If you are noticing a slowing of absorption in the data center industry, you are not alone. The exuberance of the last several years has led to massive buildouts, bringing us to a point today where supply (including land banks) is in front of demand in several key markets.
This level of competition is likely to continue to drive down price and returns. This is most prevalent in the U.S., especially in markets like Northern Virginia, Dallas, and Phoenix. In these markets, pricing has come down significantly over the past few years, and while many players have been able to reduce their cost to build by reducing redundancy (i.e., N+1) and/or improving their supply chain, construction costs appear to have bottomed and returns likely will continue to come down.
HOW DID WE GET HERE?
How does supply come to outweigh demand? It’s clear that the third-party data center model that originated in the U.S. is no longer a secret, with providers focusing on nearly all the regions one would logically assume makes sense for data centers.
The high level of interest in the sector reflects the strong demand that’s been seen from hyperscale customers over the last few years and their need and desire to expand their physical presence.
While entrenched U.S.-based companies like CyrusOne, Digital Realty, and Equinix are expanding and competing in many of these regions, so too are new names that in many cases are well funded and have strong management teams.
For example, in South America, ODATA (owned by Patria/Blackstone), or in Asia, Airtrunk (Goldman Sachs/TPG). In addition, established players are expanding into new markets such as Colt (Fidelity) in India and Vantage (Digital Bridge) in Canada (4Degrees). There are also companies pioneering new markets like Teraco in South Africa and IXcellerate in Russia.
This, by the way, says nothing of the significant competition already in the U.S. from companies including Aligned Energy, CloudHQ, EdgeConneX, RagingWire, or QTS, who each have already won hyperscale deals, or from companies like Cologix, EdgeCore, Flexential, Iron Mountain, or STACK, who are looking to enter the hyperscale market.
WHAT IS THE FINANCIAL IMPACT?
To this point, Digital Realty recently guided to 9-12% average stabilized yields on new developments vs. 10-12% in prior years. We believe returns in the U.S. will ultimately stabilize around ~8% and would be similar to those seen in more traditional real-estate segments.
While this would suggest that the worst is behind us and that the curve should start to flatten, we still have yet to see the full impact on the competitive landscape from companies who lack scale and either can’t build cheap enough or get their cost of capital low enough.
In international markets, while returns are typically higher than the U.S., outside of Europe, considering macro factors, they too are coming down. The direction of returns appears to be strongly and inversely correlated to the level of competition – which as we stated is only increasing around the globe.
WHAT DOES THIS MEAN FOR INVESTORS?
The reduction in returns though is not simply a result of growing competition, but also reflects evolving investor recognition of the sector as infrastructure and real estate vs. previously telecom and technology.
After all, it is these investors who are financing many of the companies – both public and private – and are content and comfortable with these relatively lower returns.
This by itself is not a problem and arguably reflects the maturation and validation of the sector, as well as the recognition that these hyperscale customers are typically investment grade and sign longer-term leases than we’ve seen historically.
However, there appears to be far more private funds that are flush with cash and required to deploy said capital than there are attractive platforms to acquire. This has led to some companies being funded that probably shouldn’t be at the valuation levels that they are, when taking into consideration things like quality of tenants, term of leases, land or building ownership, service offerings, as well as things like country, currency, and political risk.