Omar Oronia: I think the risk profile of these assets, it’s really what makes it enticing for folks to want to dump capital into the space. With that, you’re also starting to see more aggressive terms in the market. Some folks are going to start pushing on whether it’s credit structure, just because it’s a proven business model at this point. People are comfortable with construction loans, people are comfortable with ABSs. Now it’s just a matter of where you want to play in the space, and what risk you’re willing to take.
Jeff Johnston: That was Omar Oronia, vice president digital infrastructure at CoBank, about the evolving data center financing market.
Hi, I’m Jeff Johnston and welcome to the All Day Digital podcast where we talk to industry executives and thought leaders to get their perspective on a wide range of factors shaping the digital infrastructure market. This podcast is brought to you by CoBank’s Knowledge Exchange group.
The data center market is going through a transformational growth phase driven by generative AI, and the capital being deployed is truly unprecedented. This growth has attracted a wide range of banks and institutional investors who all want to gain exposure to the market. This investor interest has also brought new financing structures that enable companies to build data centers faster.
Omar sits at the epicenter of data center financing and has a great perspective on where things stand today and where they are headed — which makes him an ideal guest to have on the podcast.
So, without any further ado, pitter patter, let’s hear what Omar has to say.
Johnston: Omar, welcome to the podcast. It’s a pleasure to have you on. How are you doing?
Oronia: Pretty good. How are you, Jeff?
Johnston: Doing great. I’ve been meaning to have you on the podcast for a long time, so it’s great that you’re able to pry yourself away from all the cool things you’re working on to spend a little bit of time with me talking about data centers because it’s certainly the topic of the day and AI and everything that’s going on and power and energy and investments. Super pumped to have you on.
Oronia: Oh, absolutely. Thanks for inviting me. It’s an honor and a privilege. I can’t extend my gratitude enough. Thank you.
Johnston: Awesome. Great stuff. Hey, Omar, I want to just start off high-level for our listeners. Maybe you can just give us a high-level overview of where things currently stand in, I’ll call it data center financing.
Oronia: Yes, absolutely. The industry itself is still at a massive build-out phase, with the continued growth of cloud and AI services driving the current supply-demand imbalance, with power being the critical bottleneck for the data center operators. Hyperscalers like the Microsofts, the Amazons, Oracles, Googles, Metas, they’re still self-building some of their data centers, but in order to keep up with all this demand, they have to outsource the majority of their build-out needs to third party operators and developers like the QTSs, Vantages, and cloudHQ and STACKs of the world and lease that capacity via long-term contracts ranging between 10 to 20 years. That’s not including extensions.
The supply-demand imbalance has led to the largest hyperscaler pre-leasing supply. There are very few instances where you’d be able to locate supply that’s already been built out and available when a tenant needs it. That’s pretty remarkable and telling of where we are in terms of still the demand-supply imbalance. These data centers deployments are only getting bigger. What used to be considered a large deployment was 1 megawatt. Now, we’re starting to see 1 gigawatt campuses. These massive deployments require equally massive sums of capital, which increases not only the size of the facilities but also the size of the risk profile.
Johnston: You made an interesting comment around pre-leasing. This is not like a field of dreams, build it and they will come, kind of thing. Can you just spend a little bit more time on that, Omar. What are these pre-lease rates? I guess I’ve heard of numbers north of 80%, maybe closer to 90 plus percent. How do those compare to where they were, call it a couple three years ago or so?
Oronia: Oh, no, you kind of hit the nail on the head with that. We’re seeing pre-leasing rates in the 80% to 90%, and this pre-leasing of data center capacity pretty much eliminates the risk of speculative building. That’s attracted debt and equity investors to come into the space, given these healthy returns and the well-understood risks relative to the other asset classes. Now you’re starting to see so much capital being pumped in because you have clear visibility to recurring, stable cash flows.
Johnston: It’s a pretty low risk, and you’re dealing with companies that have the strongest balance sheets in the world, right?
Oronia: Exactly. You’re seeing sovereign wealth funds, construction loans, ABSs, amongst other sources of capital, are becoming more common for sure.
Johnston: One other comment you made regarding outsourcing the construction of data centers and how you’re saying these hyperscalers are doing some internal builds, but then they’re outsourcing its external builds. Just to drill down on that one a little bit more, is the outsourcing, is it a function of labor, land availability? Can these hyperscalers just simply don’t have enough infrastructure themselves to meet their needs? What’s going on there?
Oronia: I think it’s a little bit of everything that you just mentioned. You have these operators and developers that specialize in building these data center buildings. When these hyperscalers want to deploy these IT capacity workloads there, they want to do it at a much faster pace, and historically, they were able to keep up with that. They were doing a lot of self-builds.
Now we have cloud computing and AI, which is driving tremendous growth in the industry, and now they’re wanting to go much faster than what they were doing historically. That’s why they outsource it to these developers who specialize in land acquisition, power procurement, interconnection, and basically have all the infrastructure ready for them so that when they move in, it’s just a matter of turning the key.
Johnston: Wow. Great stuff. Omar, I know you spend most of your time, if not all of your time, financing a lot of this stuff on behalf of CoBank. I wanted to spend some time on that and figure out what you’re seeing and how the financing landscape of all of this infrastructure is taking place. Maybe talk to us a little bit about these hyperscalers. Are they raising debt in most cases to pay for this infrastructure, or are they just leveraging the cash on their balance sheet and their cash flow? How do you see this stuff getting paid for?
Oronia: I see a little bit of both, whether it’s debt or cash flows, but from just the top two hyperscalers, if you want to rank them in terms of size like Amazon and Microsoft, they have tremendous balance sheets that they use to build out their infrastructure. The revenues and cash flows definitely support these lease payments that they’re on the hook for. That’s how we see that risk and why the counterparty risk on these transactions is those hyperscalers and their viability, which to this day, there hasn’t been any major concerns, and we hope it stays that way.
They’re tremendously well-capitalized companies that use their weight for sure to support these massive infrastructure build outs. Whether they finance that with their own money or debt, it’s a balance of both. Just their source of capital and cost of capital, sometimes they’re able to use cheaper debt to build out. If that’s the case, why not? The ABS market itself provides more favorable rates than what a construction term loan would provide. However, the ABS market itself has certain criteria, and at that point, when you’re financing a construction site, you’re not able to transfer that easily to the securitization market until it’s stabilized and completed.
Johnston: Actually, that’s a great segue into where I wanted to go next, Omar, just around these asset-backed securitizations, these structures that we’re starting to hear more about. Maybe you can just, A, explain how that structure works. Then, B, why do you think it’s growing as fast as it is?
Oronia: Oh, absolutely. The securitization market, more specifically, the asset-backed securities, have gone from niche to commonplace in data center financings and have been used to bridge gaps between capital needs and supply. That’s expected to continue growing and gaining traction due to the surging of data center demand, the refinancing needs, and the strong investor appetite in this space.
This market also plays a key role in the takeout of data center construction financing. As construction finance data centers reach completion and stabilization and start producing those recurring levels of stable cash flow, their fitness into ABS becomes more feasible, and then effectively lowering the cost of capital due to more favorable, longer-term rates.
Johnston: Do you think we’re seeing more ABS? Is it just because of cost of capital, or does it also allow developers who are building these data centers, does it allow them to take advantage of more leverage and build more data centers faster?
Oronia: Yes, absolutely. The ABS demand is growing in parallel with high-scale leasings. To give you an idea, total outstanding data center ABS issuances will reach an estimated $50 billion by 2027, which is double that of what was done in 2024.
Additionally, while the ABS market is continuing to grow and get wider attention, it’s not the only form of refinancing capital coming into the space. It’s projected that 60% to 70% of stabilized data center refinancings will be executed through the ABS market, with the remainder being split amongst mortgage-backed securities, private placements, and institutional term loan Bs. Everybody wants a piece of the pie.
Johnston: Yes, no doubt. I would think for the typical data center ABS investor, if you will, these seem to be pretty stable cash flows, highly rated companies that are behind these things. What does that investor profile look like?
Oronia: The investor profile, it goes from banks to your insurance companies and hedge funds, pension funds, is what we’re seeing now. I think the risk profile of these assets it’s really what makes it enticing for folks to want to dump capital into the space. With that, you’re also starting to see more aggressive terms in the market. Some folks are going to start pushing on whether it’s credit structure, just because it’s a proven business model at this point. People are comfortable with construction loans, people are comfortable with ABSs. Now it’s just a matter of where you want to play in the space and what risk you’re willing to take.
Johnston: Those investors, pension funds and hedge funds and others, there’s an awful lot of liquidity and capital to go around. It’s not surprising that the numbers are projected to get as big as they are.
Oronia: Yeah, you see even the CoreWeaves right out of the world that have got well capitalized with some of the private credit side of things. Now they’ve gotten scale and are publicly traded now. We’re able to refinance some of that higher-yield debt into more favorable rated debt, just to show the willingness for investors to dump money into some of these assets and tenants.
Johnston: You mentioned CoreWeave. I was looking at some of their financials. I feel like CoreWeave is a proxy into what’s happening inside of the hyperscalers in terms of the growth of their AI business. There hasn’t been a whole lot of visibility into that yet from a reporting standpoint, but my goodness, I think CoreWeave did $1.9 billion in revenue in 2024, if I have got it right. I think they’re projecting to do $5 billion in revenue in 2025. That’s up from, I don’t know, a few hundred million not long ago.
Oronia: Yes. Now they’re in tremendous growth mode.
Johnston: Yes. Hey, I wanted to talk about some of the things that have been in the media over the last six weeks or eight weeks or so ago. There’s been a lot of reports here that the likes of Microsoft and others have been canceling data center leases and walking away from some projects, which would suggest that there’s always this concern, this risk of overbuilding and considering the capital that’s being deployed.
I think some are worried about, have we already overshot the build, and do we have too much capacity? Is that why we’re seeing some of these reports about leases being canceled and things like that? I’d just love to get your perspective, again, from someone who’s financing these things every day. How do you think about where we are in, I guess, where are we in terms of looking into your crystal ball? Because I guess nobody knows, but where are we right now in terms of infrastructure build? Do you think we’re at a point where we could see a pause or a pullback even on spend?
Oronia: That’s a good question. The slowdown of leasing activity in some markets from Microsoft and Amazon earlier this year definitely made top headline news in the industry. However, investment in AI remains a priority, and it’s expected that other hyperscalers will absorb the capacity, especially if power infrastructure is procured. I think it’s also important to note that the recent slowdown in leasing activity from these hyperscalers isn’t quite out of the ordinary, and it’s typical and common in the hyperscaler lifecycle as they digest the leased capacity.
Meta did this a year or two ago and is now back to ramping up its data center deployments. Fun fact, the canceled leases by Meta at that point of time were actually picked up by Microsoft. Somebody will absorb that capacity. If anything, this pullback from Microsoft and Amazon is almost indicative of their shift towards inference, which is the monetization of AI. As far as build times go, access to power has become a growing challenge in the data center industry and in some of these markets, which is causing longer build times and higher costs.
The standard delivery time used to be 12 to 24 months, and now those timelines can easily extend to 36 months and beyond, accommodating for the commissioning of power. Regardless, if the capacity is built and if it’s built in primary markets, then somebody will take it, whether it’s Amazon, Microsoft, Google, Oracle, one of the hyperscalers will take it.
Johnston: Good stuff, good stuff. Hey, you touched on it a little bit. I want to go back to that, just around construction and what you’re seeing there. I think it used to be you could stand up a data center in a couple of years, but I think that’s changed. We’d love to get your thoughts on how things are going on the construction side.
Oronia: I think the construction itself, especially if you’re dealing with a powered shell data center, which is just literally connecting the concrete walls to power and connection, or whether that’s a broadband connection, that’s still pretty much your standard 12 to 24 months of construction. What’s really holding this up is the commissioning of power, and the power constraints are the largest limiting factor for data center deployments, especially in the U.S., and mostly around the limitations around transmission capacity.
Transmission projects are multi-year endeavors, and rate-based issues do not really incentivize these utilities to be more proactive in meeting the future data center demand. It’s almost like they’re always in a perpetual state of playing catch-up. Additionally, the data center deployments don’t necessarily just represent a utilities-only challenge in the sector as they’re grappling with the implication of the electrification of vehicles, semiconductor manufacturing, and then a shift towards green energy.
Regardless, there’s alternative markets, such as those in the tier twos and the tier threes that attract data center developers. Alternatively, there’s also energy sources that are being considered by some of these top hyperscalers in order to mitigate some of these constraints.
Johnston: It’s a fascinating situation. I’m just wondering, from a financing perspective, are developers able to get financing to build the shell, and buy the land, before they’ve been able to secure power? Does all that need to be done before a shovel goes in the ground or before you can get financing? Meaning, do you need to have a very clear path to power before you can finance a project and start putting shovels in the ground?
Oronia: Ideally, you’d want to have the power situation procured and all buttoned up before you start sending dollars out the door. The reality is that the power challenges in the industry are very fluid, and I feel like everybody’s just trying to figure it out. I think as long as you have a clear path to that power, and if you’ve been in talks with the utilities and you’ve at least gotten a load letter in place that says, “Hey, yes, we can provide this power to you. However, you have to get in line. It’s going to take X amount of years to get it.” Then that’s enough for some of these banks to leverage and feel comfortable around it. The reality is that this demand’s not going to slow down, and I think right now it’s just a matter of getting in the queue.
You see these headlines every day where it’s like, oh, there’s a Vantage or cloudHQ, whichever developer just announced a new massive development in one of these markets. They oftentimes already have the land-- They own the land, so at that point, they might’ve just gotten clear visibility into powering that site. Now they’re starting to have talks with the tenants because that’s attractive to them, because they have the land, they have the power. Then you start talking leases, and then once that gets far along, then they start talking financing, and then that’s when the banks come in. That’s where we come in and start looking at the opportunity and providing capital.
Johnston: That’s helpful. That’s great. Boy, Omar, I’m just sitting here thinking, you’re sitting at the epicenter of this transformational technology, this fourth industrial revolution. You have a pretty cool job, man. It’s super exciting time. Really appreciate all of your thoughts today. Super insightful. Hey, before we wrap it up, Omar, any closing thoughts you might have? Any questions I didn’t ask you or things we didn’t talk about that you want to address? The stage is yours.
Oronia: It’s just amazing to be part of this point in history, a part of this, like you said, technological revolution where we’re leading the AI race as a country, and we hope it stays that way. I think there’s going to be plenty for all of us investors whether we are a bank, a hedge fund, insurance company, provide our capital to participate in the space and hope to provide some of this workloads to more rural markets because at the end of the day, we are a cooperative bank and it’s in our mission.
We’re starting to see some of these deployments go into rural markets, and these deployments are enriching those communities with enhanced connectivity, jobs, and just overall community enrichment. I think that’s just something that’s awesome to be a part of, and I’m sure that you can resonate with that as well, right, Jeff?
Johnston: Oh, absolutely. Look, when I think about rural America and AI, I really truly believe that rural America is taking center stage in our country’s AI build-out. It’s super exciting for those living there because the opportunities that are coming their way, I think, are truly once in a lifetime, it’s pretty remarkable what’s happening. Hey, Omar, thanks so much for coming on the podcast today and sharing your thoughts. I really appreciate it.
Oronia: Thank you, Jeff. It was a pleasure.
Johnston: A special thanks goes out to Omar for being on the podcast today.
Speed and creativity. That is what comes to mind after my talk with Omar. The race to buy land, secure power and build AI data centers is on in a very big way. And making sure developers and hyperscalers have their financing lined up is critically important. It is also important that flexible and creative financing structures are available to enable builders to execute their aggressive plans.
Hey thanks for joining me today and a special thanks to my fellow CoBank associates Christina Pope and Tyler Herron because without them, there wouldn’t be an All Day Digital podcast. Watch out for our next episode.
