Cool Vector Video-Podcast TRANSCRIPT: "Equinix vs. Digital Realty: A Fitch Analyst Talks Data-Center REITs"

Cool Vector Video-Podcast: Charting the rise of data centers and the digital infrastructure asset class.

Episode Title: "Equinix vs. Digital Realty: A Fitch Analyst Talks Data-Center REITs"

Episode Duration: 13:36

Originally Posted: December 5, 2024

Listen to the Full Episode here

Speakers: Harold Chen, Director of Commercial Real Estate at ⁠Fitch Ratings⁠. MODERATOR: David Snow, Cool Vector

Episode Overview:

Data center real estate investment trusts (REITs), like Equinix and Digital Realty Trust, differ from traditional REITs in their higher operational intensity and reliance on artificial-intelligence tailwinds for growth, says Harold Chen, Director of Commercial Real Estate at ⁠Fitch Ratings⁠.

Chen, whose team has assigned investment-grade ratings to Equinix and Digital Realty, says the companies benefit from surging demand related to AI. Of the two, Digital Realty has a higher concentration of hyperscaler customers, defined as big-tech, data-center customers like Microsoft, Mega and Amazon. The upside of hyperscaler concentration is longer-term leases, while the downside is customer concentration and the inability to more frequently reprice rent rates, says Chen, adding data center REITs have a history of only single-digit customer churn. Data centers that cater to co-location customers tend to have shorter leases, he says.

Data-center REITs also differ from traditional REITs in their "significantly higher levels of operational intensity," says Chen. Complexities like power, cooling and interconnectivity make data-center REITs "significantly different beasts."

Chen also discusses with Cool Vector the impact that ESG and sustainability initiatives have on Fitch ratings, and the historic challenge for data center companies to access certain forms of financing, like asset backed securities (ABS).

---

TRANSCRIPT

David Snow: Hello and welcome to Cool Vector. I'm David Snow, your host. And today we're joined by Harold Chen of Fitch Ratings. Harold, welcome to Cool Vector.

Thanks for being here.

Harold Chen: Thanks David. Thanks for having me.

David Snow: So you are a ratings analyst. Your area is REITs, and there's been an interesting trend in REITs in recent years where data centers have become an important asset that is now put into REIT structures. We expect there's going to be a lot more data center REITs in the future because the population of data centers is going to expand rapidly around the world. There’s probably going to be a lot more REITs to house these assets because it's a way for investors to participate in the growth of digital infrastructure.

I'm fascinated to hear your approach to determining the creditworthiness of these organizations and some of the things that make the asset class unique. So Harold, why don't we start off with a bit of a thousand-foot view from your perspective, given that you understand and track both traditional real estate REITs and now this newer crop of data center REITs like Equinix and Digital Realty Trust, both of which Fitch has issued ratings for recently.

From a creditworthiness perspective, what makes data centers similar to real estate? And what makes them different from your perspective?

Harold Chen: We rate Equinix and Digital Realty, and those are the major data center REITs that we look at, but we still look at them within the REIT framework. We're rating them out of the REIT team. When we look at them, a lot of things are very similar in terms of our approach and some of the characteristics of those REITs.

At a very fundamental level, they are operating in the business of renting space and receiving rent. So they are leasing space out to tenants and then receiving recurring revenues.

What sets them apart is when you compare them to more traditional real estate types—like residential or office—rather than leasing a building to one or multiple tenants, they're leasing rack space. What’s involved in that is significantly higher levels of operational intensity.

There's a lot that goes into the management of a data center, like power, cooling, things that aren’t present in some other types of real estate. On the other hand, that also comes with certain benefits.

For example, they may be receiving fees for their management. In addition to the operational intensity, there’s also capital intensity, just in terms of the specifications of the data centers themselves, the kind of construction, and the need to ensure technology remains up to date and not obsolete.

David Snow: You mentioned operational intensity. Are there simply more levers that need to be pulled and things that need to be carefully monitored within data centers? Where, if there are errors or underperformance, it affects the whole thing?

Harold Chen: Yeah, absolutely.

An office building has some amount of regular maintenance that needs to be done. There may be REITs that have warehouses where tenants cover a meaningful amount of the maintenance. And maybe the REIT will say, “The roof is having an issue. Let’s replace the roof.” So the amount of operational intensity there is relatively low. There’s not a lot of moving parts.

Data center REITs, in comparison, are a significantly different beast. You have to worry about a lot more—power delivery, cooling, interconnections. There’s just so much more that needs to be managed compared to some of these more traditional asset types.

David Snow: In a commercial office building, you typically would have a number of different companies taking up different floors. In a data center, isn't it often the case that there’s just one or not that many counterparties? Maybe they’re large, hyperscaler-type organizations like Amazon or something that might speak for the entire lease. How do you think about that? Is that customer concentration a risk? Or is it actually a good thing because the whole thing’s full and the leases go out for decades?

Harold Chen: Yeah, it's quite interesting. Especially as we're seeing this large AI wave and increase in demand, a lot of that is coming from these large hyperscalers—Amazon, Microsoft, Alphabet.

It can be a bit of a double-edged sword. There are obviously pros and cons. When we think about the pros, these hyperscalers may lease a large mega campus and hold on to it for a relatively long period of time. They tend to have longer lease durations.

Longer leases generally mean less risk. At the same time, there may be less opportunity for the data center to reprice those as the lease expires or to increase pricing over time. So you’re kind of trading pricing power for stability.

Another thing—you mentioned tenant concentration.

So as we look at Equinix versus Digital Realty, Equinix tends to have less concentration in hyperscalers. Digital Realty has more. That in itself is not necessarily a differentiating factor on a rating level, but Digital Realty does have higher tenant concentration as a result of being more hyperscaler-focused.

We are seeing both of these companies benefiting from this AI demand that is primarily coming from hyperscalers. They are doing development for large hyperscale projects using JV structures. That’s something we’ve been keeping an eye on. For them, it’s a way to reduce risk. These are large investments. If they can spread the risk around with multiple equity partners and keep it non-recourse, off-balance sheet, it makes sense.

If there were to be a pullback in AI demand, those projects may be harder hit.

David Snow: Can you just talk about the basics of how you determine creditworthiness based on the financial characteristics of the data center company?

Harold Chen: The overall way we look at a company and assign a rating—whether that’s BBB+, BBB, something higher, something lower—we’re fundamentally looking at the company’s ability to repay its debt.

When we look at that, EBITDA, cash flow. For REITs specifically, we look at REIT leverage—we define it as essentially net leverage, not including preferred stock. And then we also look at REIT fixed charge coverage.

Whether a company has demonstrated access to certain types of capital—unsecured bonds, equity, secured debt, ABS—those are all things we evaluate when we assess risk. What is the company’s operating history? What does demand look like?

Some of the strengths of the data center REITs—Equinix and Digital Realty are both investment grade. Most public REITs are investment grade. They both have very low leverage. Historically, data centers have had less access to certain types of secured financing, such as ABS, but that has been improving.

David Snow: Fitch recently issued to Digital Realty Trust a triple B rating. I saw in the notes of the rating that Digital Realty Trust’s global network has given it a competitive advantage and perhaps support in your ratings. Talk about why that is an advantage from Fitch's point of view.

Harold Chen: Both Digital Realty and Equinix have diversified global footprints. Generally, we see that as being able to spread risk out over different geographic areas. Having that diversified, wide global footprint is a plus because companies are not as impacted by potential issues in any individual area.

David Snow: It's recently rated a bond from Equinix. It's a green bond. We know that in the world of data centers and digital infrastructure, there's a huge demand for sustainable developments and sustainable energy. What can you tell us about green bonds and the extent to which you're seeing these in data centers in particular and in real estate more broadly?

Harold Chen: When we look at ESG, it's primarily through the lens of how ESG factors could impact the rating. We're not saying, “This is good for the environment, therefore it should be rated higher,” or, “This is bad for the environment, therefore it should be rated lower.” But we are looking at ways in which ESG factors could, through public sentiment or other channels, affect the rating.

For example, when we look at certain healthcare REITs, they may have an ESG score of four for some social factors because healthcare pricing is an issue in the United States. For data center REITs, we do not have any ESG scores of four for environmental factors.

As we look at green bonds, I would say that we don’t view them inherently any differently than non-green bonds. They may be useful if adhering to some of the green bond stipulations allows them to get a slightly lower interest rate or something like that.

When we look at power and the potential for power as a gating factor, we primarily look at transmission rather than production—delivering power rather than generating power to data centers. The other thing I would say is that a more sustainable building, especially for a data center because they are operationally intensive, can mean lower costs for a company. If you're able to save money by having more efficient buildings, that's a plus.

David Snow: Can you talk a bit about what you look at in the leases and the long-term leases with the tenants of data centers? How do they differ from regular real estate assets? What do you like to see?

Harold Chen: Data center leases tend to be relatively short, especially with more retail colocation tenants, as opposed to hyperscalers. Hyperscalers tend to have longer lease terms. Retail colocation leases can be maybe only two or three years. Other asset types can have long-term leases of 20 to 30 years.

Shorter-term leases mean more potential for tenants to drop out or for reduction in revenue as those leases roll. That said, that hasn’t really been an issue for any of the data center REITs. Both have very low single-digit churn. Once a tenant is working with a data center, that tends to be a pretty sticky relationship. It can be costly for them to move to a different data center, especially if they’re integrated into the infrastructure of the existing partner.

Hypothetically, the short lease term means there is potential for customer loss. It's not something we've seen to a large degree, but it's still a potential risk.

Harold Chen: As credit analysts, we're always focused on the downside. We look at the potential for an AI pullback. A lot of companies are experimenting with AI technology. It has a lot of potential in terms of where it could go, but not every company has been successful in monetizing it. So if there is a future demand pullback related to AI, that could affect some of the hyperscale developments.

The AI issue is really the 800-pound gorilla in the room. The increase in demand has been very positive for them in terms of operations, pricing, and growth. At the same time, as we’re looking at downside risks, we are considering the potential for a pullback in AI demand. That said, both companies have shown themselves to be pretty disciplined, with relatively low leverage.

David Snow: Why do data center REITs have lower levels of leverage than other types of REITs?

Harold Chen: At least the companies we look at—Equinix and Digital Realty—have historically been relatively conservative. At each rating range, we tend to give positive and negative leverage sensitivities that could lead to positive or negative ratings action. The bands we've set historically were tighter than for some other asset types.

Data centers are very specialized. They can’t easily be converted into other types of assets. So if something goes wrong, are they going to be able to get contingent liquidity? Over time, we’ve softened the leverage sensitivities. We think that these companies can now tolerate the same or even more leverage than other types at the same rating.

David Snow: Excellent. Well, Harold Chen of Fitch Ratings, thank you for speaking with Cool Vector today. I hope we can ask you to come back in the future when your expertise can help us understand the digital infrastructure world better.

Harold Chen: Awesome. Thank you so much, David.

END