The IT Breakroom Episode 2: The Power Grid Wasn’t Built for AI

Episode Summary: On this episode of The IT Breakroom, we’re joined by Neila Wilson, Director of Data Center Engineering & Operations at Visa, for a conversation on the full lifecycle of modern data centers and the growing push toward sustainability across the industry. We discuss how data centers evolve from initial site selection and construction to long-term operational efficiency, along with the challenges surrounding power, water usage, AI growth, and the U.S. energy grid.

About the Guest: Neila Wilson, Director of Data Center Engineering & Operations (East) at Visa, brings decades of experience across telecommunications, hyperscale infrastructure, and data center design and operations. Starting in the U.S. Air Force and building on her background in electrical engineering, she has worked across data center design, operations, and software development and integration at organizations including Meta and AWS before joining Visa. She has also led a non-profit and mentored entrepreneurs. Throughout her career, she has helped shape how companies connect technical innovation and AI in large-scale environments with sustainability, advocating for more intentional design and stronger alignment between data center growth and community impact.

Key Topics Discussed in This Episode:

00:00 Intro & Breakroom Question
02:38 Neila’s Background
06:52 Data Center Design & Site Selection
12:08 Enterprise Data Centers vs Hyperscalers
18:26 The Data Center Lifecycle Explained 26:36 Sustainability Opportunities Across Power & Energy
31:17: Water Usage & Cooling Systems
34:46: The Labor Challenges Behind Data Center Growth
39:07 The biggest opportunity for sustainability: Renewable Energy & Power Strategies
46:27 AI Growth vs Power Grid Limitations
51:21 The Future of Data Centers & Energy Infrastructure

Watch on Youtube:

Listen on Spotify: Click here

Episode Transcript:

Meredith: Thank you for joining us on the IT Breakroom. Super excited to have you today. I’m really excited to get into your experience in general. You come from a great background — all the way from data center design to the operation side, even software integration. You just have a wealth of experience. So we’re definitely going to talk about your experience and where you are today and what you’re passionate about. But I do want to intro with a little bit of a more personal question — our break room question of the day. I’m interested to know who your favorite musical artist or band is.

Neila: Yeah, I have been thinking about this, and there are so many I love, but I was raised up on Prince, so that’s like my all-time favorite. But as far as bands that I love to see in concert, Train was really awesome in concert and they have such fun songs. That’s who I love to go see probably a lot. But there’s so many great artists, and it’s so hard to pick just one.

Meredith: It is hard to choose. I will say Train is one of those bands they play on the radio all the time, and they do have some really catchy music.

Neila: Yeah, they are a lot of fun. You’re not gonna not smile at one of their concerts. It’s a good time.

Meredith: I’ll have to check that out. That is the first time I’ve heard that as somebody’s favorite, so that’s fun. And in the spirit of May 4th coming up, I do have to ask if you are a Star Wars fan.

Neila: I am a Star Wars fan. Now, I’m not as much of a Star Wars fan as I am a Trekkie, but I do respect that Star Wars was kind of like that first movie I really got into sci-fi-wise when I was about seven years old when it first came out.

Meredith: Nice. My goal is to watch the newer Star Wars this weekend. I have not seen the new versions yet, so we’ll see.

Neila: Yeah, I was a little heartbroken when Han Solo’s own son had killed his father. I was a little heartbroken over that.

Meredith: Yeah, that’ll do it. All right, well, let’s get into a little bit about who you are, your background, and then we’ll segue into our discussion today about data center sustainability, community impact, all of that good stuff.

Neila: Yeah, great. Well, I started off a long time ago without expecting to necessarily be in data centers today. In terms of electrical engineering, fiber optics was really what fascinated me initially — that drove me towards electrical engineering. I happened to get an internship early on with a telecom company that translated into my military experience in the Air Force, where I really solidified my telecom background at a test base for comms and new communication testing. From there, when I came out into the engineering field, it just continued to build on design of fiber optics and co-locations and points of presence. Originally in the telecom industry, we didn’t call them data centers — we called them central offices, co-locations — but they were housing network and data connectivity. And eventually they grew into spaces that were focused on actual data transmission.

Working with Sprint originally, that’s where we had one of the first significant data centers designed, and I was responsible for the fiber optics and then later managing all of the data center operations, NOC organizations, and so on. Eighteen years later, when Facebook reached out to me and asked that I consider an opportunity with them, I was kind of surprised — what does a social media company have to do with data centers and telecommunications? I was not quite making the connection. Once I understood the opportunity, I was really blown away at the scale of a hyperscale data center and the opportunity to come in and lead their global engineering organization responsible for the design of the connectivity infrastructure.

Then going on from that into some of the other aspects of their engineering, design engineering, and construction components of the data centers, and then stepping into AWS’s design and being able to see how the energy components were really a key portion of how we had to think about data centers going forward — beyond just the bandwidth and capacity within the data center. Physically, we had to start really being much more conscious about how the power consumption was impacting, especially at a place where they’re building hundreds of data centers on existing campuses and new campuses.

Just between Meta and AWS, it gave me an opportunity to really see what the global footprint looked like for data centers, what those impacts would be, and a lot of considerations of how we had to work with local utilities and how we had to really rethink our designs, especially once AI came onto the scene. All of those experiences over 30-some years got me to where I’m at at the moment — just really enjoying the opportunity to share a lot of my own experience with people and helping as we, as a community around the United States in particular, where we’re seeing a lot of new build impact. Helping communities understand what that really looks like, understand why AI, what that direction is, that’s changing the landscape, and how we can really consider a sustainable approach to data centers in our communities.

Meredith: That’s awesome. I’m super excited to get your perspective because you’re really on the ground in the data center. You have a well-rounded wealth of knowledge from all different aspects of it. Before we talk about the sustainability aspect, I want to get a full picture of the data center life cycle and what that looks like from design all the way to operations. When you’re starting from conception, is it like, “All right, let’s find a piece of land we need to be on,” or, “Let’s design this and then see where this fits”? How does all of that work from that very starting point to deployment, ready to go and we are operating?

Neila: Great question. There was a time when we really just wanted to find a piece of land, and we would take whatever we could find, and then design something to fit within the parameters. One big shift was coming up with the design ahead, because you needed to really design to a standard to drive efficiencies and increase productivity by saying, “This is what we know we’re going to need in the data center. These are the parameters we need to design to. Now let’s find land that can meet those requirements.”

What that helped shift was: there’s a team that does just looking for land, so to speak, and we could give them more structure and guidance of requirements to say, whether it’s land you’re looking for or lease facilities, look for these minimum requirements so that we could plug and play a design with some marginal modification. Some things you have to tweak and point a different direction or something to get a few inches here and there, but the considerations could be a little more cut-and-paste when it came to the designs that were deployed.

That said, finding the right location is imperative in terms of looking at where existing infrastructure is. Where does fiber already exist? Where do facilities and utility — where’s the power cost? Doing a map nationally of what different energy costs would be, how much they’re charging, what are some of the infrastructure qualities, do they have the type of renewable energies that may be integrated, and additional resources such as labor markets. Those factors are all relevant.

You’ll often see that you might look for where it is someone else, like an Intel or Google or an AWS, that’s already broken into a market, because then the assumption would be that if they’re there, the infrastructure has already been laid and invested. So you can come in. That’s why you’ll often see clusters in certain markets, because if it was good for one, it’s probably good for another. They’ll try to come in behind it oftentimes. But in a new deployment, those are some of the key factors you have to look at from a cost perspective.

The other thing is that you’ll often see them come in with blind bids, where they’re not saying who they are — because as soon as you say Facebook or Meta, now the price of the land goes up by four times.

Meredith: Yeah, I will say on the marketing side, when I try to research who’s building the data centers, where are they coming, I’m always like, “Dang it, they don’t have the names of any of these people!” I’m limited to my knowledge. That is interesting.

Neila: Yeah, those are real factors in the front end, and the designing aspect is really consideration of how you’re going to use the data center facility. There was a time where CPU compute power was kind of just the thing. So you were planning for your network and your CPU, and that was pretty much a standard approach. But once AI came on the scene and GPUs had to come into play and now you needed liquid cooling, you really had to pause and rethink your design completely and where those needed to go — what resources you would need to have, and things like that.

The larger these data centers get in the hyperscale world, you’re seeing these mega campuses, and they’re often just really looking for these very large swaths of land — and hopefully not any significant residential communities around that will negatively affect them. But they also found that just placing a data center out in some ridiculously rural location isn’t beneficial either, because people have to work there and they don’t want to live way out in the boonies. So there’s a lot that has to be balanced on the consideration of the upfront design portion of it.

Meredith: That makes sense. Before you talk about the operation side, can you paint a picture of the difference between a regular data center that’s been established compared to these hyperscale infrastructures? How big of a difference actually is it?

Neila: Yeah, so you have what we’d consider enterprise data centers. People have closets where they have equipment for their businesses. But then there are things where a company is small and they’re not going to need a lot, so they may just have cloud resources, where they’ll go to an AWS to provide their data center requirements.

But then you have enterprise — and an enterprise data center is oftentimes driven by proprietary data that they need to keep within their walls. They’re going to be on a smaller scale, relatively speaking, like 22,000 square foot per pod, and no more than maybe two campuses in a region. Or I should say, I call them campuses, but they’re probably just one data center per market, like two locations for redundancy, one end of the country and the other end of the country.

Versus a hyperscaler, where we’re talking major, major megawatts, gigawatts of capacity requirements, and they’re going to have campuses with 18 pods per building, and they could have four buildings on a campus. So it’s exponentially much more in size, square footage, power requirements, and just the total number of facilities that they’re going to have on one space. They’re going to have multiples of those across every state almost, as a hyperscaler, and then globally — versus an enterprise that just is trying to establish redundancy between one site to the other.

Meredith: And is the main reason or trigger that a company would need to develop hyperscale infrastructure — is it really the need for compute power? Is that the driving force behind it? Or are there other factors that really go into play there?

Neila: Yeah, so an enterprise customer is unlikely to have the same kind of drivers for what you see the hyperscalers driving around. They’re going to have a lower demand for an AI architecture. They are going to have growth, but it really depends on how their product portfolio is rolling out.

What I’m seeing is that most enterprises are building AI infrastructure mostly for their own internal application use. The difference is that a hyperscaler is usually customer-facing on their solution. If we use Meta as an example, they have WhatsApp, Facebook, Instagram, a number of online platforms — and all of those platforms have some level of software coding and development and integration with AI such that those AI tools can be utilized by consumers. All of those platforms have to be facing out to user communities, versus other companies where everything is facing internally for them to be able to use.

When you have external facing, you then have to put data centers closer to locations where you might have to hand off services to end users, or where you have a lot of data that has to be collected and transferred to a number of other people. If you think about social media and all of your cat videos — all of that has to be stored somewhere. Every picture, every video, all of that information has to be stored. That oftentimes grows these data farms essentially out into larger footprints.

Meredith: I feel like a lot of times when we hear “it’s in the cloud,” we just assume it’s somewhere, when a lot of times the data center is housing a lot of what is actually in the cloud. So it is kind of funny to think about it that way.

Neila: It is. Even when we speak about autonomous vehicles — those autonomous vehicles have to get information from a data center, and there’s data center applications and software that are managing those autonomous vehicles. All of that has to come back to a data center. Transmission of that data is required for us to do any number of things that we do on a daily basis.

Meredith: Sure. And I think a little bit later, we’ll talk about who does the responsibility fall on when we want to use these social media apps. We want to use the latest and greatest technology, but there’s also concern about what’s happening with all the data centers being built — but we’re also using it ourselves and pushing for all the greatest updates. So I think we’ll touch on that too, because I’m interested to hear your perspective.

Neila: Yeah. Well, to take you through your original question of the life cycle of the data center — you have that upfront design, and then you have the construction phase. Those construction phases are where we’re bringing in a lot of other partners that do construction, and that’s where we see a lot of other resources. Those can be where that longer window of time is happening, where the community is actually seeing construction happening and the break ground. All of those components are starting to be placed on whatever that property is. There’s different ways in which we’re trying to drive speed, but standards and sustainability for the materials and things.

Following that piece of the construction phase, then it becomes operations once it’s completed and ready. A little bit of the difference there in a hyperscale is that when it’s operationally ready, it’s really already fully loaded with racks and equipment and all of those things. It’s plug-and-play ready, versus an enterprise that’s generally like a shell, and it’s ready to receive new equipment and requests as time goes on over a longer period of time.

Meredith: Sure. And I was just about to ask: what stage do sustainability talks come into play? Is that at the design phase, or is it during — I mean, I’m sure it has to come in in early stages as we’re constructing, they kind of have to know the game plan. What does that look like?

Neila: Well, sustainability looks different in every phase, but it’s relevant to every phase. When you’re in the front end of the design, you should be designing for consideration of how your design is going to consider sustainability measures, such as how do you do closed-loop cooling. What type of equipment are you designing into your facility? Is it rated for energy efficiencies? The way in which we design the actual property use.

For example, instead of hauling off the dirt that you dig out for the base construction, consideration first: you’re going to lower your design of your facility so that you have a bit of a berm around it. Now you don’t take the dirt away. You can mount that dirt up. That creates a sound barrier, a visual barrier, and saves cost on haul-away. But those kinds of things have to be considered up front. Safety, how the fire suppression unit is going to be designed and what kind of materials it’s going to use, how much of it is going to be a modular design for quick deployment, and things like that.

Once you get past that, the actual execution comes into play where you’ve worked with your partners and vendors to talk about the packaging material, how things are going to be shipped to you so that you have reusable or recyclable packaging. How the actual walls will be prefabricated, for example, and then shipped as a whole piece so that you don’t have as much time on-prem. You have better use of resources as far as labor and skilled resources that can be managed off-site.

Then operationally, we talk about sustainability in terms of: how do we manage how we discard trash? What’s the lifecycle on the assets themselves, and how we destroy the actual drives and dispose of the physical assets? How do we manage the continuous trash removals? Everything that comes in there isn’t static. It has to come out. We have to consider what all of that really looks like. We also have to consider the deliveries of everything that’s coming in.

Can we work with partners to have more efficient means by which things are being delivered to the site, versus having five or 10 different delivery trucks that have carbon emissions, as well as just the time and resources it takes to address each person coming to the site? How can we work with them to reduce the number of deliveries so that we don’t have so many truck rolls? And that’s the same on the front end of construction. How do we reduce how many ships things have to be delivered on from across the sea? Can we do manufacturing here, close within proximity of a particular data center? So that we are reducing our carbon footprint.

Meredith: That’s a lot of moving pieces and a lot of different teams I can imagine that have to be involved in this process. Is there an effective way that you’ve seen to streamline all of that? Because it seems like there’s so many different aspects and lots of ways businesses could reduce their carbon footprint — I wasn’t even thinking about the shipping of materials and reducing that. How do you measure that? How do you get all these pieces together?

Neila: Some of the positions have gone to establishing sustainability teams, and they focus on ways in which the organization can be accountable to driving towards the reduction of carbon footprints. There was the 2030 initiative that said, “Hey, all of us are going to move towards a 2030 target of what the carbon emissions should be.” So then you set that as a target within your organization, and everybody has a stake in it and has a responsibility to drive that out within their own processes. That’s how you spread the responsibility around to help everybody keep their eye on the prize on how to do that in each of their respective lanes.

But a lot of times it comes back to organizations just articulating really clearly that this is important to them — to be sustainable, to drive out carbon emissions wherever possible, and to always be thinking about how we can create a more closed loop, or full-circle pipeline, of how we do what we do.

Meredith: Sure. That kind of segues into my next question about what the different opportunities are for sustainability at each of the different stages. You touched on it a little bit on the design side of things, but what are some of those sustainability opportunities when it comes to operations, when it comes to water, energy, all of those different aspects?

Neila: Yeah, well, this is the exciting part. I think it’s fun. First of all, there are a lot of global initiatives around sustainability where people are getting it right. First, there’s power. Next, there’s water. Then oftentimes we talk about labor resources as a sustainable measure as well. But then there’s, in general, the lifecycle management of the assets and things like that.

When we talk about what opportunities are really out there, the energy part is the part that we’re talking about the most right now. There’s some really great things being done that are exciting across the globe. Some of those things are, for example, installing or balancing energy demand by having on-site power generation, having microgrids. You’re hearing about nuclear small modular reactors — this is not your grandfather’s nuclear. This is a new-age nuclear that’s got lots of safety protocols and is much smaller in scale.

You’re also seeing where data centers have gone into partnerships with nuclear plants and they’re in proximity to those, so the transmission lines are much shorter. These are renewable resources, and having deployed renewable resources and energy is a benefit that allows for both the offset of power impact in the community, but also the ability to have PPAs (power purchase agreements) and provide energy back into the communities where there’s excess. Those are things we really want to push, that are able to leverage these energy technology innovations.

Google, Microsoft — they’ve all really done a lot of great things. There’s something over in China where they’re leveraging all of the different renewables all at once: solar, wind, all of the hydro potential. What I thought was wonderful is that they deployed solar panels across their desert. Because there was a lot of dust on the solar panels, they were rinsing them off. The water that rolled off of them ended up causing the desert to go green. Then they had the problem of really high grass growing, so they brought in sheep, and those sheep kept the grass low. Then they found that it just created a really eco-friendly environment that also was large enough to sustain huge energy demands that are being driven by AI.

They’ve also submerged data centers to bring down temperatures and cooling. I think Google did it. They kind of abandoned it. But those kinds of opportunities really exist, and those are where local communities and legislation have the opportunity to help drive more initiatives around how our utility companies can do more to partner with renewable energy resources.

The other thing is that data centers are like ninth on the list of power consumers — next to manufacturing being number one. You talk about auto plants and other things like that, and textiles — they use a lot of natural resources. But it depends on what market you’re in. If all you see are data centers going in, that’s where you see your concern for how much power is being consumed, and your concern about water. But the reality is the data centers use the least. The opportunity, though, is that there’s clarity and understanding of how the data centers are using resources.

One thing to know is that oftentimes a data center is working with the utility, giving them land and giving them money to build out their infrastructure. So they really shouldn’t be passing that cost on. But the other side of the coin is the United States power grid is just old. It’s been in need of updating for a long time. That’s where we see them realizing they can’t just get away with the infrastructure that they’ve had — they have to leapfrog into the future, and that’s expensive. So they’re having to do a lot of those investments.

Water is oftentimes designed into what we see as retention ponds on larger hyperscale data center properties. That collects the water that can be used. It also offers an attenuation method, because you can put fountains into the water that help dissipate sound. If you watch A Quiet Place, that movie — the creatures can’t hear you if you’re sitting in front of a fountain. It’s the same thing at a data center. If you have attenuation walls, those are designed to block noisy things like mechanical systems, but very lovely, pretty ways to do it are to put fountains around on your property. The water creates a natural buffer — and it’s much prettier to look at. It also creates some security, because if somebody were to try to come onto your property, you have a moat around your property.

Meredith: Right. Is that water by any chance drinkable? Or is it water we can’t repurpose for that? It’s more what you just described.

Neila: Yeah, so in a data center design, generally, it’s not using potable water. It’s not going to be water from your drinking water system. It’s going to be all of the runoff, things that nobody really wants to drink. Then it’s used in the actual cooling systems. That’s the primary use of it.

When you hear “liquid cooling” in the data center for AI, liquid cooling is not water. It is a liquid chemical combination necessary to avoid any kind of breakdown of the equipment, but it offers the ability to run liquid through and offset the heat. But in terms of the water, the primary water that’s used on a data center, it’s measured in your WUE. So your power is measured in PUEs, and we keep those numbers targeted to be low for efficiency. WUE allows for us to measure our efficiency in water utilization. That water is usually in a water tower that you keep on a property. That water is used as backup water from evaporation that happens in what is a cooling tower.

The cooling tower really sounds like a waterfall. It recycles the water through the facility, offsetting cool water with hot water. The hot water comes out of the facility and cool water comes back in through the pipes. That water gets treated so that it doesn’t break down the piping system. But there’s very little water that has to be taken from any new water sources or community water sources, because that is a closed-loop water system. It really just keeps recycling the same water except for what’s lost in evaporation, and then that water can be brought in to refill that. Those are waters that most people wouldn’t be drinking, and can be leveraged from the retention ponds as well.

Meredith: Interesting. Now, you also mentioned the labor side of sustainability. I’d love for you to talk a little bit about that.

Neila: Yeah, so it always comes up, particularly when we’re talking about deploying data centers, because the technical resources aren’t necessarily everywhere that we go as far as new build locations. But the technical skill sets for fiber optics — a lot of the resources that we’ve had have retired. They’ve aged out of the system.

Meredith: We had one of our fiber optics guys retire last year, so…

Neila: Yeah. They’re in that age range of the 50s and 60s of when that kind of was the thing to be, and a lot of people got into it. But what we’ve seen over generations are fewer people in trades, fewer younger people interested in doing this type of work. They look towards many other avenues. It has created somewhat of a skill deprivation for the demand.

Maybe in the past, we would say we had enough. But because the demand has become exponential and there’s so much building going on everywhere, we need really everything from fiber splicers and those who understand fiber, to those who understand plumbing and can manage the aspects of how to run the liquid cooling piping, to the electricians. There’s always more electricians needed for any operations site. The majority of resources we have on site are made up of electricians and mechanical people in the back of the house, and then our site technicians who are having to troubleshoot and manage all of the cabling connections and the actual assets and platforms.

In a normal pre-AI world, we probably had enough of these. But with the boom in construction, that means more data centers that need more people. That means more new construction that needs more talent to execute on those sites. Not every market has the same amount of qualified resources. Then the challenge becomes: how do you create a market that drives the development of the skill set, so that you can use local talent and not have to ship people in from other places to do the work?

That’s the sustainability piece. How do you work with communities to create pipelines of talent that are relevant to the needs of those businesses that have come into that market? AWS has some really great programs where they provide community development. A lot of construction companies will work with local communities to pipeline talent and help to inspire the next generation. But you have to be really intentional about your labor commitments — to bring in non-college-degree people as well as some college people, but creating those opportunities, whether it’s leveraging veterans that have gotten out of the military recently, to new college grads, to new high school grads. Just creating different levels of opportunity so that we continuously have people available to do the physical work that can’t be AI’d.

Meredith: Yeah, it can’t be replaced. There’s a solid market for that right now. That’s so interesting. I never really thought about the labor aspect, but with people talking, you know, they’re scared to lose their jobs and things like that — trade school is a good avenue to go into now.

Neila: Absolutely.

Meredith: Well, in your opinion, where is the biggest opportunity for sustainability when it comes to all these different stages — from design to operations to labor? Is there one area where you think there’s the biggest opportunity?

Neila: Yeah, I think we’re going to continue to see that the biggest opportunity is to solve the biggest problem — and the biggest problem we have right now is power and energy availability. Coming up with solutions on the front end: how the assets, the equipment themselves, are more efficient so that they don’t demand as much power. There’s evolutions that have been happening to reduce that power, with the density of the equipment that is being released — servers that are more dense but not as power hungry.

The other side of that is how we leverage renewable energy and ensuring that we’re driving our utility companies to really own a lot of those renewable energy partnerships. I know there are companies out there doing really great renewable energy solutions. It isn’t going to be one silver bullet. It is leveraging all of the renewables that are available to us to really solve the United States energy problem, and really getting better monitoring of how we’re using our power effectively.

You’re seeing companies coming out of the woodwork as far as new ideas on how to solve for energy utilization, how to better monitor it so we can make better decisions around it. And the energy companies, utilities — some utilities are fast-forward, moving quickly towards where they need to get from an infrastructure standpoint, where there are also smaller energy companies that just aren’t able to adjust and they’re going to end up being kind of left behind on some of this next-generation growth that we are seeing across the country.

The country as a whole has to have a solid grid to be able to sustain everything from what is what we consider terrestrial data connectivity. But you’re also seeing what I call celestial — all the satellites that are leveraging, you’re seeing a lot of capacity going up into our orbit, because the sun is a never-ceasing energy source. It is very viable for a lot of the big transmissions and data center requirements that are being able to be designed into satellite architecture that’s around the orbit.

Meredith: And you have some experience with the satellite, celestial energy too, don’t you?

Neila: Yeah, I have the benefit of getting some exposure with that, in terms of working with a gentleman who has started his satellite company, and having an opportunity to sit in on some of the advancements he was doing, and some other people that were working with NASA on satellite deployment. It’s really a fascinating thing to have been able to see. They’ve already started to leverage the moon for storage of data, because it creates a different level of security and opportunity to erase borders. We have borders here in the United States on Earth, but on the moon, it’s kind of like shared space essentially, right now anyway.

There’s just a lot of interesting dynamics that we can leverage between what is satellite technology and its connectivity back to what are our terrestrial data centers that we have to have here. They’re not completely independent of each other. They’re still — just like wireless cell phones, they don’t just exist out in the ether. They still have to connect back into the network.

Meredith: Back to something, yeah. Interesting. I didn’t even think about that, so I’m glad you mentioned that. All of this does tie back into our conversation about community impact and what people are thinking about what’s happening in the world. I also want to get your perspective because I know you’re in Virginia, correct?

Neila: I am — data center alley.

Meredith: I was going to say, you’re in data center alley — tons of growth in Northern Virginia. Have you yourself seen the impact? I don’t know if you live close enough to that area, or it’s just something you’re familiar with since you live in the state, but have you seen the impact of that at all?

Neila: I definitely feel the rise in my power bill, so I understand it. But I get the benefit of having been on both sides of it here, in terms of having the conversations with the utilities, understanding what investment was being made by data centers to the utilities to help them meet the demand and requirements. But then also being on the consumer side and feeling the pain of really high — what feels like very high — power bills.

But what I look at as a whole is, I see that a lot of those costs are not necessarily being driven exclusively by data centers, but simply because, as I stated, the landscape is changing as a whole. These utility companies have been woefully behind for many years, and now is the only opportunity they have, where they have to really shift. Now that cost is being seen — versus had there been more incremental evolutions in the utility infrastructure over time, we wouldn’t have felt it all at once.

That’s what we’re really seeing now. I would like to see more offset from a government federal level to help offset a lot of the cost. Because the reality is, this grid is a national security requirement, and we need, as a country, to invest in where the grid needs to go for us to be able to support the technology that the country wants to be able to have as a primary boon to our economy. It’s an opportunity, I think, that shouldn’t be simply expected of each individual community and county to have to shoulder, but to be seen as much more of a national opportunity to invest in the grid.

Meredith: Sure. So do you think that’s the biggest disconnect right now — between how data centers are being built and the energy concerns — the disconnect being that the power grid has not been worked on as much as it should have been throughout the years? Or do you think it’s also an issue of how quickly AI is scaling in general? Or a combination of those two things?

Neila: When I reflect back on the last couple of years, it seemed like it took people a while to really embrace where the AI conversation was going and the reality of it. There were a lot of investors who had cold feet on how long it was going to take them to really get their arms around what it meant to support and invest in AI. You also had utilities that really couldn’t conceptualize, I think, about how much power was going to be necessary for the future of this technology. A lot of people were behind the eight ball, even though they were being told, “This is coming, this is here, this is happening.”

A lot of that disconnect is because people just really can’t see behind the curtain of how hyperscalers are driving the industry forward as one thing — but not just hyperscalers in the sense of social media, but also as a cloud provider. The bottom line is, really hyperscalers are driving it because the technology has driven us to the state that we’re in to say: we want electric cars, and we want those cars to drive themselves, and we want those cars to be able to play music and to give us our laundry list and to schedule our babysitter or whatever. To do all of those things, it requires a lot of data. All of that data has to be someplace, and it has to be able to talk to each other.

It took the industry — it took the energy industry in particular — a long time, I think too long, to really see ahead to what was coming. The fact that the medical industry is able to do surgery remotely via an Oculus headset — all of that requires transmission. That requires a lot of data and it requires a lot of information. Without having utilities and tech kind of coming together to understand really how technology is being used, they really just weren’t ready for the exponential growth that we’ve seen, especially coming out of COVID — because that’s really where we saw this shift. It was really the light-bulb moment where somebody said, “We can do this, and we can make it available to everybody.”

It was like when the internet came online. I got a lot of questions like, “I don’t get AI, and I don’t understand how it’s going to be applicable.” I said, “Well, do you remember when the internet came out? Nobody maybe really understood it, but everybody started using it. Because the human mind is just so creative and innovative, it became the wildest dreams that anybody could come up with.” The same is true of AI. But with that also, it just means that people have to move a lot faster in the decision-making, and getting us to where we need to be in terms of thinking further ahead of what could be — versus being stuck in the mud of “Is it going to happen?” It’s already happening. Get on board. Let’s move it forward.

Meredith: It’s interesting, because I know back when the internet came out too, there were tons of people who were very anti, very against it. And now it’s so normalized and standard. It would be absurd for somebody to not use the internet at all in their life, completely off the grid. That would be insane.

Neila: Yeah, people choose to, if they want to, that’s fine. But most of everything we do cannot exist without the internet.

Meredith: If they want to, I’m all for it. That would be impressive. It doesn’t really fit my personal lifestyle. I like the convenience of things. But more power to those people. Where do you see this going in the next few years? Do you think we’re going to be able to bridge that gap between the utilities, or is it way off in the future? What do you anticipate?

Neila: Well, I think we’re moving there. I think it’s going to come down to a lot of the legislation holding utilities accountable, and new developers accountable, to investing in the right type of energy and helping support being able to be permitted to use the new technologies like nuclear and SMR and all of those things. The faster those kinds of things can be allowed and available and bring costs down so it’s accessible, those kinds of things become a part of the fabric.

I see, because tech is being much more talked about and adopted from a government perspective of understanding, there’s some hard facts that are going to have to be associated with it. We’re not going to be able to stay in a lot of the old coal kind of approach mindset to energy and resources. We have to start thinking in terms of: how do we take ourselves and launch ourselves into the future?

We’re not by far the first at all to do any of this. There are a lot of other countries that are doing some really wonderful things in terms of renewable energy, in terms of understanding scale and growth of how to support technology in their environments. There are a lot of places we can learn from and partner with — of leveraging things that work well in other countries and adopting them here, to get the most out of our infrastructure.

Meredith: Yeah, I’ve seen some positive case studies just in Europe, Finland, Sweden — where they’re able to power smaller communities using the leftover data center energy. It will be interesting. Lots of good case studies. I’m hoping to see some progress on that. Is there anything else you wanted to add to the discussion, just in regards to sustainability, community impacts, anything going on?

Neila: Yeah, I think technology is here to stay. It’s something that’s a part of who we are as a country, for sure, as a world. We really have a good plan of how we can build towards technology while being responsible for the communities that we live in. There’s so many win-win opportunities, just as you mentioned, in other countries where they’re using the offset of the energy to heat the sidewalks and the streets. Building communities with intentional, smart designs, so that the communities really benefit from any of the tech growth that’s happening — I think there are clear ways to do that. I just would love to see more planning and collaboration of all of the players, so that the communities win in the end.

Meredith: Yeah, that’s an awesome perspective. I feel like I learned so much in talking to you today, just to get the full picture. Both of us — we’re part of our own communities, and Georgia has a huge data center footprint too that’s growing pretty significantly. So we’re kind of right in the middle of everything going on. Super interesting to get your perspective on everything, and thank you so much for coming on the podcast.

Neila: Thanks for having me. It’s been a pleasure.