Wes Cummins, Applied Digital CEO (Full Interview) | SD Focus

Sure. So I grew up in a small town in Idaho. I actually grew up on a potato farm set a little bit outside of town. The town was about 200 people. My graduating class in high school was 24 people. So I grew up in the same type of community in which we're building a lot of our projects in North Dakota, and have a really good feel for how small communities work. And it's actually a really big point of pride for me to bring the cutting edge of technology and the opportunity that that brings to small rural communities, just like the way I grew up when I grew up.

You know, we were, I would say our town was one of the last ones in America probably to get high-speed internet. And so that was kind of, you know, I grew up through that with dial-up. I think my parents had dial-up internet until 2015. So we were really left behind in what was happening as far as the technology evolution in the world. And so to be able to be actually on the front line of that and bring the benefits and the large benefits that we bring, and how well they integrate into the communities in which we work, has been a real source of pride for me and kind of a full-circle moment. And finding what fits really perfectly for small communities. So we don't have a big disruption, just a small enhancement. I would say big enhancement, actually, in the benefits we bring.

Sure. So the evolution of data centers over the past few years, frankly, this is what creates the opportunity. So the data center of the past is in. And what we've seen since the 90s really roll out across America and across the world were data center footprints, typically built in population centers that served all of the applications that you use today. I don't think people fully understand how integrated their lives are with data centers. Almost any application you use on your phone, anything now you use on your TV, the way you do banking, the way our banking system operates, the way in most cities the streets operate, all of these things are operated out of data centers historically in cities, in city populations, high population density cities is where these data centers were built.

And now, as we move to the next evolution of compute-focused data centers, so we call them HPC or high-performance computing, and really what these are focused on are generative AI applications. They can be traditional cloud applications in video streaming or serving up your office applications. They can do actually both of those things. But the uniqueness of this workload allows us to go to more rural locations, more remote locations, because the latency requirement is not as stringent as it was in the previous generation of data centers. So two things have happened. We were able to go to different locations, which is what you're seeing with how our company is building. So we were building two locations in North Dakota, and then the other part of it is the data center scale is typically bigger from a power perspective, not necessarily from square footage, because you actually use more power per square foot inside the data center than you have historically. But from a power perspective, the footprint has gone much, much larger. So if you roll back 5 or 10 years ago, a 10 or 20 megawatt data center was actually a pretty large data center build. But now we're looking at 100-plus megawatt data center builds in these locations, and even potentially larger than that, because that's the way the world is going. But it's a more recent phenomenon, just in our ability to move this type of infrastructure into the locations that we're moving it into.

Yeah. So typically what we find, and it's the way we find sites overall, is where is power available, and typically where is access to unused power available on the network. So when you look at Ellendale, North Dakota, our first really large-scale AI data center, HPC data center development in North Dakota, what we found there was a substation where roughly two gigawatts of wind farms feed into that substation. And when the wind is blowing, there's actually not enough transmission capacity to take all of that power to an end user. And so oftentimes what you would see there before we arrived is the owners of the wind farm would run the node negative from a pricing perspective, and then they would curtail the wind turbines. So you would have basically all of this electricity generation that went unused, or potential electricity generation that went unused after they curtailed the wind turbines. And so that was a perfect setup for us. We were able to actually bring the load to the point of aggregation. It didn't cause MDU to build any large infrastructure that was a burden to the ratepayers. We located right next to the substation. And then what we've seen there over the past 3 or 4 years that we've operated is we've actually saved ratepayers money every year. It was 5.2 or 5.3 million the first year we were there, 5.8 million the second year.

Well, this was electricity again that was going unused, excess capacity. And then we come in, we locate next to it, we use that electricity that wasn't being used, and then we paid the tariff to MDU for delivery of that electricity. But they didn't really have to build much infrastructure for that because we're using it all at a single point. So, you know, imagine how much infrastructure, power lines, you would need to build if you were serving that amount of houses over rural North Dakota, right? There would be a massive investment from the utility. They would need to get a return on that investment. That would be harder. You know, the ratepayers would see that because the offtake is not paying for that, so it gets spread across the rate base. But when you're dealing with a project like ours, not much infrastructure gets built, but we pay a lot in tariffs to the utility that then gets shared back with all the users. So if we can more efficiently use power, the network, more efficiently use the infrastructure if it's already there, then we can actually be a big benefit to the network from that perspective.

But so that's what we found at Ellendale. That's what we have found at Toronto. That's what we found in Harwood, which is near Fargo. Those are the types of locations that we look for that are going to be mutually beneficial to everyone, and that have really the capacity that we need from an electricity standpoint. So that really is what drives us. So the secondary driver is fiber, fiber optic communications available nearby, or will take minimal effort to bring them there. For Toronto specifically, it's actually a lot more work to bring in there than it was for Ellendale or Harwood. But given what we've already done in the region, it's kind of made that site more attractive from a fiber perspective. So it really starts with power, then fiber. And then I always say that at the end, I want to operate in communities. Our company wants to operate in communities and places that want us there, that want the benefits that we bring. That's another big part of what we do. You know, Ellendale wants us there. They want all of the tax revenue that we bring, all the things that we'll do for their community, for their schools, for all of those. And then during construction, there's a thousand-plus jobs on site, but that's not permanent. What we end up bringing is a few hundred jobs that are permanent, that are high-paying jobs. We train locally for the vast majority of those jobs. So again, another big benefit economically for communities that have historically been shrinking. It's not overwhelming. That's why I said we found the exact, I think, right match for small communities. And what we bring, big dollars from a tax base perspective, jobs, the right number of jobs. So it's not overwhelming. High-paying jobs. We typically pay two and a half to three times the average household income in North Dakota for the job. Train everyone locally. We source locally for all of our contractors for actually building sites, not our GC, but all the subcontractors. So we try to keep all of the benefits there in the communities that we operate in. They fully recognize all of those benefits that we bring. And so power, fiber, and then someone who wants us there.

So I would be very transparent on this. There is more traffic. There's just no way to get around that. There's more traffic. We work locally from a traffic perspective, making sure we do the absolute optimal routing, but it is going to be somewhat of an inconvenience. But like we take even into consideration things like when is school drop-off and school pickup if we're in that vicinity, right? So that we make sure that we're doing everything we can. You're always going to have things that don't go the right way, but generally we do things very well from that perspective. But there is that burden that's pretty short-lived for the long-term benefits, because once these are operating, you'll barely know that they exist. It would, you know, pretty much like a warehouse, doesn't really create any noise. Then you have a few hundred people that come and go every day. Light traffic from that perspective. So once they're there, it pretty much feels like it doesn't exist in the location. But there are those things, we do work through those as much as we possibly can.

And then when you look at infrastructure, one of our key focuses is making sure that we're an improvement to all the infrastructure of wherever we land. So what does that mean to us? First responders, for sure, right? That means there's a lot more funding, either directly from us or your tax dollars that go for first responders because we need that. But it also improves what you get for the community as well. Communication networks are going to improve dramatically in the location because of the type of fiber that we need. So all of the services that come along with that and all the benefits that come along with that, especially typically for rural communities that don't get those types of telecommunications infrastructure, we bring that to the location. And then on the electricity side, like I said, we've worked within the system. Despite what we've done previously, we've been a benefit to the system. And then the type of installations that we build actually are really unique because they have full backup power redundancy on site. What does that mean? It means, you know, we have commitments to our customers to make sure that these sites run. We have leeway to have them down 23 seconds a year. Right? It's a really short amount. So the systems are extremely robust and redundant to the point that they actually can go completely off grid and run for a time like 50 to 70 hours. So when you have a big strain on the grid in, say, a big snowstorm or a big heat wave, we could actually take our facility completely off grid. It'll keep running. We can turn that on, like any electricity we're using, or a portion of it, or whatever the grid can handle, back into the system so that for other users that would need that. So we create this extra resiliency inside of grid operations. So there's a lot of benefits that we do bring. But there is a bit of a headache for sure on the upfront when we're doing construction from a traffic perspective.