Building Data Centers With the Community, Not Around It

The Empire of Lights is brought to you by Comcast and Babst Calland.

Transcript:

This is the Empire of Lights, where we are exploring all things- ... around the data center movement, particularly here- ... in Western Pennsylvania. And I'm your host, Jonathan Kersting, with the Pittsburgh Tech Council- ... and TechVibe in the Huntington Bank Studios. And today I could not be more excited to talk to Brian Regli- of the new Springdale development. It's been in the news a lot. People are learning so much. Yep. The Dynamo. Very excited to learn what's going on today, Brian. Who's better to have than the guy that's actually making this thing happen? So- Well, Jonathan, there's a deep team of people that are making this happen.

Of course, yes, yes, yes, yes. So I'm, I'm really pleased to tell you that story about that team and all the things we're trying to do. That's ... I wanna know. Absolutely. Yeah. First off and foremost, like, just tell me about your- yourself real quickly. Your, your experience. You got a really cool, diverse background.

So, so thank you. Yeah. I mean, look I am someone who grew up going to construction sites. Okay. My family has been in building for a long, long time. Awesome. And my grandfather was a mason, uncles [00:01:00] architects, literally- ... this is what I did growing up. But a lot of my early career was actually in the telecommunications industry and the utility industry, working on regulatory frameworks, understanding how government and business can work together to make change happen.

And I'm a Generation Xer, so I'm in my late 50s, and so I lived through- There's the Gen X thing ... there it is, it is a thing. Boom, boom, there you go. Lived through fiber, lived through the dot-com bust, lived through the ups and downs of all these various cycles. But the one thing I learned along the way is that infrastructure is a crucial part of any economic development strategy- Absolutely

of any social development strategy. And so any infrastructure development that leans into innovation can be good if it's built the right way. About a year ago, I, I was fortunate enough to be approached by Davidson Kempner, which is an investment fund out of the New York and London areas that has a data center investment strategy.

And they had purchased a coal-fired power plant, a former coal-fired power plant in Springdale, Pennsylvania. Yeah. And from the ashes of that demolition, they saw the opportunity to build a next generation hyperscale facility. Yeah. I mean, perfect spot because you got utilities there, you [00:02:00] got the space, like all these things that make it a great spot.

Like- Exactly ... what was once a very stranded utility- Correct ... or, or a piece of infrastructure- Correct ... can now become, can be put to life again. So keep going, yeah. And, and this is, I think, the important piece of it, right? Yeah. This is a brownfield that right now has gotta go through the Act II process at the state, which basically means there's remediation that needs to be done.

It's a contaminated site. It's literally 100 years of coal-fired power plant operation on this site. Ooh, so there's some wicked stuff there- There is some wicked stuff- ... that has to be remediated ... that has to be remediated. Yeah, of course. But by the same token, if you have the capital required to do it- you can move it from being an unproductive piece of land to a productive piece of land. And when we- Yeah ... started talking to the community about this actively in May, June, July last year, our goal was to create as much community engagement as possible. Thank you for saying that because- 'Cause it's important

it's just so important. I mean, that is literally the foundation, so yeah. I agree with you. A- and this is where I think, frankly, the, the AI community and the data center community can do a better job and is not doing the job they need to do. Okay. Part of the challenge is not just to show up and say, "Hey, look, we got a gazillion dollars and we're gonna drop a data center in your community- Right

and boy, you're [00:03:00] gonna love it." 'Cause that's not fair, I like that. People live in those communities, and you have to understand what their requirements are. And frankly, every site's a little bit different. It's not like the same thing that you're gonna build in Springdale is the same thing that you would build in Arizona or in Nevada or someplace else.

The only way you can do that is by sitting down with the community and asking them, "Hey, what's important to you? Is it trails? Is it public infrastructure? Is it your schools? What are the things that we can do that as we bring capital and labor to this site to transform this site, would create a better outcome that even though you may not love a data center, even though you may say that you don't trust me or like developers in general, at least you can walk away from the conversation and say, 'All right.

They heard me. They're listening and they're gonna try to do what they can, or at least they're gonna make certain adaptations. Okay. So that to us was the important part in the fall. A- and ultimately now what we're moving forward with is the next round of this, where they've given us permission to build, which is great.

Awesome. Great that we got the yes. Okay, step one, step one, step one. Right. Okay. But now there's step two, three, four, five. Yeah. You love how that works, right? [00:04:00] 10 trillion. Yes. Absolutely. So, so we're now excited to be moving forward on the land development process. So now it's getting the permits so that we can actually start doing the work, and there's regulations and rules that we have to follow, and the joy of this experience again is we get more public engagement.

So we appreciate all the work that you and others are doing in the region to talk about data centers and what they can do. But fundamentally, it's still gonna be about, folks like you and me and others going out into the communities, having conversations, and figuring out what these developments really mean for the communities where we locate them.

I lo- I, I just love hearing that- ... because I think that is just the right approach. At the Tech Council, we're all about making sure technology moves forward. And we look at the opportunity- ... that data centers, no matter where they're built- Yep ... are bringing to our manufacturers- yep ... here in Western Pennsylvania. And I, I mean, we look at the fact that, like, literally we have members with two and three-year backlogs- ... 'cause they're making parts and pieces for the power grid- Yep ... for the pieces that go into the data centers themselves. And so we get very excited when we see something happening in a, in a town like springdale. Springdale, where basically, like I said, here's a piece of land that was- [00:05:00] It needs to be remediated- Right ... so it's being cleaned up, and it's being put back- ... to use, being put back on the tax rolls- Right ... like all the good stuff. Right. Some employment to goes, that goes- Yep

along with it during construction and then a little bit- Yep ... afterwards. Absolutely. So there's key things there, but it's also providing a center that we need. I mean, I pull my phone out and I pull stuff up- ... it's because there's a data center behind it. Not just for AI either. Just for all types of things.

Well, see, now, and now you're getting into something that I think is specifically relevant for Western Pennsylvania. Yeah, okay. Right? There, there are things that you build here, there are things that we do here that lean into that digital economy. Some of it is the information economy. Yes. You have software companies here.

They're brilliant companies that, you got Gecko Robotics down the way here. You got companies like Duolingo. You got all sorts of companies here that spun out of CMU that could do really fantastic things if they had access to the compute, and if that compute was readily available within the community.

Interesting. Okay. So there, there's a piece there that's really important to lean into, but then even beyond that, there, there are health systems here. There are banks here. There are other institutions here that could really benefit from the application of these large language models. And if we can make customer relations better, if we can make healthcare delivery better, if there are things [00:06:00] that we can do in Pittsburgh because we have both compute and infrastructure and the people, then that's gonna hopefully supercharge all kinds of- Yeah

changes in the ecosystem in the Pittsburgh region. I think that thesis is real. I think that the folks, for instance, at the AI Strike Team, Joanna Doven and that group, I mean, they're doing things to promote those ideas. I think those are great ideas, and fundamentally, we're a part of that story, and we're really pleased to be part of that story.

Yeah. It's very- ... it's so interesting to us. So let's talk about some of the key stats as to what's, as what this- Yep ... data center would be- Sure ... 'cause I think that's where a lot of people get concerned. They- ... talk about things like water usage- ... noise- Yep

all those types of things that make a community concerned. Obviously, no one wants their water to be drained, and- Yep ... people wanna make sure that, that they're not living next to something that could be bad for their health and so forth. What are the key stats? Like, give us the facts as to what, how this data center- would actually be integrated into the community. What resources- ... will it use? So, so let's go through the top line- Yeah ... right, Jonathan? So this is a data center that at its height, when we have it fully deployed and fully built out, will require about 350 megawatts of power. [00:07:00] Now, within that 350 megawatts of power, about 200 to 240, depending on what chips are deployed and what companies we partner with- Sure

are gonna be used specifically for the compute itself. So you're talking about chips that are drawing power in order to do the work of a hyperscaler in the deployment of these large language models. The balance of that power is used for things like heat rejection. It's used to run the systems within the building, and that's a lot of power when you really think about it.

Ultimately, the hardest problem of these data centers is not necessarily the power, it's the heat. It's the heat, right? And how you reject the heat. Yeah. So companies have begun to deploy technologies drawn from billions of dollars of investment that are more and more and more efficient in the area of power and water and sound, and what we're gonna benefit from in the stage of development that we're at are all of those contractors and all of those vendors doing that hard work to deliver a better and better product.

So concretely You asked about the cooling systems. Yes. We're using what's called a closed loop system, and Jonathan, I know you're familiar with [00:08:00] those, but for the listeners that are out there- Hell yeah. Tell us more- Let's talk about this, right? 'Cause when I learned about these- Yes ... my mind was like, "Whoa, okay."

It's, it's a completely different way of thinking about the cooling process. And yes, data centers in the past quite often use groundwater or other types of source water because the biggest problem here is the chips, and boy, they're running at a certain temperature, and the, the water that goes past them might come out at 130, 140, 150 degrees, depending again, on the intensity of the activity, and you got to bring it back down to like 80 or 70 degrees in order to run it back through the system.

You're not going to simply pull groundwater in that circumstance because also you need to make sure that it is viscous enough, it's got to move through the system fast enough. And it's pure enough. You don't want to have the impurities of the groundwater. So the closed loop systems exist to make sure that the system itself does not draw from outside resources that could contaminate or affect the process of cooling those chips.

Ultimately, what you have in a data center like ours is a one-time event of about 500,000 gallons, about the amount of what's in an Olympic pool- Okay ... that charges the system, and that lasts for between 5 and 15 years- Wow ... depending on the integrity of the system. So [00:09:00] it's like swimming up an Olympic-sized pool- Pretty much it

the amount of water. Okay. Pretty much it to start with, right? Gotcha. And then you're putting, essentially the equivalent of antifreeze in, so it moves a little bit faster, and you're not necessarily creating any environmental hazards 'cause you've got a contained system that's controlled. And I think this is one of those innovations that have come out in the last really five years- That's what's so cool is, yeah

that change, yeah, it changes how you think about a data center. Yeah, 'cause an argument that was valid five years ago is no longer valid because the technology has marched forward- Correct ... and it's addressing this, 'cause that's what I love about human beings. We solve problems. And that's a really important point.

Yeah. That's a really important point, and actually, let me just take a quick diversion on that 'cause I think it's a really, it's a crucial element for everybody to consider. What you have today is not what you're gonna have tomorrow or the year after- ... as long as we continue to innovate and work together and collaborate.

It's the moment that you stop working on a problem that you never get back to the problem, and then you never solve the problem. Good point. And the infrastructure that we're trying to build has to be responsive to different challenges, not just today, but in the future. So let me shift over to power.

We've gone from water. Yeah. Let's go to power for a second, right? Let's go to power, 100%, yes. Power is something that, every data center [00:10:00] is going to draw, and when I stand in front of communities and people ask me, "Hey, you're responsible for the increase of power bills," I usually say to them flat out, "Yeah, data centers are causing your electricity bill to go up."

And frankly, there's a lot of folks out there that, that would say, "Well, wait a second. You're developing data center. Why would you say that?" Yeah, exactly. Why would you say that? Because I think that's actually the truth. Okay. A- and the issue that I think many data center companies are ignoring right now is when you look at the capacity market as it is constructed, and I don't wanna get too far into the weeds- I hear you.

Of course ... of regulation, but it's kind of fun, right? Oh, it's so much fun. It is. It is a joy, right? Yeah. So there is this interconnection service organization called PJM, and that's, you know- Heard of those guys ... those are the guys who basically run that capacity market. Well, that capacity market was never designed to deal with the market conditions we have today.

So every time there's an auction, the auction price maxes out because somebody's projecting a future demand that may or may not materialize 'cause these data centers actually have to be built. And so when you're in a position where the market does not show the real demand, it projects a future demand, and the future demand is entirely speculative, of course the price is gonna [00:11:00] go through the roof.

Interesting. I've never heard it that way before. So- Okay ... all of the speculation around how many data centers are gonna be built, oh my God, if there's thousands of data centers, well, then no one's gonna have any electricity. The market is just reacting to that as markets do. But the reality is a lot of the projects, I would say, frankly, the vast majority of projects that have been announced are not gonna get built for the next three to five years.

It takes a long time to build these projects. So- It's not that our data center or any individual data center is creating the market condition. The collection of those data centers- Interesting ... and the, the volume of requests in the utilities have this impact. So the question that we all need to ask ourselves from a responsible perspective is, how does each individual data center get developed to reduce that impact?

And there's a variety of ways to do that. Tell me. Yeah. Okay. How, how does that work? Well, well, number one, and this is where I, I wanna lean into what Governor Shapiro and his team has done with the grid principles. First and foremost, there are certain types of rules that at the state level, from a policy perspective, and hopefully from a legal perspective, get implemented because the requirement to curtail when there is a [00:12:00] power stress on the grid, which means the data center has to reduce its load at any given moment- because the consumers need it before the data center, that's something that data centers can do, and they should be responsible for doing. The ability to pay for infrastructure that's unique to the infrastructure, this is a whole process that everybody's talking about right now. If, if there is infrastructure that is being built for a data center alone, the data center should absolutely pay for it.

The ratepayers should not have to pay for it. That just makes perfect sense. So, so- I mean, why would... Yeah Why would you not do that? Why would you do that? Right. Yeah. So, so when you go into the market of, power and, and components and substations and materials, you're also driving up costs because you are now demanding things that three or five or seven years ago, there wasn't as much demand for all of this equipment.

It's good for the vendors, but it's bad for the utility companies 'cause now twice as three or four or five times- Mm ... more expensive than it was in 2015, '16, '17. So we all have to contend with this, and any company that steps forward in the data center space that doesn't take at least some responsibility for being part of the dialogue, I think is ignoring a fundamental truth, [00:13:00] which is you can't be part of the solution if you don't believe you're part of the problem.

Yeah. And we all have to be part of the solution. Interesting point. That makes total sense. But here's the thing- Yeah ... for Pennsylvania. We export 35% of our electricity. That's the number that I've heard, which is pretty amazing- Which is pretty amazing ... when you think about how much , we export. Yeah.

We are the second-largest power producer in the United States behind Texas. That's amazing when you think about that. And so when you think about- Yeah ... the strategic advantage for Pennsylvania, and for Western Pennsylvania in particular, this is where we need to do the power thing right. If we can do it right, and we can coordinate between the regulatory agencies, our independent power producers, our grid companies, and the data center companies, we will be the strongest, most efficient, and most effective market for building these things over time.

But we have to work together to do that, and we can't sit and yell at each other and- ... complain about each other in that regard. We've gotta actually maintain that dialogue and keep solving the problem and keep moving forward on the power side. So what we're doing in our data center is to start working on how those curtailment policies might work, how to create flexibility with the grid, how to use the [00:14:00] interconnection that we have with FirstEnergy, and frankly, the adjacent infrastructure we have with Duquesne, so that if we can generate some power on the site that makes it available to people like that- I was gonna ask about that.

Is there ability to generate some power on site? There is some, yes. Okay. But this is not like a mega center, multi-gigawatt center- Right ... that's got, like, thousands of acres. This is what you would call in the business an edge data center, where it's gonna be close enough to an urban core so that the fiber has very, very sort of minimal latency.

But it also then cr- sort of creates an opportunity for three or four or five data centers like this that are kind of a shared infrastructure for the different types of companies- ... and organizations that may need- Right ... either co-location services or hyperscale compute. And this is where we think, at least in the investments that we are considering in the region and the one that we've announced so far, puts us in a market to really deliver on that kind of promise for the customers that might need it.

Very interesting. What about noise? So noise, again, critical issue that folks need to contemplate. The main source of noise are the chillers because the chillers for heat rejection, you gotta go through that secondary loop, get it up to the roof, blow a bunch of fans through it. Those fans, just like any fan in your air [00:15:00] conditioner- Cool it down, bring it back out

cool it down, bring it back in, right? And the loop continues, yes. There's also issues with generators, but generators are usually not run very frequently. If they're tested, they're tested monthly. You can build enclosures. It's the chillers that people are concerned about, and they have the reason to be concerned about that.

Again, same thing that we talked about before, the technology is changing and the vendors are catching up. Just go to York, Pennsylvania, right? Okay. Johnson Controls. Yes. One of the biggest companies in the region. They've hired an entire team of acousticians, and literally- Love it ... all they're doing is trying to figure out how to tune these things- Thanks

so you don't get those low rumbles. So there's a point where we're privileged to be talking to this group, in part because Turner Construction, which is, one of the largest data center construction companies- ... in the United States, and they're doing, like, 14 billion this year. Their Pittsburgh office is their design build partner in this.

How cool is that? And they've got a great team. To have them right here in Pittsburgh. They're right here in Pittsburgh. Yeah. And they have a team of people that do nothing but source equipment for data centers nationwide, globally. So when they make a phone call to Johnson Controls, it's like, well, here's a big customer- Yeah

Turner Construction, that's making a call to Johnson Controls, and then we can have a dialogue about how they're building the [00:16:00] things that they're building in York and how they're gonna be used in Pittsburgh. Yeah. So over time, I also believe that sound, again, mitigated just like water. There's gonna be a moment where people will not even realize that they're near a data center.

I don't know if that'll be tomorrow or next year or the year after. I think our data center is gonna really be good at this. It may not be the best, it may not be perfect But the ordinance in Springdale is, we have to be below 85 decibels. We're gonna be well below 85 decibels. The ambient traffic around the data center is about 55 decibels.

We believe we'll be well below that as well. Even below 55. Yes. Gee, that's- So we think that ultimately we're gonna end up being part of the background noise that people are not necessarily even gonna notice. Yeah. But again, the engineering is still underway. It's part of what you have to do during the land development process.

And then ultimately vendor selection will be completed, and, and then we'll be able to really talk about the ultimate outcomes. But the one thing I would want to emphasize to you is that under the conditional use approvals that we receive from the borough We are actually paying for third-party consultants to validate all the things that I'm telling you here today.

There's a point where I [00:17:00] show up in front of a community, and the community's like, "Hey, that sounds great, Brian. You seem like a really nice guy, but I don't trust you, okay? Here's an engineer." That engineer shows up at a conditional use hearing. They get sworn in, and that person testifies to a certain outcome, and they, "It's great.

I, I, but I don't believe you're an engineer either." Right? I get it. Now, that's another threshold, but I get it. Fine. You choose your engineer. We will pay the borough to review the work that we do, and your engineer will say if we're right or not. Yeah. And ultimately, there's a point where that kind of dialogue and that kind of interaction hopefully will raise the, the dialogue and the debate about what's real.

That's, and that's what it's about. Yeah. Yes. And like my, my key takeaways from this are just, like, there's so much transparency going on here. We're trying. That's what's, it's just so important to me to see that- ... and the idea that, like, you're seeing standards we're going to exceed- ... those standards.

That's the goal. And ... doing this in a, in a thoughtful way- ... is just, it's not just, "We're gonna build this great thing, and you're gonna like it." No. It's like no, no. No. So a- and it's a, it's a tough journey to travel there. Well, it is. And, and sometimes, you know the Simpsons episode where the guy's like, you know-

like, "Hi, I'm, [00:18:00] I'm building you a monorail," right? I mean, it's Phil Hartman. I miss Phil Hartman, right? It's like, yeah, some days you feel like, "Okay, what am I trying to sell here?" And I mean that in the best of senses. But I, I think if you're being intellectually honest about your team and this, this deep team that we have developed through Turner and Mascaro Construction and all these other companies that are working with us today, fundamentally, it's, it's about explaining the engineering of it, the optimization of it.

Ultimately, these are just big engineering and optimization problems, and by saying just, it feels like I'm diminishing it, but, but fundamentally, it's, it's like any good, solid team exercising their intelligence to address a set of issues. We can solve these problems. We can build it well. So let's figure out what that standard is.

Let's codify that in the laws and the rules that are in this Commonwealth of Pennsylvania. I love that you say put that in the laws and the rules- Let's do it ... because that's what we want. Right. We want to enforce this. Exactly. We want to make sure that people are safe. And frankly, and frankly, from my perspective- Yeah

I wanna make sure that everybody does it well because, if not, it's gonna be, frankly, disadvantages- It's, it's, it's bad on you ... disadvantages us, right? Exactly. Right. Last thing that I'd like to [00:19:00] talk about a little bit. I'm not sure how deep you can go into this- ... but the idea of, like, looking at, like, as you put this data center together- Yeah

in Springdale, like- ... how you're gonna be using local labor, how you're- Yep ... gonna be using locally made parts and pieces and concrete and everything it takes to put a data center together. No, that's the goal. I mean, the reason why we chose Turner's local Pittsburgh office is exactly that. Okay. We know that they have the skills and the resources here.

They can lean into their global network. Yeah. But fundamentally, they're gonna source as much as they can from Pittsburgh, and we know that there's gonna be a local hire provision in the contracts that will require us- Great. Yes ... to have a certain number of people- ... from the community. This will be a union job because it's just too big.

Even if we didn't want it- Yeah ... to be a union job, there's just no way. A- and fundamentally, we need to train more people and get people off the bench. We need to get them into the workforce. So Mascaro Construction, one of the largest construction firms in, in, in the Pittsburgh region, they're partnering with Turner.

We've been talking to the trades about who we're gonna need, when we're gonna need them, how we're gonna need them. This'll be thousands of jobs on the construction site. We're gonna have to coordinate all of that. So leaning into all of those things is critical. But meanwhile, [00:20:00] leaning into some of the other things you're talking about, yeah, there is local manufacturing for parts and components, and we've gotta work with the community to do that.

There certainly are schools where people can come out and get trained, and the high schools have to be- Okay ... an important part of this. Yeah, for sure. You have research work at both Carnegie Mellon and at University of Pittsburgh. I mean, I, I don't know if you saw, but University of Pittsburgh just received more than $300 million from the National Science Foundation for this grid, essentially, electrification, how do we do grid infrastructure better?

And that's gonna lean into supply chains in very unique ways. All the way through, yeah. I mean, it, it's a really unique moment in Western Pennsylvania where this intersection of new, fresh capital and, and hopefully, you know, the, the great history of Pittsburgh as a builder of things can come together in a way that I believe could be transformative to the tech community, and it's a, it's a great privilege to just be a part of it.

Well put. I tell you what, like, you've given me so much insight here. And I've learned so- ... much today. And my key takeaways are, first off, I keep going back to this, the transparency- ... blowing my mind. Being able to work with the communities- really show the benefits, really work with them- ... to make sure they h- [00:21:00] have- ... the trust. Learning about the whole closed loop system- ... and what that means. I mean, it shows you fill up a swimming pool worth of water- Yeah, yeah ... it's gonna last for five years. Like, we're not- Yeah ... just continually pulling water out of the ground.

Yep. Addressing the power concerns. And you gave me your best Troy McClure. Come on, man. Can't, can't beat it. Can't beat it. That's what I'm talking about. That's why we do the Empire of Lights is just for that. Simple as that, so. Well, I gotta tell you, Jonathan, it's a privilege to be here. Thank you for having me.

Fun stuff. I really appreciate you taking the time. I know you're busy. We're excited to have you here at the Tech Council today. Sounds great. Yep. Great stuff all the way. Brian, you are the best. In case you forgot, this is Jonathan Kersting with the Pittsburgh Technology Council, bringing you the Empire Lights.

We're trying to have as many conversations as we can around all things data center- ... because we need to have dialogue. Absolutely. Just like you said, Brian, so cool stuff. 100%. I'll see you guys on the next one