How can data center developers meet soaring demand for AI capacity without locking billions of dollars into buildings that may no longer fit tomorrow's workloads?
In this episode of Tech Talks Daily, I speak with Steve Conner, president of EdgeCore Digital Infrastructure, about the decisions sitting beneath the rapid expansion of AI infrastructure. Steve has worked in and around the data center sector since 1998, including the dot-com era and the later growth of cloud computing. That history gives him a measured view of the current rush to build large facilities quickly.
Steve argues that AI has intensified what he calls shiny object syndrome. The opportunity is large, but training, inference, and cloud workloads do not all ask the same things of a building. Rack density, floor loading, cooling, electrical design, available space, network distance, and proximity to cloud regions can all affect whether a campus can adapt when customer requirements change.
We discuss why EdgeCore has chosen to preserve flexibility in its facilities. A training-focused building might be made smaller because dense racks require less floor space, but future inference workloads may need a wider footprint. EdgeCore therefore accepts additional space in some designs, reinforces floors for heavier equipment, and enables liquid cooling even when a lower-density workload may not need it immediately. Steve presents those decisions as insurance against expensive retrofits or stranded capacity.
The conversation also examines EdgeCore's recently secured $1.5 billion in financing. Steve says the covered buildings were fully leased and designed to support mixed cloud and AI workloads. For him, the financing reflects continuing demand both inside established cloud regions and in surrounding markets, while the mixed-use design gives the customer options as requirements develop. These figures and interpretations remain EdgeCore's account of the investment.
Site selection is another major part of the equation. Power availability receives much of the attention, but Steve adds network distance, workforce availability, long-term political support, and relationships with utilities and local authorities. He describes looking beyond crowded locations such as Ashburn while remaining close enough to established cloud regions to support different use cases.
For me, the most valuable part of the discussion concerns communities. Steve says developers should begin meeting local leaders and understanding local needs before purchasing land. EdgeCore's examples include support for chambers of commerce, first responders, hospitals, fire services, and workforce development. His argument is that a company cannot simply purchase goodwill after construction begins. It has to be present early and demonstrate that the relationship runs both ways.
We also address public concerns about water, emissions, energy demand, and jobs. Steve argues that many modern data centers use cooling systems that do not consume water for routine cooling, though his comments apply to the facilities and designs he knows and should not be generalized to every data center. He also notes that AI facilities consume substantial power while arguing that developer-funded transmission upgrades can benefit other users of the grid.
The episode closes with a wonderfully plain analogy. Steve describes the data center as the plate rather than the meal. The infrastructure serves whatever workload the customer needs, which is precisely why the plate must be designed for a menu that keeps changing.
Are developers doing enough to prepare AI data centers for changing workloads while earning the confidence of the communities around them? Listen to the episode and share your thoughts with me.
Useful Links
[00:00:00] Scale your business with agentic AI, with limited risk. With Denodo's AI data layer, your agents are provided with real-time company data and guardrails for company protection. Create the business you always dreamed of with Denodo. And you can do that by simply visiting denodo.com to learn more. But now, back to my guest.
[00:00:27] What happens when the data center built for today's AI boom becomes tomorrow's stranded capacity? Well, Steve Conner, president of EdgeCore Digital Infrastructure, joins me here today to explain why speed-to-market can no longer carry an infrastructure strategy on its own.
[00:00:46] EdgeCore recently announced $1.5 billion in financing for two Northern Virginia data centers that are designed to support 114 megawatts of load. Great Scott, as Doc Brown might say. But seriously, today's conversation will look beyond the headline numbers.
[00:01:06] Steve will explain why operators must plan for changing rack densities, heavier equipment, liquid cooling, inference workloads, energy constraints, and communities that expect developers to arrive as partners. And he even has a data center visible from his backyard, so he's more than qualified to talk about this stuff. And he also has a data center visible from his backyard, which is one way to make sure your work follows you home.
[00:01:35] But listen out today, we've got practical lessons in building flexibility before billions of dollars become concrete. So enough from me. Let me introduce you to Steve right now. So thank you for joining me on the show today. Can you tell everyone listening a little about who you are and what you do? Yeah, sure, Neil. So my name is Steve Conner. I am the president of EdgeCore Digital Infrastructure. I basically run EdgeCore's back and front components of the business.
[00:02:04] So I'm responsible for the sales team, operations, and a lot of other things that just make the business run. Historically, I've been in and out of the data center industry since 1998. I started with, I would say, data center 1.0 with Exodus Communications. When the dot-com bomb happened, I left for a bit. And then I came back in 2018, joined Vantage. Spent seven years there. And then when Lee came over to EdgeCore, he brought me over, and we've been running ever since.
[00:02:34] I'm so glad you had time to sit down with me today. There's a lot I want to talk with you about. And I'm glad that you mentioned data centers there because there is a lot of noise around AI right now. But obviously, they put data centers right at the center of the technology conversation. And you must have seen so many changes throughout your career. And I was also reading that you've argued that the leadership playbook is changing just as quickly as the technology itself.
[00:03:00] So what is different about building infrastructure for this AI era that we all find ourselves now compared to the previous waves of, I don't know, digital transformation, mobile, or cloud growth? Yeah, I think with AI, it has introduced a lot more of the shiny object syndrome. You know, in the past, you know, from the 90s through the aughts in the teens, we had growth. And it was substantial growth.
[00:03:29] But it was something that I think was paced. And as leaders, you could work through, I would say, diligent growth. Now with AI, there's a lot of opportunity to go out there and just really build big things really fast and really just cater to the moment. And I think as leaders, we have to pay attention to that. But at the same time, we also have to think about pasting.
[00:03:56] What's the work going to look like in a year, two years, five years? AI is definitely going to be here. But as we've seen in just a span of two years, it's going to change. So we've got to think about, you know, is it going to, how's training going to exist? How's inference going to exist? And still, cloud's going to be here. So from our standpoint at EdgeCore, we've been thinking about how to look at the data center positions, the actual campuses,
[00:04:23] and pay attention to not today, but two to three or four years from now, and try to position those campuses such that they can meet any of those AI components that we just talked about, but also potential cloud components and any other twists in the road into the future. So many great points there, because I think for years, speed to market was seen as almost the ultimate competitive advantage. But again, you've suggested that even that isn't enough anymore.
[00:04:52] So why is anticipating future demand become more important than simply just building capacity as quickly as possible? So I think there's a couple of things that we have to pay attention to there, Neil. And when we look at just the anticipation, I'll give you three things that we have with eDevelop. One, we've got to pay attention to, you know, the change in technology. There's going to be changes in the capacity of the racks and the workloads themselves.
[00:05:20] So as we look at designs, we have to start paying attention to, you know, larger densities, heavier racks. And as a developer, as a real estate developer, I want to do my best to build my building such that it's not built for just one use case, but potentially two or three use cases in the future. That's a hard task. But if you spend a lot of time listening to your customers and spending time with their engineers, do that. So that's number one.
[00:05:47] Number two, we've got a lot of press coming out around data centers. And, you know, we as an industry have the ability to go where we're wanted and help people want us at the same time. So as we look at, you know, the AI movement, we have the ability to go in and do a lot of forward transactions with the community before we even think about purchasing land.
[00:06:12] So if you do that long play there, you can actually build up that goodwill with the utilities and the community and make sure that when you land those data centers, it's something that the people want rather than just something that they have to tolerate. And then lastly, you know, speed has always been a thing that we've looked at in the past. But now with power being something that is constrained and will probably be constrained for a good portion of the next decade,
[00:06:41] you really have to start thinking around how are you going to find sources of power? And that may be generation, new ways to connect it to the grid, and really pick your spots and make sure that you're working with the utility and ultimately working hand and globe with them to hopefully do kind of on-site, in front of meter activities or other things like that that can help you be prepared for those future movements that AI may bring to us.
[00:07:11] And of course, every conversation around AI infrastructure will eventually come back to power. So how are leading operators thinking differently about energy availability, grid constraints, and sustainability when planning that next generation of data centers? And I would imagine this is something you're passionate about as well. You're living right in the heart of this space too, right? Yeah. So I'm going to speak for export, but I think it's pertinent for almost every developer now.
[00:07:41] And I think we've seen that in the press a lot these days. When we look at siting these days, when I was working in the data centers, even in the early 90s, but even in the early late teens, we assumed that power is going to be available. And we treated our utilities partners as just transactional entities that we would work with.
[00:08:03] But as we advance to today, where power is scarce, either via transmission issues or in some instances generation issues, we are starting to think as an entity how we partner with the utilities first. So rather than just treating them as a distant counterpart, we get in early and start working with them as partners
[00:08:26] to start anticipating what our loads are and making sure that we're understanding the grid impacts that they're looking at as we work through what we need as we build on that. And we're doing that right now with that short. The other thing that we're doing as an industry is when we look at sites, it's no longer how close are you to a substation or to a switching yard. You have to do that still, but you're also looking at how close am I to a generation source or how close am I to gas lines,
[00:08:55] whereby you can work either by yourself in some instances or with the utility to intentionally bring your own power and either behind the meter or in front of the meter, a generation would own it. And that's something that we did at Vantage when I was there. And it's something that everybody's contemplating now. So you have to be constantly thinking about different sources of energy, how you're going to get that energy to your site.
[00:09:23] And when I speak to people, I don't talk about power anymore. I talk about energy because I'm talking about molecules. I'm talking about electrons as well. So we're looking at energy when we think about it from an export standpoint, and it could be whatever source of energy is. And I would imagine that site selection has become far more complex than Hades finding available land.
[00:09:48] So what factors now determine whether a location will still be the right choice 10 or even 15 years from now as AI workloads will inevitably continue to evolve and demand more power? What goes into that? What should they be thinking about? I mean, that's a good question. I think we look at sites. I'll speak for experts. We look at sites.
[00:10:12] We obviously start with that tenant that I put forward where we're looking for places, geographies or jurisdictions that are currently pro-data center. We are anticipating that they be pro-data center to the long part. And we'll put a lot of time and energy with the community. When we're looking at sites in those communities, we are then looking for a couple of things.
[00:10:35] For EdgeCore, we're looking for a network distance that can be meaningful regardless of whether it's a cloud component, inferencing issue, installation, or it's a training instance. So we've been looking at sites that are quasi-approximate to cloud regions. They'll have to be in the region. Ashburn's very crowded. Northern Loudoun County is very crowded.
[00:11:03] So we've been looking a lot further south. So that's what we've been doing. We also, when we're looking at siting these sites, we're looking at communities that would appreciate the new jobs that we're going to bring to that equation, but also those places that can support those jobs. I think we'll probably talk a little bit later about some of the things we're doing about workforce development, things like that.
[00:11:29] But we at EdgeCore, we're very pro workforce development. We've got a couple of programs that we can talk about a little bit later, whereby we're building our workforce. So we want to make sure that the communities we're working in can support that growth inside the community. But to summarize it for you, we have to have access to power, obviously, or energy one way or another. We want to be in a community that's going to work with us as we grow.
[00:11:55] We want to be in a place where we have a workforce that we can work within or develop. And then lastly, we want the site to be quasi-approximate to or proximate to a cloud region so we have optionality with respect to the workforce. And it's also an incredibly exciting time for you guys.
[00:12:14] I was reading before you joined me today that EdgeCore recently secured $1.5 billion in financing to support hyperscale AI-ready facilities. So what does an investment like this tell us about where customer demand is heading and also how the infrastructure of requirements for AI are different from some of those traditional cloud deployments?
[00:12:38] So that $1.5 billion that we secured is interesting in itself in that it's, one, those buildings were fully leased. So we have that in the equation for this scenario. And it's interesting in that both those buildings are mixed use case workloads.
[00:12:58] So what it tells us, I think, and look at that type of investment is, one, you know, there's a lot of appetite out there for capacity, both in the cloud regions and outside the cloud regions. We've seen it in state, across state, across state. And two, it tells us that in places like Ashburn, that there are opportunities for both an AI workload and a cloud workload to still exist.
[00:13:23] We've made those buildings that that $1.5 billion is covering capable of supporting both use cases should the customer want it. And I think that was what was kind of a hidden tagline in that $1.5 billion is simply that it is allowing that customer a lot of flexibility. But you're going to continue to see a lot of investment across the board for AI.
[00:13:49] You're going to continue to see a lot of investment across the board for inferency, specifically, I think. And regulations and increasing community concerns are becoming increasingly important, especially as data center development begin to accelerate. So how can operators balance that need for rapid expansion with responsible development? And most importantly, those long-term relationships with local communities as well.
[00:14:15] Yeah, I think this is something that the industry is starting to come up to speed on. Edgecore was an early actor when it came to community development. And having been in the business for a while and having sites that were generally just zoned for industrial, we would just follow the rules. We would go in, get the sites, and just build based on code.
[00:14:42] At Edgecore, what we did early on in our existence is bring in somebody who was part of community development. Our lead of public policy was an economic developer. And what we do to make sure that we can get in and build those relationships is before we even think about buying a piece of land, we look at the counties, the municipalities that we think that are good for us, and we go in and start doing that work.
[00:15:11] We get in and start doing the legwork, understanding who the city councils are or supervisors, and then understanding what their needs are. It does require a couple years of upfront work in order to get those things done. This is why I said, you know, in the beginning, we're talking about how leaders have to think differently. And by being there and being present, that's the key in the list. You have to be present. You can't just buy your way out of this. You have to be in the community.
[00:15:38] That sets you up for when you do buy that piece of property or you do, you know, you start building that building. The business, EdgeCore, has already put time and energy into the community, Silver Noon. So we've done that in all the genres we were in. Mesa, we're one of the largest contributors to the Chamber of Commerce there. In Story County, we're known throughout the county for supporting their first responders.
[00:16:04] And in Louisa and Culpeper, we're in the process of helping with different aspects of their hospital system and their fire department. So I think as a community, from a data center community, we have to go in and start prioritizing people, prioritizing education, and getting in early before we start building. And that should help quite a bit.
[00:16:30] And I was also reading one of the risks that you've highlighted is stranded capacity and investing billions in infrastructure that no longer matches market demand. So how do you build enough flexibility into today's decisions when AI technology is evolving so rapidly? So we listen to the requirements. So as a company, I'm speaking once again for Edgecore.
[00:16:55] We spend a lot of time with our customers engineering teams, just understanding, you know, what they're thinking and what they want to think about for the future. You know, we'll look at the NVIDIA reference architectures, although those are super challenging to design too. But we'll spend time with students alike. So the Microsofts, the Anzots, and the Googles, and things like that. And just look at the, get forward looks on their lines.
[00:17:21] And what that allows us to do, Neil, by really just doing that deliberate exercise with the engineering team is get a feel for how they need to do it. So when I stated that we had this opportunity for stranded capacity, there's two aspects of that. One, you know, when we look at what the industry has done to date with these large learning model facilities, they are designed specifically for those models.
[00:17:49] So in the future, if something were to change there, the retrofit associated with those facilities could be enormous. And, you know, as a developer, we have shied away from those things. And then secondarily, going back to the engineering, we are building in aspects of AI flexibility into our buildings that will allow us to flex. And I'll give you an example.
[00:18:20] The natural tendency in an AI building is to shrink the building. The densities of the racks are such that you could just make the building smaller. We are actually doing the opposite. We know that that building may need more space in it to support future workloads because inferencing workloads are a little wider. So we've made the decision that it's okay to sacrifice a little building space, make the building a little bigger, to make it such that that building can support a different use case in the future.
[00:18:49] And then also, in parallel to that, we understand that those racks may be heavier. So we reinforce the floors by understanding that, you know, if we're doing inferencing, floors are going to be overbuilt, but that's okay. And then lastly, when it comes to the buildings, we've integrated a lot of the bells and whistles around liquid cooling enablement. If we don't use it, it just makes it easier for us to maintain SLAs for a lower use case.
[00:19:17] But if it is a heavy use AI model, we're prepared. So by, you know, not building to a specific use case and then by paying attention to what the engineers are telling us from a requirement standpoint and being willing to, you know, make a few sacrifices that will allow for flexibility.
[00:19:40] That's what's giving us the ability to flex, I believe in the future and make our buildings, speaking for export, a lot more adaptable. And for everybody listening, anyone that scrolls down their news feed on a regular basis will see that everybody's got an opinion on AI and data centers and infrastructure.
[00:20:02] And I'm curious, as someone working right in the heart of this space, is there an area of AI infrastructure that you think doesn't get enough attention or equally any myths in the industry that you might see repeated online? Anything that spring to mind there? Well, I think when we talk to, you know, when we listen to the myths, I mean, one of the things that, you know, kind of raises my hackles, Neil, is the water conversation. Yeah.
[00:20:28] I think I've read a number of times where, you know, one AI transaction, so you say, you know, five million dollars or something pretty similar, right? In today's data centers, now I can't speak for passive systems because they did do things differently back in we've had teens and early aughts. Yeah. Most of today's data centers don't use water for cooling. We have cooling systems that are very similar to your home air conditioning. We don't put water in our home air conditioning units.
[00:20:58] They just, we just put them in and they work. We've got two portions of it and away we go, right? Yeah. I get a center operates the same way. There's just, they're just bigger and I've got a lot of units on my house, right? So whenever we're doing things like that, an AI transaction costs in most modern data centers, zero water because of it's your home air conditioning unit.
[00:21:19] So those myths like that and, you know, the pollution and things like that, we don't run our generators rarely ever at all unless, you know, we have a power outage, which is rare in most instances. So I think, I think those are two of like, like the environmental ones that I just, I would love people just to take a minute and actually, if they ever want to come toward data center, I'm always happy to lose.
[00:21:43] And then I think from AI workloads, when we look at these models and what they're doing and things like that, a lot of people are concerned about, you know, these models taking over the world, taking people's jobs. I use AI a lot in what I do and what it allows me to do is aggregate a lot of information pretty quickly, but it absolutely is not something that can take away what I do because that information needs to then be interpreted and managed by a computer.
[00:22:12] So I look at AI as an enabler for anybody in any community to just do things a little bit quicker. And then I think lastly, when we look, look at, you know, those workloads and, you know, they're going to suck all the energy out of the grid. I mean, it does consume energy. I mean, that's something that we just have to accept, but these AI workloads in our communities actually ultimately help with a couple of things.
[00:22:41] With respect to energy and infrastructure. AI workloads require a lot of energy. They require upgrades to transmission. And these are things that we in the data center industry are having, we are paying for now. These are things that are part of the bill. But that infrastructure is used by everybody, whether you're, you know, manufacturing plant, whether you're a car dealership or your home. So everybody benefits from the equipment.
[00:23:09] And then because we have steady loads, it helps the generation organizations make sure that they have, you know, a steady capacity. And that's their biggest challenge. So, you know, there's a lot of things that, you know, people worry about with AI. But I think at the end of the day, as people get more comfortable with AI and the fact that the data centers are just the platform. We're the plate. Like the data center is the meal. You can't have any meal without a plate. So these ones are just plates.
[00:23:39] We just serve up whatever our customers want us to serve. But as AI gets to be more, you know, adopted and people get more comfortable with it, I think people are just the rumors and the myths with StarCline. And I would recommend anyone listening that wanted to educate themselves on this. I know you do a lot of great work here on your LinkedIn channel, for example. I think it's Density Digest.
[00:24:07] A lot of frequently asked questions and education pieces there. And for anything else, people listening wanting to learn more about what you're doing here, where can they connect with you and find out more about Edgecore? So if anything that we talked about today for Edgecore, please, you know, obviously visit us at, you know, edgecore.com. There's a lot of things that we publish there just from a thought leadership standpoint, information on our campuses.
[00:24:34] And, you know, we've got a lot of people on our team that have tons of time in the industry. And then, you know, you can hit us up on our LinkedIn page. And then if you want to see some of the things that I've written, you can hit me up on LinkedIn. My LinkedIn link is Steve Conner, C-O-N-N-E-R-V-A. And I cover a lot of topics by itself. Awesome. Well, we've covered a lot in a short amount of time today.
[00:25:00] Anyone listening wanting to find out more information about all things data centers and the great work that you're doing there, I would put those links down, urge people to check that, and let me know. Let me know your experiences as well. And we'll keep this conversation going. But, Steve, a big thank you for starting it today. Really appreciate your time. Neil, completely my pleasure. Thanks for the opportunity. I think Steve's message is that foresight has become a design requirement.
[00:25:28] An AI facility built around one narrow workload might be fast to market, but it will be expensive to adapt when demand changes. But Edgecore's response includes larger buildings, reinforced floors, liquid cooling readiness, several energy options, and community relationships that are developed before land is even purchased. And I think the wider lesson here applies beyond data center.
[00:25:56] Speed creates value only when today's decision preserves tomorrow's option. And I appreciated Steve's reminder that infrastructure developers cannot just buy their way into public trust. They have to show up, understand local needs, and remain present. So a big thank you to Steve for joining me today. You can learn more about Edgecore at edgecore.com. Follow Steve on LinkedIn. And reach me at techtalksnetwork.com.
[00:26:26] But over to you. Would your current infrastructure plan survive a sudden change in workload or energy availability or customer demand? Something for you to marinate on as I walk off into the sunset. Thanks for listening, as always. Speak to you soon. Bye for now. Bye for now.