, you've never been this old before, so you don't know how to handle this. It's kind of the same thing. We've never been in a situation where we had to solve for these problems. So Yeah. There is going to be a substantial amount of try it, and and I'm a kaizen practitioner. I'm all about trying and improving and then measuring and then trying and improving and measuring until you figure out the formula. And these aren't going away. That's that's my what my main thing that I tell folks is we this is a a long term problem. This is not okay. We're gonna be through building data centers in seven years. That's right. We're going to be evolving the data centers. Right now, I'm I'm working with a fellow on redesigning a data center that I helped install in Oh, wow. Nineteen ninety nine. So what do you do with these legacy colo facilities that were built for relatively low density? Because that's what that's all we had. We had three kilowatts per rack. Well, now the latest NVIDIA machine is is two hundred kW in a rack. So how do you how do you evolve Architouch infrastructure rather to support older properties that need to evolve? That's another tribal knowledge thing that we have to extract from people who know how to do that And train the younger generation to go into these facilities, let alone building new ones. So it's so long term interesting. Figure out. Well, let me ask you. So what is what is the biggest difficulty in a big retrofit like that where you're trying to update a nineteen ninety one old data center, old nine. No. Nineteen ninety one. Okay. Sorry. Sorry. Well, what what are the big difficulties in updating? It's just power. Right? Okay. Yeah. Because most of those designs were built for a hundred watts per square foot. Right now, that's simply not even close to anything. So let's say you got five thousand square foot data hall, and it was built for a hundred watts per square foot. That is the power that the developer asked the utility to provision for. And you can't just go back and say, hey. We need to our our power increased have increased ten times, so we need ten times the power. They simply don't have the infrastructure locally built. Transformers, transmission lines, underground utility feeds. It's just not there. So the challenge is, I think, of taking a older facility and figuring out how to make it a hybrid facility so that one third maybe is high density, and then the remainder is legacy low density equipment. Maybe it's, you know, not maybe it's cold storage. Maybe it's something that doesn't take a massive amount of power. But you're still able to rent that square footage out to a consumer. So So this would minimize the power upgrade you would need. You'd power up just that one section that's high density. The rest of it can remain. Yeah. So you have to just figure out the the what's available. Yeah. Can you buy out the rest of the building's leases so that you can have their power? You know? There's a lot of dynamics to figuring out old or older properties because there's a I mean, we have, you know, what, two thousand some odd colo facilities that were built during the relatively close to the dot bomb era. Yes. So a lot of that a lot of those properties are aging out. So what do you what do you do with them? That's interesting. Yeah. I've been so I've just been looking up a lot of these old, dot com boom era facilities. And, yes, what you're saying does make sense because a lot of times, they'll get bought out. You know, there'll be some merger and acquisition of the data center developer space. And always when they're retrofitting and upgrading these centers, they're always buying out neighboring sites. So that Yeah. That does make a lot of sense. That's interesting. It does interconnection, like how long it takes for the public grid to approve them and hook them up to the public grid. Does that also create delays for these kinds of upgrades? Yes. If it's available. I mean, if you Yeah. I I know several buildings in the Charlotte metro area where my business was, they were pretty much tapped out of the local availability just from a a big, you know, big wire, thirteen five distribution from the pulp, you know, out on the street. They they were just done. So that that's the kind of stuff that I think is so interesting, let alone getting down to the rack level. So Yeah. There's a lot to figure out there, but a lot of opportunity flipping back over to the AI conversation. Everybody doesn't need a hundred kilowatt rack. Okay? Sure. Yeah. But if you watch the news, you would think we all do. You would think everybody needs a liquid cool NVIDIA, whatever the latest is, a four hundred stacked, you know, to the ceiling. You don't need that unless you're doing real high computational stuff. Right. There's so much legacy equipment that will continue to utilize air cooling. We're still gonna have the crack units on the perimeter. All of that's still going to be in effect for eighty five percent of the market. That hyperscaler stuff is such a niche market. Yes. It's huge right now, but there's still a tremendous amount of what we would call legacy gear And and equipment and infrastructure that's still going to be used. My email server and my banking operations do not need a liquid cool Grace Hopper chip. Okay? They just don't they need, you know, five k w in a rack, and I'm good. And there's a lot of that out there that I think people lose lose track of. Yeah. That's right. That's why it's so important, I mean, to to get voices I mean, to for your your voice to be heard. I mean, people who are closer to the ground who understand the actual material, like, you know, what the technologies are, you know, you see a lot a very different picture than what's in the media and this kind of thing. I mean, it's interesting because I I do think, yeah, what's in the media basically just amplifies the hyperscaler segment of the market. Everyone's concerned with what they're doing. But the colo is not going away anytime soon. If anything, like, these smaller kind of less high profile segments of the market, like Edge is huge, and that's just gonna keep growing. Right? Colo is not gonna go away. I mean, I'm actually interested in what happens like, the cloud the cloud service provider market is really just AWS. Right? I mean but but colos and cloud are those sort of, like, complementary markets, or does the growth of cloud take away from colo, or has it in the past? Those are questions I'm still I'm still trying to answer. Yeah. Well, I I think they're complementary in a way. I think that one it's kinda like I I like to use this term, a surgeon needs a patient, and a patient needs a surgeon. Everybody can't be a surgeon. Everybody can't be a patient. So there has to be a little yin yang in the equation because you simply don't need that for everything. You talked about edge, especially edge five g and six g for cars, for everything that's going to work in the five g spectrum. You five g doesn't go far in a in a very high high quality way. So, yes, we're gonna have a whole lot more edge, but it's gonna be in weird places. It's not gonna be like we have historically thought it was gonna be in a nice little square room that's purpose built. It may be stuck in the elevator room on top of a building, and we gotta figure out how to solve for that. It it may be on the side of the road in an enclosure Like, we have now for the traffic control signals. There's just so many different variations in where you can where we will put edge product, and nobody talks about that. We talked about we talked about nine hundred gabillion gigawatt, you know, something in the middle of a cornfield. Right. Not what my local utility is pulling underground out here right now. You know? You gotta it's a big it's it's a local and a big picture topic. Yeah. That's right. That will the edge data centers, I mean, my impression is that they just need to be I mean, you just put it somewhere near a substation. Right? Anywhere near an electrical substation would be great. And, they're small. You can kind of fit it inside a a moving pod or even. And so, yeah, they get a lot less press. But I feel like that's the fastest growing market simply because of all of these technologies that we're developing as well as the fact that edge combined with cloud means that you can stream, you know, your Netflix with really low latency no matter where you are. That's gonna be yeah. And and look at, if you will, when you're driving, and I spend a fair amount of time driving because I like to drive if I can. The number of miles of fiber that we are currently pulling is incredible. These big people see them on the side of the interstate a lot. These big rolls of orange tubing or yellow tubing. That's all that's the conduit for the fiber. So when you see two yellow pieces of conduit coming up out of the ground together with nothing attached, that is soon to be something. It's going to be an interconnect. It's going to be an edge spot. It's going to be something. And if you start noticing all the fiber pull that is going on, this this country is being wired underground in such a way fiber wise that Most people don't grasp what it takes to pull fiber in an existing urban environment. And if you look at the the repair work that has to be done as careful as they are Yeah. They still rupture stuff because you're underground boring. Yeah. And it's a beautiful industry, and thank god that people are willing to do it. Yeah. Yeah. They've gotta work around a ton of existing infrastructure. The fiber itself is extremely delicate. Yeah. And it's so many strands. I mean, this is very complex. Well, mode multimode fiber to me is just incredible. The amount of data that you can shoot down one glass strand, and we're doing a bundle, you know, three quarters of an inch of glass. It is crazy to me. I like I'm an air conditioning dude. I can't think like that. So I'm just glad that you're technology. It really is kind of an amazing moment we're living through. And it's interesting because I so my husband is a is in is in math, and so he's a little bit closer to people who work on the software side of this stuff. And if you talk to the people who are actually working in various, you know, AI startups who are coming up with new algorithms, coming up with new new ways of of machine learning, none of them think this is a bubble. And it's so interesting to me because, you know, if you talk to the if you just read the media, everyone's afraid of an AI bubble coming, etcetera. But people who are close to the industry are all saying no. I mean, even if there is you know, like, you know, Chattypitty is just one one form of text based learning. There's there's all these new things that the AIs can do. Like, we're heading for kind of a Star Trek age where you will go to the doctor, and the doctor will be a robot. And they'll immediately scan your body, and it will be completely, you know, like, high accuracy, low involvement, etcetera. I mean, I I do think that there is some kind of mechanism by which, you know, small startups have to blow a lot of hot air in order to get funding. And so maybe you're just getting a lot of that hot air runoff when you talk to those people. But it does feel like if you're just trying to get a barometer for how bullish we should be on AI, the more of those AI people you talk to, the more of a true believer you become or the closer you get to Silicon Valley. You know? A lot of those people in that Silicon Valley bubble are extremely, you know, true believer about this stuff. Look at that topic, what you just said. Elon Musk will tell you that shooting rockets into space and selling cars are nothing more than a means to fund his robots, and that's where we're headed. I mean, Tesla will be a robot company long after it's a car company and long after he does cool stuff like landing the rocket back on the, you know, on the barge. All all of that is to fund the r and d that is going into building his robot vision, which, I mean, it's already here. You can you can buy a freaking robot right now if you want one. So I feel like we're like, so if you look at it in a baseball game analogy, there's nine innings. We're like the first batter. We're not even in the first inning good yet. We're just on got the first guy warming up. And there's so much more to go. And Yeah. Compound knowledge, Julie, is as you know, once you learn something and then you apply that knowledge to the next problem, you get the solution easier. And then you compile that knowledge and put it on the next problem. It's just a compounding effect, and AI is a genius of that. And I use Grok. That's mine because I'm a I'm a Musk I won't say fan, but I'm intrigued by him. I like Grock. It that's my my AI machine of choice. If you just look at what it does with a question that you keep improving upon the question, that you keep getting a better and better answer, And we just started. I mean, can you imagine what this is gonna be in five years when, I mean, my neuro chips implanted, and I have a thought in a cheeseburger up here? I mean, how cool? Yeah. Yeah. I mean, it's fascinating. It's fascinating. It's fascinating. It's terrifying for people too. Be right? Because a lot of this, people are thinking, okay. Like, it's gonna lead to more inequality. Right? Because the fastest way to commercialize artificial generalized intelligence is to sell to corporations for cost saving, for labor saving. So they can fire half of their accountants and replace it with the platform. And so but, I mean, that comes with every technological advancement. Right? If you look at the Industrial Revolution, a bunch of textile workers were put out of business, and a lot of them went around destroying looms. Right? So we're kind of in that early moment of kind of Refined Teslas. And right. Exactly. Yeah. So, I mean, it's interesting. I do think there's a lot of complex distributional issues to be solved. Right? Like, should we be talking about universal basic income? Or even just at the more small scale, how should we we you and I talked about this. How can we ensure that across all energy regional markets, data centers are paying their fair share and rate payers aren't subsidizing the energy cost of data centers? Now, of course, that's a complex question. Right? Because, like, right now, I can use ChatGPT for free. Why? Because they're using my data to train for it. Right? They need this. It's kind of beta mode. Right? But at the same time, nobody is thinking, oh, well, I actually should be paying higher residential energy prices so that I can subsidize ChatGPT. Right? That's not in the public conversation. Now I'm not saying it should be, but I am saying that this is something we all use. Right now, how we understand energy costs of AI is just filtered through this. There's big tech, and they're putting billions in it. And one day, this AI is gonna put you out of a job. Right? So we're very much in that same language of, like, I wanna burn down the loom, the mechanical loom, because it's because I'm scared of it. Right? And that's how energy price talk is being handled as well. I'm I'm such a non scarcity mentality person that I don't go there because if that were true, then we would still be chiseling on on granite tablets notes instead of writing on a computer, or we would still be in a horse drawn carriage. I mean, we have to evolve. And in order to evolve, you have to figure out what's next. And then our country was built on figuring out what's next and how do I support that effort. Okay? I mean, that people get hot. They don't wanna be hot. So I'm in the air conditioning business. Somebody had that great idea to do that. People are they want automobile, so I'm gonna start an automobile service company. So we have to figure out how to support whatever is in the current domain. And I I just I don't have I don't know if people worry about their privacy, but they've got Alexa listening to every damn word they say at their house. Yeah. Alexa, I promise you, has already scooped up your your credit card number you ordered something, your social when you were talking to the doctor. They got all that stuff. So Yeah. Yeah. Yes. We should be cautious, but not scared in my this is my opinion only. And it's just the way I think and operate is is not from a place of scarcity, but from a place of abundance. And I think that we all will do better if if we try to adopt some of that mentality and embrace it. And, you know, to your point about paying for it, I love this topic. You don't drive your car or or ride your motorcycle without a tag. Your tag pays for the right to or the ability to repair the roads, maintain the roads. Why would we not do that for our infrastructure use? Why would we not want to help the burden? Because I'm not I'm not anywhere close to a free loader. I hate free programs because nothing's free. Somebody is paying for it. Now it may not be you. It may be me through my taxation of a product or whatever, but nothing's free. Everything has a cost. And, you know, whether however you wanna start distributing tax burden, we still have to pay to maintain the infrastructure of fiber, and and we've never had this much fiber. Right. Is that gonna be strictly at that company's expense? Yeah. Or is it like oil distribution? When we burn natural gas, you pay a pipeline fee hidden in there somewhere to Colonial Pipeline or whomever to maintain their infrastructure. So there's a lot to talk about on on nothing more than infrastructure and and cost of maintenance, so to speak. Yeah. That's a great point. I mean, that that, the vehicle registration fee is a really great point. Most people don't think about that as an infrastructural tax. So why don't bicycles pay it? They're they wanna be treated like a vehicle. So if you're gonna ride your bike on the road, there should be a tag on it because you're using the public driveway. Yeah. That's right. It's tough. Was hated when I say that, by the way. It's tough because you basically have two different corporate models for doing things. Right? So the power industry is very much about this infrastructure, and the infrastructure is paid for by the public. Right? That's how the electricity industry runs. But then then you have things like the software industry. Right? Software for AI is still running on the same beta mode kind of model where you offer Google for free. Everyone can use a search engine for free. The price you pay is that your data is helping to train the next generation. Right? And so I think that's what leads people to have this idea that, yes, I should not be paying for the power costs of the AI data center. Right? But I should be able to use Cheetah and, ChatGPT and whatever, you know, products that these these tech companies come out with, I should be able to use that for free. At least among the younger generation, I think I feel like when I talk to my students, for example, nobody's concerned with privacy. Right? It's older people who are concerned with privacy. No one's concerned with my data is being used out there for profit oriented stuff. There's this great book by a Harvard Business School professor about how the new capitalism is that your data is being sold to private companies, right, and that that data is now powering algorithmic capitalism. And that that's that's an that's a very compelling argument. But if you talk to twenty year olds today, nobody cares that their data is being used for this. Right? Convenience is everything. But the flip side of that is, like, yeah, if your data is being used for free by these companies, and then that's that's its own economy. Right? We see I give my data to Google, and they give me these free services, but nobody is thinking those free services have an electrical bill attached to it. So who's gonna pay that? Right? And that's, I think, part of the tension between why data centers are so controversial is it's the clash of these two different infrastructural models, I guess. Well, think think about the additional cost of a chat GPT search versus that of a Google search is five times as much. So it costs five times as much for for the same question asked on either platform. And That's right. If you're gonna ask stupid questions like is Mariah Carey dead, Use Google. You don't need ChatGP for that. You're burning extra and and maybe that's just a public awareness campaign. Yeah. Making sure that you're using the proper tool for the proper query because Yeah. It really is a huge difference in in operating expense when you when you use, chat versus just a a dumb search. I completely agree. I completely agree. I mean, in general, I do think there needs to be some kind of public campaign for awareness around the fact that every you know, these online activities are not free. Right? Like, you are incurring some kind of carbon funded, carbon fueled cost that every time you do one of these searches, and people should think about that. Right? The know, these are these companies offer it for free to use so they can use your data, but at the same time, the data centers are running on full power. I mean, they're actually straining to, as you say, like, fulfill every stupid query people put through. Yeah. I mean, so this idea that we should be cognizant of, you know, the actual infrastructural cost of our online activity is I think that would go a long way in kind of, deescalating this very tense antagonism we have between, you know, the private big tech companies that are taking advantage of us and then public residential users who are paying for data centers, for example. I mean so I do wanna just end with with saying that right now, like, the public discourse around data centers, it actually has a big effect on policy. Right? We have this black and white picture of residents hating data centers. Right? And what it leads to is actually really antagonistic policymaking around data centers. In Texas, for example, there's a lot of people who who want data centers. Right? There's a lot of local officials who actually want that revenue and that economy that the data centers are gonna bring. Right? So they're giving away a ton of tax incentives to big hyperscalers to build out these data centers. But on the other hand, there's kind of this schizophrenia among policymakers at the on the other hand, but they also signed into legislation senate bill six, which is, you know, this very kind of like, well, but we don't want residents to pay for data centers. We don't want data centers to drain our state of electricity. And SB six actually gives ERCOT, like, the authority to just pull the kill switch on data centers every time there's an emergency. Right? So this leads this real big data center to public society, civil society kind of antagonism, where data centers, you well, first, roll out the red carpet for them, but then another governor comes in and then puts a kill switch on them. And it's it's kind of this tough way, very binary way of dealing with data centers, right, where policies themselves are really, on the one hand, really nice and preferential, on the other hand, very antagonistic. And that reflects the public discourse, I think, of I think I believe that's spot on. I really do. And I think you'll see the rise in BYOP, bring your own power, is going to be ever more, in conversation of how yeah. You can come, and we'll give you the tax break. But how are you going to generate your own power? Right. And I think that's gonna be a a really cool conversation. Julia, I know we could go for a while longer, but you have professor stuff to do, I'm sure. And, really, really do I just really appreciate you taking time to join us today and talk about this because I know it's such an interesting thing because it's like a a an octopus. You know? You start on one conversation, and then eight tentacles later, there's so many pieces to it. You know? It's a very, very ever moving mosaic. To try to it's like a puzzle, a mosaic puzzle that's that's moving. And, trying to hit it in all the right spots is is going to be a challenge, but really appreciate you bringing daylight to a lot of it, for our listeners and and for me and giving us good stuff to think about and to, kinda noodle on of how this affects us locally and and, you know, at large as a country. Thank you. As ever, I'm indebted to people like you. I mean, I learned so much about this whole industry from people like you who are much closer to it and to these really kind of big, you know, seismatic shifts that are happening to our society. So thank you. I I really learned a lot as always. I appreciate you coming and sitting. Folks, if if you wanna learn more about Julia, she's on LinkedIn. Check her out. She's always out there poking around and prodding and asking good questions and posting some some good content to make you, you know, kinda make you use your noodle a little bit, which is what we all should be doing and not settling for what we hear, but finding out facts and making your opinions about what you read and understand, not what somebody told you or what some influencer told you on TikTok. So get out there, do your homework, and be a smart consumer. Sound like Dave Ramsey all of a sudden. Don't know why. That's right. Yeah. But I I really do appreciate it. Check her out. Julia Chang, LinkedIn, associate professor of sociology at the University of Maryland, the the Terps. And if you wanna learn more about me and my stuff, it's dreg crumpton dot com. You can also check out two books I've got authored and published called Deep Knowledge and deeper knowledge. So you can check out both of those at Amazon, Barnes and Noble. And if you wanna read about them before you buy them, you can go to deep knowledge dot m e. Deep knowledge dot me. Read all about it. So thank you for taking time with us today, and we will look forward to talking to you on the next episode of Street Afterthought.