S&P Global Offerings
Featured Topics
Featured Products
Events
S&P Global Offerings
Featured Topics
Featured Products
Events
S&P Global Offerings
Featured Topics
Featured Products
Events
S&P Global Offerings
Featured Topics
Featured Products
Events
Corporations
Financial Institutions
Banking & Capital Markets
Economy & Finance
Energy Transition & Sustainability
Technology & Innovation
Podcasts & Newsletters
Corporations
Financial Institutions
Banking & Capital Markets
Economy & Finance
Energy Transition & Sustainability
Technology & Innovation
Podcasts & Newsletters
Blog — 8 Jul, 2022
By Dan Thompson
Last week, Jim Chanos, President of Chanos & Company, announced he was working to raise hundreds of millions of dollars for a fund that would bet against (short) datacenter REITs listed in the US. Chanos, who is known for shorting stocks over the years, originally found fame when he predicted the fall of Enron more than 20 years ago. Equally as famous as his Enron prediction, though, is his bet against electric car manufacturer Tesla, which didn’t go so well. So when it comes to this latest bet, against datacenter providers, will it be Enron or Tesla for Chanos and his investors?
In his public statements to the press, Chanos’ main thesis for betting against datacenter providers is that, as he puts it, the companies are now losing out to the public cloud providers. Despite the fact that those same cloud providers are currently large customers of the very datacenter providers Chanos plans to bet against – which he acknowledges, by the way – he views datacenter providers and public cloud providers as “enemies,” concluding that “The real problem for datacenter REITs is technical obsolescence.”
This charge is, of course, nothing new – people have been predicting the death of the datacenter since the public cloud providers (Amazon AWS, Microsoft Azure, Google Cloud, etc.) began their rise to prominence more than a decade ago. Traditionally the argument goes one of two ways: First, enterprises will abandon their own IT infrastructure in favor of moving everything to the public cloud, leaving the poor datacenter providers with nothing left to host. As we’ve seen over the last 10 years, however, this never materialized. Companies now leverage hybrid infrastructure, with assets in the public cloud and hosted in both company-owned datacenters and third-party colocation facilities – including those owned by datacenter REITs. Second, datacenter providers would become too heavily reliant on cloud providers as customers, and one day those cloud providers might bail on them as customers, for whatever reason, leaving the datacenter providers with all of this built-out infrastructure but no one to rent it to (see the first argument).
Chanos’ argument seems most closely aligned with the latter scenario, and in fact, he goes so far as to say that the reason the cloud providers will leave is because they prefer building out their own datacenters rather than renting from the likes of the datacenter REITs. When the cloud providers do rent space, he supposes, they’ll beat up the REITs over price so bad that they’ll be left with nothing but the scraps of their eroding margins.
Before we go into the supposed merits of this second argument, when we say "datacenter REITs," what companies are we even talking about? Admittedly, this landscape has changed dramatically over the past 12 months. At the beginning of 2021, there were seven prominent, publicly traded datacenter REITs in the US: CoreSite, CyrusOne, Digital Realty, Equinix, IronMountain, Switch and QTS. In August 2021, though, Blackstone Funds got the ball rolling with the $10B acquisition of QTS, taking the company private. American Tower – another REIT, but traditionally focused on owning cell towers – followed suit in November 2021, with the $10.1B acquisition of CoreSite. CyrusOne was next to go, when it was acquired by KKR and GIP for $15B in March 2022. And then finally in May 2022, Switch was acquired by DigitalBridge Group – another REIT, but with a portfolio that spans datacenters, cell towers, fiber cables and more – for $11B, leaving only Digital Realty, Equinix and Iron Mountain standing.
So which REITs will Chanos target exactly? He didn’t say, but it seems unlikely he would pick on American Tower, since only 6.9% of its revenue comes from datacenters. Similarly, Iron Mountain only sees 7.8% of its revenue from its datacenter business. Prior to the Switch acquisition, DigitalBridge saw 21.8% of its revenue from datacenters, and the Switch acquisition looks to push that number well north of 50% – but Switch has never been known to heavily service public cloud providers, Chanos’ main critique of the industry. This leaves Digital Realty and Equinix, the two largest datacenter providers in the world, both in terms of operational square feet and revenue.
What, then, is the real risk to these datacenter REITs of the public cloud providers turning on them, leaving them to ruin? To understand this, we must first understand the way the public cloud providers grow their infrastructure. Public cloud providers do, in fact, like to build their own datacenters, as Chanos asserts, but this is nothing new – they’ve done so for years. At the same time, though, the various cloud providers also rent space from datacenter providers. Why would they do this? There are a number of reasons. Sometimes the cloud providers rent space to gain entry into new markets quickly, especially overseas markets. Sometimes it is easier to rent space due to lack of local real estate opportunities (i.e., no cheap real estate), like in Hong Kong, or even downtown Los Angeles. Sometimes they rent space for speed-to-market reasons, like in Phoenix, where each of the public cloud providers have bought land in the greater Phoenix metro, and then signed large contracts with datacenter providers soon thereafter because they already had capacity available to rent. Speaking of those contracts, when the cloud providers rent entire datacenter buildings from the various providers, they’re doing so under multiyear contracts, which can be anywhere from 5-15 years. Furthermore, it is rare these days, although not completely unheard of, for datacenter providers to build datacenters speculatively – meaning they already have an anchor tenant for the facility, or at the very least, very warm leads on anchor tenants, before any dirt gets moved.
The net of all this is that between 2015 and the end of 2022, 65% of all new rented datacenter capacity globally is wholesale – the segment of the datacenter industry that services the public cloud providers. So yes, the various cloud providers do like to build their own datacenters, but at the same time they also rent datacenter space, and a lot of it.
It is worth noting, however, that the cloud providers are not the only customers of these datacenter providers. In fact, the public cloud providers actually attract other customers to these facilities, because companies want to be as physically close to cloud providers as possible in order to improve the experience of the workloads that move back and forth to the public cloud (that hybrid model discussed earlier). Also of note, Equinix only started servicing public cloud providers through whole-building-style rental contracts just last year. In 2021 Equinix launched a new datacenter product, which it calls xScale, specifically targeting hyperscalers, a group of large tech companies including Amazon, Microsoft and Google, which are known to grow their digital footprints at a rapid rate. Interestingly, though, Equinix has so far financed these projects through joint ventures with private equity funds, which spreads the risk around.
None of this is to say that churn can’t or doesn’t happen within the datacenter industry, even in the early stages of development. Switch somewhat famously lost an anchor tenant in Atlanta, delaying its first building project significantly, but the company went on to rent the space to different customers. Equinix also famously lost a sizeable customer a number of years ago from one of its LA facilities, and while it's true that space took a while to rent again, the company did finally rent it all, and is now expanding the facility further to accommodate more demand. And so it goes with churn.
Finally, what about those falling rental rates (prices), that Chanos is concerned about? This, too, is nothing new. At 451 Research we’ve been discussing the falling rental rates in our market reports for years – and that’s just it; these rates are really a market-by-market discussion. It’s a simple supply-and-demand scenario. When providers have space sitting that they want rented, or there is a customer shopping the market that they believe will bring additional business with them, they will get aggressive with pricing. When a provider maintains a dominant position in a market, and no other space is readily available, they can command a price premium.
Interestingly, the global supply chain issues seem to be helping datacenter rental prices (in the favor of the datacenter providers), since they are effectively slowing datacenter development, thereby creating supply constraints. This, too, seemed to be a factor in all the rental activity that took place in Phoenix earlier this year. So yes, falling rental rates can be concerning, and it does eat into providers' margins, but a quick look at the profits of the two datacenter providers in question suggests that it has not been an issue thus far; certainly not the big issue Chanos suggests. Additionally, there doesn’t appear to be any looming change on the horizon for pricing, and if anything, pricing pressure will get a break due to these ongoing supply chain issues.
Can Chanos be wrong, though, yet he and his investors still profit handsomely as he has in “feasting on the returns of these stock ideas for years," as he has said? Well, that’s the funny thing about the stock market, where in today’s world a simple tweet can make a stock soar or sink. This, too, we’ve seen before. In mid-2018, self-described “short activist” Blue Orca Capital released a scathing report about Chinese datacenter provider GDS while holding a short position on the company’s stock. The report claimed the provider was misleading investors on the amount of space it had actually rented out, and also outlined some shady financial dealings. To those of us in the datacenter industry, it was clear that Blue Orca didn’t understand the business model of datacenters providers, but to investors it didn’t matter, and GDS’ stock fell 46%. GDS explained away the claims, and investors slowly returned, but Blue Orca profited in the meantime.
The fact of the matter is this: the datacenter industry has proven itself to play a critical role in economies around the globe. Whether hosting enterprise infrastructure directly or hosting it by way of the various cloud providers' infrastructure, datacenters underpin everything in the way business is done today. Of course there is ebb and flow within the datacenter industry, just as in any other industry, but we see no major changes on the horizon in the buying habits of the companies that rent datacenter space.
Simply put, there’s a reason private equity firms keep buying up datacenters and datacenter providers, and that’s because they offer steady and consistent growth. Looking ahead, 451 Research estimates the compound annual growth rate of datacenter capacity globally to be 9% from 2021-2027, a majority of which will be built to continue servicing public cloud providers. This underscores the fact that the public cloud is a complement to the datacenter industry, rather than an enemy of it.
Campaign
Podcast