← Currents
Currents
Ep323: Data center financings: Part II
00:59:11
Four industry experts discuss about the latest trends, challenges and opportunities in data center financings. The four are https://www.linkedin.com/in/claus-hertel-479619/, managing...
Loading summary
Transcript65 lines
- [00:00]
A
Foreign.
- [00:05]
B
Welcome to Currents, a Norton Rose Fulbright podcast. Today we have for you a rebroadcast of a recent webinar that we held for our clients. The webinar is a discussion among four industry experts who are discussing the latest trends and challenges and opportunities in data center financings. The four panelists are Klaus Hertel, managing director at Rabobank, Melee Elleri, Chief investment officer at DC Blocks, Mike Johnson, a director at SocGen, and Tim McGuire, vice president of Capital Markets at Rowan Digital Infrastructure. Our moderators are my partners Christine Brzezinski and Christy Rivera.
- [00:44]
C
Thank you everyone for joining. I'm Christine Brzezinski, project finance partner in Norton Rose Fulbright, based in New York, and I'm joined by my colleague Christy Rivera, also a project finance partner at Norton Rose Fulbright, based in New York.
- [00:57]
D
So we had a really great webinar in April with this panel, and we were really lucky to have all of them agree to come back. For part two, we have Klaus Hurdle at Rabobank, Mali O' Leary at DC Blocks, Tim McGuire with Rowan Digital, and Mike Johnson with SocGen. I'm going to go ahead and let each of them introduce themselves and their companies. Klaus, can you kick us off?
- [01:20]
E
Yes, indeed. First of all, thank you, Norton Rose Fulbright, for inviting me back for part two of this data center financing panel. I had a really good time on part one, so I think this will be a great discussion. Just as background, at Rabobank, I head up the energy and core infra coverage team for North America. And basically we cover power, renewables, and on the infrasign digital infrastructure, which is primarily comprised of data centers, which has been a really busy part of the business for us in the last couple of months. We're basically product agnostic in terms of the sector. We do everything from PF to warehousing to ABS to capital markets and hedging. So we try to find the best solution for our clients in each of those sectors. I'm part of a global team. We have a team in Europe and a team in Asia, and. And the team here in North America comprises around 40 to 45 professionals. Thank you guys for having us.
- [02:20]
D
Great, Mellie.
- [02:21]
F
Hi, everyone. Thank you, Norton Rose, for hosting us today. My name is Mellie Hillary. I am the chief investment officer of DC Blocks. We are a hyperscale data center platform focused in the Southeast United States and looking forward to the discussion.
- [02:39]
G
All right, thanks.
- [02:40]
D
And Tim.
- [02:40]
G
Hi, everyone. I'm Tim McGuire. I'm the Vice President at Rowan Digital Infrastructure, hyperscale data center developer. I'm responsible for all of Rowan's capital markets activities as well as other strategic finance areas including evaluating new investments, risk management, deal structuring and hedging. Excited to be back for this encore presentation. I'm sure it'll be a great discussion.
- [03:03]
A
And Mike, good afternoon. Appreciate the Norton Rose team for having providing the opportunity to have me back on the podcast. It's been a really exciting period in the industry since we've last spoke. I think we've seen record volumes and some really interesting structures come to market. So looking forward to today's discussion. As a background, I'm a director on SG's TMT Finance and Advisory Team. We're a full fledged TMT and infrastructure financing platform that's been active in the space, very active in the space for the past six to seven years. Team here really looks through to core assets and is focused on bringing a full capability, full suite of capabilities to our clients sort of throughout the whole life cycle of assets in the TMT digital infrastructure and core infrastructure space. And that includes your regular construction finance. Through Project Finance we have leveraged finance capabilities as well as the full ABS CMBS private placement capabilities as well. And we're working closely with ECM and hedging to really bring together a full suite of solutions for our clients in the sector. So again, really excited to be on and thanks for having me join.
- [04:19]
C
Great. Well, we had a great webinar in April. It generated a lot of Q and A that we unfortunately didn't have time to get to. So the questions in this panel largely draw from that Q and A. So let's kick it off then with some questions about general market activity. Klaus, this one we're going to start with you on this one. Just talking about uncertainty in today's market and how that's impacting the ability to get financing. Do you see banks and private credit still staying on the sideline or do you just see higher financing costs at the moment? And can you talk about market volatility generally and how that's impacting the financing environment?
- [04:59]
D
Thanks.
- [04:59]
E
Thanks Christine. I think in general there's more market volatility around renewables and quote unquote regulatory risk. I think some of that has subsided since OP3, but I think there's more volatility around there. The Saudi Renova's mouse switch over, I think on the data center side, I think it's been fairly stable. As Mike just mentioned, there's been a lot of activity. There's been very large deals that are in process or have closed in the past three months, and I think that's still going to go on for the next 12 to 18 months. A lot of these deals have investment grade characteristics and in some cases actual explicit ratings, which opens it up to a wider investor universe. So if you think about banks, banks don't need a rating. They like to see a rating. We love ratings, especially good ones for institutional capital. And private credit doesn't need ratings either, but it certainly helps and it certainly opens it up for institutional investors. In terms of financing costs, I think they've been fairly stable in the low to mid-200 over SOFR. So I don't think we've seen too much movement there, especially given the large amounts of capital that have been required for some of these deals. From my perspective, I think the volatility kind of goes around the competition for deals and kind of clear market is maybe my personal biggest concern. There's a lot of competing deals in the market and there's, you know, there's not an infinite source, you know, source of capital. So my question is, how does the market absorb all this paper? You know, one of the kind of things coming around, around the corner is I think, recycling of capital because there's deals that have been seasoned now, projects that have been stabilized. So I think, you know, portfolios of projects can be refinanced to the ABS market or the RBS market. So that can free up balance sheet capacity for existing lenders. So I think that's one of the bright spots on the horizon. But overall, I think good sponsors with financial backing, good financial backing will attract the needed capital for projects over the near to medium term. I mean, the other kind of internal bank processes, we've seen lenders kind of trying to figure out how much of this paper you kind of want on your books, right? So these are huge amounts of, these are huge amounts of bank loans, right? So if you, if you have a project finance book, you know, maybe Your book is 10 billion. How much of that do you want to subscribe to? Data centers. If you're a real estate lender, you know, maybe Your book is 100 billion if you're a US real estate lender, so you can possibly do a bit more. So I think internal credit committees and kind of internal portfolio management decisions, those are kind of things that, that will come into play over the next year.
- [08:07]
C
Maybe.
- [08:08]
D
Let's, let's talk about on the hyper, hyperscaler side. Do you see? You know, and Tim, I'm going to direct this one to you. With everything going on in the market Good or bad. Do you think we're seeing any pullback on the hyperscaler side? Are they playing back on their AI capital expenditures?
- [08:25]
G
No, we haven't seen that at all and I think in fact there's evidence for the opposite. I'm sure a lot of people are aware of the deal that Meta recently did with Blue owl to place 20 odd billion of rated bonds. That deal was very well received and I think is actually trading at a significant premium in the secondary market. But I think we'll see more and more of these megadeals because the appetite from the hyperscalers for projects that can deliver and energize, you know, by 2027 is just very, very strong. All these businesses are, you know, very profitable in their own right. Speaking of the, the hyperscaler business models, these aren't companies that are kind of all in on, on some AI model that's yet to, to. To. To f. To figure out how to monetize that, that AI interest. These are, are, are our businesses that have a tremendous amount of, of profits coming in regularly and could those AI or cloud expenditures just from internal cash flows? So we, we haven't seen that slowing down at all. On the lender side. We haven't seen any evidence of a pullback either. I think more people are getting up to speed on our space and there's just a growing awareness that hyperscale data centers in particular really are core infrastructure and not niche real estate asset class. So we've seen those big financings get absorbed pretty well. You know, there's a, a little bit of extra margin needed to clear the market at that kind of size. The pricing and leverage available still really attractive from the borrower's perspective. We've seen those be pretty consistent throughout, you know, a year where there's been a couple of disruptions. But at the end of the day, you know, for the investors or the lenders, securing that AA type credit at a pretty substantial premium to the underlying corporate credit spread is still a pretty attractive proposition from their perspective.
- [10:21]
A
Yeah, I jump in there, Tim. I fully agree with what you've been saying. SG has been the forefront of some of the big AI data center financings in the space. And what we've seen through our distribution channels, whether through institutional investors or the bank markets, is exactly what you mentioned with respect to the AI lender pool really broadening out and folks opening up to, to the space. So, so not only are we seeing an increase in the number of transactions, but, but the sizes are increasing as well. I Think our first, first one was in the realm of, of 600 million and now we're seeing some, some of these transactions in the billion plus range and they are being absorbed by the market. I think lenders generally are proceeding with some level of caution with respect to overall exposure to individual names given. You know, as we're seeing the individual deals are quite large in the space, but I think to a point that Klaus brought up as well, as assets start to stabilize, we'll see a healthy recycling in the industry and folks continue to be able to support on that front.
- [11:30]
C
Great. So Mike, let's stick with you for this one. What are you seeing now in terms of the current mix of lenders for these projects? In addition to commercial banks, Are you seeing more pension funds, infra funds in sovereign wealth funds and how does that affect the way that these deals are structured?
- [11:49]
A
Yeah, yeah, absolutely. I think what we're seeing some of the exciting development in the industry since we last spoke. Our team here at SG has been at the forefront of this as well with respect to obtaining some ratings for construction financing in the data center space. And what that does is it opens up a lot of capital for the types of investors that you've just outlined. I think we're also seeing sort of a widening of appetite from the bank market as well. So some more and more players are coming to the space in terms of volume. It's necessary. Right, Because I think we've seen construction financing double over the past couple of years and we're in for another record here. So as additional projects get bigger and more come to market, we're going to need to broaden out space as well. But from our perspective at sg, we're seeing through our distribution pipelines, both institutional investors as well as bank market. And at times some prefer individual types of structures more than others. So sometimes the rating does help to bring in that type of capital and really round out the group. But yeah, I think we're seeing both depth and breadth of the market with respect to the lenders that are coming to these assets. And then to answer your question on the, on the structure, sorry, really quickly, is that I think the group has approached the market, has approached these financings through a lens of hybrid between real estate and project finance. So the lender mix doesn't really necessarily change the structure at times there's some minor intricacies that are relevant to the asset or market that the asset's located in and you sort of tweak and pull different levers in those situations. But for the Most part, we're seeing a lot of the construction debt come with an abundance of project finance discipline applied to the structure there.
- [13:46]
E
Yeah, the one interesting aspect of the ratings is typically you get the ratings, you know, once the project is stabilized or cod. And I think it's a new development, the rating agencies are willing to ascribe an IG rating prior to that. So if you look at a project, I think the agencies and banks in general have gotten more comfortable with construction timelines and construction risk so that you could bring in earlier alternative capital or permanent capital, given the earlier rating cycle and ratings ascribed to these projects. Thanks.
- [14:25]
D
Let's talk a little bit more about the structure. And you guys mentioned earlier that the markets getting more comfortable with these, like they're with infrastructure project financing. A lot of these deals started out with guarantees with completion or construction guarantees. Mellie, I'm going to turn to you. Can you explain kind of, you know, why we're seeing these guarantees and how the recent trends. I mean, are we seeing more or less of these guarantees?
- [14:51]
F
I think the guarantees sort of entered the environment as a bit of a legacy outlook to how financings have been considered for our space. You know, when data centers first came about, we were building smaller of them and many of them. And aggregation of a collection of assets was resulting in structures that were borrowed more from the real estate side of the house and certain banks. And also, you know, it is truly a real estate asset. Right. It's a property that you're building on land and you own the asset outright. You put mortgages you can secure, you know, put security, the hands of the banks. And so real estate really got things going in terms of financings. And so, you know, what they were really accustomed to was providing these construction completion guarantees. And when, you know, the projects got bigger and bigger, able to stand on their own two legs as, you know, single asset, single borrower, single tenant. You know, you're looking at these massive projects and the project finance market has, you know, entered and took over really. I mean, I think today 90% of the, you know, large construction projects are being done in the data science space. But project finance structures and terms, they sort of accepted this legacy term from the real estate side of things. You know, I don't love it as a developer. Of course, it provides some comfort to the banks. That's honestly, I think it's more of a historical thing. I don't know if it's necessarily. I mean, let's ask the banks on the panel. Curious to hear their take on it, but I feel like it's more of a legacy thing than anything else on the merits of the, the, you know, the, the completion guarantee thrown in on a typically non recourse project finance structure.
- [16:51]
G
Yeah, I, I, I agree with Mellie and obviously another developer here, this is Tim, but I'm a real estate guy myself so was, wasn't really surprised that these were a part of project finance kind of package. But you know, what we build is even the, the outer edge of like the turnkey know data center scope is still quite a bit more straightforward from an engineering and construction perspective than some of the other assets that get financed in the project finance world without completion guarantee. So I would also love to get the bank's perspective on this.
- [17:26]
A
Yeah, I could chime in here. I think one of the points with respect to the completion guarantees that we're seeing the space is an evolving asset class that I think came from obviously sort of starts in real estate structures whereby when you compare that to some of the core infrastructure project finance asset classes tended to have a little bit more of sort of turnkey, EPC type, fully wrapped construction contracts. We're seeing that more and more in the space. But I think there's two elements to why we still see the guarantees. One is sort of the early days of the industry didn't necessarily have that fully wrapped contractual structure that you see for core infra assets. And then the second being sort of the nascent nature of some of these hyperscale, or at least the volume of the hyperscale market, such that banks don't have a robust and long tenor of historical loss data. So these two elements I think feed into, feed into the need for banks to still push for those guarantees. I mean, I think we're in a space as well now where some of these financings are exceeding the enterprise value and market cap of some of the banks that are financing them, let alone some of the developers that are developing them as well. So I think we're seeing a trend in the sector to really focus in on, you know, the risks that are there, such as the overruns themselves and sort of the cost, potential cost delays. So I think the industry is getting constructive on that and not acquiring unnecessary risk exposure. And I think the industry will continue to move that way similar to what you see in renewables, although I'm not advocating for that anytime, anytime tomorrow, but just saying that that's probably what we've seen in other asset classes and renewable sectors and what we might see in the future for data centers as well.
- [19:33]
C
So drilling down on this a bit. Tim, we'll start with you for this one. Other than the completion guarantee, is the financing typically fully non recourse to the sponsor?
- [19:43]
G
Yes, it is. For regular way PF deals, we don't see any recourse to the parent unless there's some kind of development hurdle or at least contingency that hasn't been met as of financial closing. In those cases, you know, we've seen lenders ask for some credit support from the parent, but that would drop away once those hurdles were cleared. I think it's a little bit different in the case of warehouse or a Devco facility where there's typically going to be some credit support required from above in order or in exchange for the increased flexibility that those types of structures provide. But for the single asset, you know, PF deals fully not recourse but for the completion guarantee.
- [20:23]
D
Perfect. Let's stick with you for one more on the financing structure and what is included. Are you seeing the acquisition cost of land included in the financing and when are you including that?
- [20:36]
G
Yeah, well, by the time we get to the project finance market, we typically are going to own the land and probably have done some of the pre development or pre construction work, like mass grading, some of the horizontal infrastructure might be underway. As far as our ability to include that in our tenants budget, we have been able to do that for our yield on cost leases. And so those are our revenue generating and so we've been successful in getting banks to look at that as a part of the overall cost. When we look at land kind of as a separate financing from the project finance. So in the case of, and maybe you need to buy land, you know, nine months or 18 months in advance of clearing the hurdles that would unlock the PF or getting a lease in place for the pf, that's where, you know, I think we're starting to get some more. They're starting to become more interested in putting some kind of land acquisition facility in place. Land's gotten very expensive in a lot of the tier one markets. Once a location is proven to be viable for data center development, you know, the price rockets higher. We've made some Texas ranchers and farmers very, very wealthy people. I can't say what percentage of developers are using land financing facilities like that, but it's definitely something that I think more of us are considering. The main challenge there is if it's a really short tenure, the lenders, particularly if it's like a private credit lender, they have fees and a minimum multiple requirement that can be pretty hard to stomach if it is a short fuse. The effective cost of capital on that can be really high. So, you know, we'd be looking to structure it as a facility. We can recycle that capital rather than, you know, project by project basis.
- [22:31]
C
Great. So we, we touched on this a bit earlier. But Mellie, curious to know, are you seeing banks book these more as real estate transactions or as project finance transactions? It sounds like from what Mike was saying, these are now trending towards project finance. Curious to hear your thoughts on that.
- [22:52]
F
Yeah, I agree with that assessment. I think, you know, a lot of the activity is seeing the prevalent, you know, maybe like 90% of the liquidity coming from project finance method, but it really depends on the problem that the lenders are solving for for the developers. You know, for example, a new project with a sort of single asset and single tenant that's going to go through construction and has, you know, pre leased 15 or 20 years of contracted cash flows. I mean from, from an off taker that's you know, high credit to class's point earlier. But also some that are even not IG credited are getting these things. You know, typically follow that project finance method, you know, sculpting to the SCR etc. Over the cash flows of the project or the totality of lease. But on the other hand, you know, there are also borrowing base or you know, DEFCO loans that are care more about the contracted run rate cash flows in comparison to the the debt service that's required or the outstanding loan size as well as the loan to value metric. You know, which is typically beneficial for more corporate style financings where a pool of assets are baked in and baked in equity can be helpful to borrow more flexible capital capital for either incremental growth and portfolio or even in some instances, you know, some strategic capital for land and power development opportunities. Similar to what Tim was talking about. You know, yes, you can do bespoke land financings or you can do these Devco financings where a certain portion of the proceeds could be used towards those strategic land and power development opportunities inside of a pool with some, you know, leveraging of cash flows coming from other assets that are lateralized inside the same pool. So it really depends on the problem that's being solved for. And yeah, in these pre leased projects with tenants are known and you're trying to put in a massive financing for a massive single data center. Yes, project financing has been the prevalent method.
- [25:09]
D
You talked a little bit about how we're financing different pieces of this Mike, I'm going to turn to you. Have you seen as part of these early stage development site structures for these data center financings, are there financings for, for example, for the credit support under some of the contracts or are you looking at Power, Gas, data center? Are those all separate financings?
- [25:31]
A
Yeah, exactly. Happy to touch on that. I think it's probably one of the more exciting aspects of the industry right now and where we're going to see a lot of focus in the coming year and the end of this year as well. Just given the fact that power is such a constraint and in the space and what developers and asset owners are trying to do is solve for the quickest delivery of capacity to their, to their tenants, to their clients. And we see these problems being tackled a number of different ways. I think Nellie alluded to one of the more common ways, one of the more common tools that we've brought to market in the form of devcos. This is a popular structure that allows borrowers to unlock liquidity for early stage assets by leveraging a whole pool of assets. So some may be stabilized, some in construction and some with the need of earlier stage development capital for, for the likes of equipment purchases or deposits with your utilities. And these typically structured to include LC's and revolvers for these purposes specifically. And we've been pretty active in the space. And one thing I think that it's worth highlighting as well, outside of your Devco structure, SG has a full team here working capital and cash management that'll help with clients just supporting the standalone lcs as well. And so while there is a market for land financing and that can be done, albeit at, you know, fairly expensive rates for earlier stage assets for certain clients, we're able to structure these solutions with our dedicated team to provide LC's supporting those deposits on an earlier stage basis. And for what lenders are looking for with respect to that power is you're depending on the utility region that you're in, you're looking for power agreements, LOAs and really a path to sign leases with tenants if there's not one there already. And a big thing I think for lenders in the space it tends to be focus on markets that are either tier one or have a path to strong growth and underlying demand components. So with all that being said, I think the short answer to your question is yes, we're seeing the various elements of the data center power components being financed and there's a couple of different ways that that the market's tackling it.
- [28:08]
C
So what do you Guys see then as the average ticket size for data center financing and what are you seeing as the tenor? Tim, let's start with you.
- [28:20]
G
Well, I think, I mean partly the ticket size is a function of the, the overall deal size, right? There's, there's some tiering involved and some of the larger deals. And by that I'm not just talking about like the you know, $20 billion mega deals, but you know, a billion plus is going to require kind of multiple tiers and different players at each tier. So we've typically asked our, our lead banks to underwrite about a third of our deals which be anywhere from 200 to, you know, $500 million commitments where our deal sizes like everyone's are going up. But that's kind of our first, our first handful after that, you know, they commit to that size but they, they want to sell down, right? None of, none of these banks, they as capital markets transactions and so they're not looking to hold, you know, that, that kind of size. And so there's a sell down that occurs next tier below those leadership banks or would be the joint leader Rangers Jlas. We've asked for like 100 to $150 million commitments there. And then if needed there's you know, up to two, maybe three additional tiers where the check size as well as the economics kind of are reduced at each step down on that ladder. But really we look at the bottom of the pyramid here. There's a lot of people out there who want to write checks in the, call it 15 to $75 million range. That's a fairly big range. But there's a lot of different types of participants in that part of the capital structure. This is really where we've been doing a lot of work to try to broaden a segment because all those investors are buy and hold, right? They're not further syndicating that out or further kind of the final source of the liquidity that kind of flows up through that capital structure. The end of the day someone's actually got to want to hold this paper. I'm specifically talking about the project finance market, but we have done a lot of work to build that base of liquidity that benefits everyone through the mega banks and including the borrowers themselves. I don't think I answered your question about the tenor. We've been using a four plus one term structure. Four years gives us enough time to kind of accommodate any, you know, realistic project delays. And that extension is typically contingent on, you know, project completion, funding, reserve accounts.
- [30:39]
D
And so on the Banks might cost that consistent with what you guys are seeing on video?
- [30:45]
E
Yeah, four plus one is definitely, four plus one plus one is definitely tenor for the project financings. But as Mike kind of alluded to earlier, I mean there's a whole ecosystem of different levels of financings that are, that are happening around data centers. It's not just project financings. It's financing the whole kind of life cycle of a project or developer. So we kind of look at it again from a life cycle with project finance and the capital markets being the end of it. But earlier stage we're fairly active in the supply financing as working capital. So we have supplier finance programs so we can offer that to developers who are procuring equipment. So that's typically a very attractive product because they need to procure a lot of equipment and it definitely supports their balance sheet. So and those are short term, Those could be 180 days to one year and then they roll over. Then you have the development facility which are typically two years. And you know, that's again early stage kind of ntp and those are a little bit higher risk. You're taking development risk. But those typically go into project financing or warehouse. So the warehouse is a combination of development and pre stabilized assets and maybe some stabilized assets that will be taken to the capital markets at some point in the future. And we've seen tenors there in kind of the three year range. And so there's different pockets of capital for different parts of the life cycle of a data center. So that's kind of the short answer from my perspective.
- [32:28]
D
Perfect. You know, not all data centers are creative equal. Right. So how are the banks looking at financing of a hyperscale data center compared to say a located data center?
- [32:41]
F
Mike?
- [32:43]
A
Yeah, So I think what it really comes down to for those, obviously there's, there's a number of differences with respect to where the assets are located, the types of tenants and even quite frankly, maybe most drastically the size. Colocation data centers can add a lot of value to the tenants and even the broader network through carrier hotels and the like, given their proximity to user bases. And what we see in the hyperscaler side, especially with AI, is sort of the broadening of the geographical spectrum with respect to where these assets can be located. That said, when looking at financings, for the most part, sort of the strong points of colocation include the location itself. Right. Having a more proximity to the end users as well as an element of customer diversification. Oftentimes there's an anchor tenant as well. So These COLO facilities, whether they're serving enterprises for cloud purposes, can have an IG sort of backbone supporting those cash flows, whereby the hyperscalers, you know, you're looking at the single tenant, so you're either analyzing the cash flows with respect to that one single tenant over the term and then looking at, you know, the market overall and considering where the asset's located and the ability to potentially replace a tenant if needed. So those are sort of your various credit considerations on the revenue side for both. What it comes down to, I think to get to the point on the differences in financing is that COLO will have a little bit of a price premium and slightly lower leverage and as a result taps into a potentially separate pool of capital that is a bit advantageous to the space. Just because, as we were speaking about before earlier, rounding up the 50 and $75 million tickets is a bit easier than, than the larger ones sometimes given the volume of deal flow in the market right now. So there's an advantage there. And then for the hyperscalers, there's obviously an advantage given sort of the robust tenant and sponsor relationships across the banking universe. And so I think it shakes out to a bit of a pricing premium and difference in leverage and attachment points. When you look at Pecola, I think.
- [35:13]
G
It goes back to something that a couple of us have mentioned, including at the top of this call, like are you financing an infrastructure asset or a niche real estate asset? And I think in the case of the hyperscale customer, what makes it feel a little bit more like core infra is you've got a 15, 20 year lease with a very high investment grade rated, well known, profitable, you know, company as the, to put it in infrastructure terms, the off taker. Right. And so those cash flows are pretty close to bulletproof once the development's been completed and the tenants taking over possession. And I think that has unlocked some pretty attractive financing terms for us both in the project finance market as well as in the kind of more, you know, I guess what you call permanent debt, not really permanent, but longer term, you know, fixed rate kind of debt. I think if I'm, if I were to put myself in the shoes of like an insurance company where there's a tremendous amount of capital that's kind of been on the sidelines waiting for maybe it's credit rated term loans or securities, when I look at making a 15 or 20 year kind of bet. I think you feel a lot better about the hyperscale segment because you know that while valuations might go up and down. Like your cash flow is pretty much, pretty much baked and known. And it looks a lot more like a, like a long term bond or a credit tenant lease than it does a real estate loan.
- [36:50]
D
That's great. Klaus, we're going to come to you for a minute. So you know, everyone's, we've talked about, there's a lot of interest in the data centers. We've talked about these great leases, the tenants, a lot of activity in the market. So you can refi these out. Why then at least, you know, this was a question we had at the time we did it earlier. If these are such attractive deals, why are they pricing wider than renewable deals? Is it the demand or is it something else?
- [37:16]
E
It's a good question. I don't know the exact answer to it. I've some theories of it. I mean, I think it's just the amount of capital that needs to be raised. I mean, look, I mean you have a big renewables deal is very large renewable deal is 2 to 3 billion. And that could be divided amongst 8, 10 banks. Right. It's not out of the realm of possibility to see a $20 billion deal every couple of months. So that's a lot of capital. So I think to attract kind of the universe of banks needed to get that deal done. And we're not talking about eight banks, we're talking about maybe 50 to 100 banks. Some banks will hold 250, maybe three. Some banks will hold 25 to 50. You need that every last dollar to get the deal done. And again, that's not just for one deal, that's for a couple of deals that are coming to market. So I think that's the main reason, you know, conversely, you know, renewables, from my perspective, I mean, we do both. I think it's a, it's a different asset class, it's been around longer. But there's risks in renewables that you simply don't have in a data center. I mean you have the tax equity, you have, you know, back leverage, you have different tax equity structures. It's a much more complex asset class. And you know, if you talk about solar, you have basis here for curtailment. You have all these risks flying around that you need to kind of figure out so why it's priced tighter than data centers. I'm not sure, but I think it has something to do with the amount of capital that's needed going forward. Be curious what other folks think.
- [39:04]
A
Yeah, I think that's fair. I think you typically would expect the potential for Compression in pricing as an asset class matures. And I think somebody mentioned earlier in the discussion, lenders are viewing this on a relative value basis, clipping a decent amount of spread in excess of the pricing of say the bonds of Microsoft or in AWS for that tenanted data center. So there is some benefit there. I think there is a sort of stabilization in the space for the exact reason that Kyle outlined. And, and it's driven by the demand for capital right, right now. And so while we're seeing the industry mature, we're seeing volumes really outpace sort of any expectations that, that folks may have had. So it tends to be a really capital intensive and high demand industry that is maintaining a stabilized level of pricing in the space.
- [40:04]
E
The other thing that just came to mind is, you know, renewables and energy prices are going up. I think everybody agrees on that. But PPA prices historically have been, have been fairly tight and margins for developers on the renewable side are super thin. It's hard to make a really good return. So I think on the data center side, I think there's more room to play with respect to cash flows and returns. So I think, you know, as a hybrid scaler, I think they're willing to pay up to get the data center completed and on time and on budget. So I think, you know, I think there's more room. So probably you could pay a little bit more.
- [40:49]
C
Great. So that's actually a great segue. We're going to move now to Lisa's PPA from development. Mellie, we'll start with you. What are you seeing in terms of who's responsible for supplying the equipment? Is it the developer or the tenant? And how far out are either developers or tenants planning for longer lead time on equipment?
- [41:14]
F
So I'll start with the last one. You know, the longer lead time, I mean, can't exactly speak for what all the developers are planning, whether they're planning appropriately or not. As you know, the equipment planning is a very, you know, depends on risk appetite and capital availability as well as the availability of said equipment within those time frames as much as anything else. So that said, I think the, I'll call it advisable planning cycle is for a two year development timeline as that is when the hyperscalers start, you know, getting interested in pre leasing capacity, at least in my personal experience, you know, usually if you have, you know, longer than two years to delivery of your data center, which, you know, long lead equipment can very much be the, or the equipment can very much be the long pole intent, you know, it's kind of rare to see a tenant be interested in pre leasing that site. You know, maybe unless there's a reasonable ramp up over time in the same campus or some other externality going on, that's helpful. But you know, usually that two year time life cycle or construction timeline is what you want to aim for when you're planning for your supply chaining of the long lead equipment. Sorry, did I answer that question? I think there was a second question at the beginning of the two questions you asked me.
- [42:53]
D
Who's, who's providing the equipment?
- [42:56]
F
Oh, well, developers are investing in the equipment, right? That's kind of what we do for a living. We have to get in, get in line for whenever that equipment's going to be available and sign up to selling, you know, certain. Sometimes there's deposits, sometimes there's not, depending on the equipment, depending on the availability equipment at that particular moment in time and the competition for that equipment. But you know, developers are the ones who are signing up to, you know, finding that equipment. That said, you know, there are, I mean, it also depends on the structure of the lease. You know, we at DC Blocks have been mostly doing, you know, turnkey data center developments with turnkey leases where, you know, we invest through not only the corn shell, not only the powered shell infrastructure, but also the mechanical and electrical plant. Therefore, that plant requires mechanical electrical equipment. So, you know, we have to plan for that. Some developers are finding great success and just building the powered shell, which is, you know, of course a smaller component of the build out. But you know, you can plan perhaps a little bit more in advance with tenants as they deem more economics with respect to those powershell opportunities. But of course, you know, smaller deployed capex and therefore smaller deployed, you know, aum that asset under management that you're getting returns from just, you know, great, great business models and both of them just two different appetites. And you know, I'm not saying we would never do powered shells, by the way. Don't get me wrong. I mean, that's something we explore every day. But it's just a matter of what you're looking to invest into, whether you. That determines whether you as a developer go ahead and order some long equipment or not. And in the case of power shell, your tenant that's coming in would typically be the one bringing in the equipment.
- [44:50]
D
Thanks. Can you. I want to stick with you, Mellie. Can you just really quickly tell us about the different types of leases and you know, how the payment structure by the tenants?
- [45:00]
F
Absolutely. There are a few Variants. This actually, you know, the prior question was a good segue for this. You know, I think first, the amount of capital deployed and the demarcation of whether, you know, the developer is building and owning a PowerShell or a turnkey data center with mechanical electrical plant or mep we call it matters. Second, usually connected to the, to that first comment, whether the tenant or the developer is responsible for operating the site matters. Typically a powered shell is fitted out by the tenant itself and the developer wouldn't operate the tenant's equipment. So you know, it would follow a triple net structure where the tenant pays for the base rent plus is responsible for the cost to maintain and operate the site. Property taxes and insurance, those are the three ends in the triple net. And of course with most of these, you know, hyperscale developments, I mean any of them really tenant is paying for the electricity bill. If however the MEP is owned by the developer, there are a few more flavors ranging between, you know, it could still be the triple net where either tenant might operate the site or hires the developer to separate operate that same site or even tenant might hire a third party to operate the site. But in any case, that same triple net structure applies where the tenant would be responsible for operating not just the, you know, you know, not their own equipment, but the equipment that's owned by the, the developer. There could be a modified growth structure where the developer is also the operator for the lease and you know, prices in the OPEX and taxes and insurance while the tenant still pays for electricity usage. Or we've also seen some instances where, you know, it's a bit of a hybrid or in between where you know, you could do for example a double net structure where the, those two ends are, you know, where the developer operates the site but the tenant, the tenant covers the taxes and the property taxes and insurance of course, as well as the electricity usage like all the other structures.
- [47:10]
C
Klaus, let's move to you. Are you seeing debt increasingly gain size on releasing assumptions for data centers?
- [47:20]
E
We have seen for the smaller deals, the $1 to $2 billion deals, I think we give credit and banks have given credit to releasing assumptions. So if you look at a 50 year lease, we'll give credit to that five year assumption of releasing and there's another releasing assumption beyond that, but we give it to the first one and that's similar to kind of, you know, again, you know, analogous to renewables land where you give credit to Russian cash flows. But I think in some of the larger transactions that are taking place, again just to circle the amount of capital and the amount of banks that are, that are needed and that maybe do not have the transactional history and the data center space. Those are sized in my opinion sometimes within the contractual, the original contractual term, which is we mostly 15 years.
- [48:20]
D
I'm going to follow up on that one a little bit. So you've talked about how they are willing to advance a bit against the post contract. How is the sculpting happening? How is the DSCR sculpting rate applied to the cash flows that are contracted versus uncontracted?
- [48:34]
G
Tim, we actually haven't seen any difference. We have financed three projects where we have a 15 year initial lease term followed by three five year options and we've succeeded in getting lenders to underwrite that first option to a total year end profile of 20 years. I think there's a couple of things that were important and in achieving that. These projects were in the Northern Virginia Cloud availability zone, not Virginia itself. And the tenant's a household name with a double A credit rating. So I'm not sure you can get similar treatment without those types of mitigants. But for those projects we were successful. Argument that the likelihood of renewal was very, very high.
- [49:20]
F
Yeah, I would agree with that. I think when you're, when you're financing the, the market has been important as the, in the feedback that we've received from lenders. You know, if you're in a tier one market versus tier two and the credit rating of the tenant also matters. That said, you know, good credit rating tenant, you know, in the investment grade or high investment grade range, I think they're still able to get that, you know, first initial lease term of let's say 15 years plus the first renewal term of five years included in the debt sizing experience. Even if that lease is in you know, tier two or even three of these instances market. However, you know, if you go into the sort of the non investment grade zone, maybe you're getting that additional five years in your sizing for know a top market. But almost certainly it'll be more sensitive as you go into, you know, the, the non tier one markets. If you have that non investment grade sort of tenant and you might only be getting credit for your GSCR sculpting for that first, you know, 10, 15, whatever years of lease you have with that tenant, that's and of course you know, that's not the only risk element. So there are other certain components that can be structured on the lease or with the tenant or with the borrower that eliminates Some of these risk elements that could potentially reasonably have lenders providing and have resulted in lenders providing credit for that renewal period, the first renewal period, let's say, of that extra five years in the DSCR sculpting. I think it's a, it's always a holistic view and you know, what are the risk elements, what are the mitigation strategies? Why is this such a strong investment thesis for all the lenders to come into when considering whether or not to give credit to that extra renewal period?
- [51:28]
C
All right, well, we're coming to the end of our time here. So as a final question, we'd like to do a round robin and ask each of you the big question. Is this a bubble? Who would like to start? Mike, I'm going to put you on the spot.
- [51:43]
A
I could jump in there. Happy to kick that off. Look, I think some of the underlying trends that we've talked throughout our discussion have highlighted sort of the structures of these loans, underlying collateral, which is a high demand asset, and you know, in popular markets, tier one markets and ultimately sort of. I think the key theme on these are the credit worthiness and the strong revenue profiles of the underlying tenants. When, when we look at sort of the use cases of data centers and I know that there's been a lot of buzz and concern around AI, I think the latest with respect to the overall market is that AI is less than 15% of total capacity. And, and while that's expected to go up at a CAGR above 5% a year, the regular traditional use case of data centers is expected to also go up at like a 3 and a half or 3% CAGR through 2030. So the everyday use of compute is such an essential piece of our lives both at work and at home. And I think AI is just starting to scratch the surface of that. So while we see a strong growth and the market is sort of seeing rapid multiplication of debt quantums in the space, it's really supported by strong underlying demand and for the most part creditworthy or if not creditworthy, near creditworthy and really strong revenue profile businesses. So in my opinion, no. Very happy to hear the rest of the group's takes on that as well.
- [53:29]
C
Anyone else?
- [53:30]
G
Okay, I think the answer is is, is no. But there might be some bubblicious pockets. I think when we look at the importance of cloud to pretty much everything we do throughout the day, whether it's, it's working on shared documents, on our SharePoint, it's, it's, you know, watching Netflix, it's Yeah, anyone who has kids with their Snapchat, there was actually an AWS outage, you know, a couple days back that took those services offline. And I think it was pretty surprising for people to realize how much real people ran on these cloud services that aren't labeled as such, but really run on that, on that critical infrastructure. And I think that underscores the need for just more reliability, more robustness in those cloud networks. And that's what our customers continue to say is the, the cloud services business is, is really booming and the, the backlog of unfulfilled revenue is, continues to grow and is in the hundreds of billions. So we are really focused on a lot of those cloud services companies and helping them meet that demand. You know, when we look at some of the AI stuff, I think there's, there's maybe a bit more of an argument that, you know, you know, you see some, some pretty frothy, you know, valuations. At the same time, I think the, the potential for AI to, to change the world and practically unlimited, you know, addressable market is, is, is very real. And you know, maybe some of the companies haven't quite yet figured out how to monetize it, but I think there, there's a day in the, the not too distant future where a lot of these AI applications are integrated to everything we do throughout our day. So I think there's a huge amount of economic value that's kind of sitting there. I'm sure along the way to realizing that economic value there will be some ups and downs, but I don't think we're in a bubble right now.
- [55:35]
F
I guess I'll chime in and turn over the class for the final word. But as much as I not a huge fan of hot takes, I guess I'll try to put a little bit of a, maybe a safer spin on this. I think I understand why people would think that, oh, there's potentially a bubble. You know, there's a growing concern over the, you know, success story of how much growth there is in AI. I'll say two things. One is, you know, those who got burned from, let's say the Internet bubble and, or the dot com bubble and, you know, some other types of bubbles. This feels a little bit different to me just because of how integrated to everyday life of both consumers and companies it is and how much of a value proposition there is behind it. But also, you know, those who are, the companies that are spearheading this AI advancement are not just the, you know, basically thinking back to the previous bubbles they're not the sort of companies that only do this. I mean, these are companies that have invested in vast infrastructure and, you know, advanced, really the largest businesses on the planet, even before AI came about as a, you know, know, leader of investment dollars around the country and around the world. So I think from that perspective, I feel, you know, a little bit safer. But also, you know, as developer, you know, two developers and two banks who are putting their balance sheet to work. We are putting our money where our mouth is, I guess is the second thing I would say. You know, we, we believe in this growth. We believe that there are sound fundamentals associated with where we're investing all this capital. And so, you know, I, to me, I don't, I don't, I'm going to agree with the other folks who have already spoken. I don't see it as a bubble. And, you know, our actions are backing that up as well.
- [57:49]
E
I agree with everything that the panelists have mentioned. Yeah, this, this touches upon every, every aspect of everybody's lives on a daily basis. Right. So I think AI is not going anywhere. And the only thing I could see is like, if there's a new technology that comes along that substitutes a data center or servers and the chips, et cetera, or super cloud, I don't know. But I just don't see a substitute right now for the foreseeable future. So, again, yeah, I don't see this as a bubble at all.
- [58:29]
C
Great. Well, that's all the time we have. Thank you so much to our panelists for the great conversation insights. We really appreciate it.
- [58:40]
B
You can find us online at www.projectfinance.law or send us an email at currentsordonrosefulbright.com Please rate, review and subscribe on Apple Podcasts, Spotify or your preferred preferred Podcast app. Our show today was produced by Emily Rogers.
- [58:56]
F
Stay ahead of the Currents.