First Half 2017 Data Center Update
Digital Realty and Equinix deals out value for all of 2017 data center acquisitions combined.
Q2, 2017 Acquisitions
When analysts stated that the mergers & acquisitions (M&A) market would continue to grow in 2017 (after a robust 2016) they weren’t lying.
The Digital Realty Trust acquisition of Du Pont Fabros is yet the largest data center acquisition valued to date. What I find amazing is that the total acquisition only encompassed 12 data centers. While I know that the transaction is based upon revenue and current customer contract values, the construction value is far less from a fixed asset standpoint. While the total transaction was $7.6B, the deal has an equity value of approximately $4.95B based upon DuPont Fabro’s 77.8M shares outstanding as of April 2, according to Thomson Reuters. Digital Realty Trust operated 157 data centers internationally and the acquisition only adds 12 data centers to the portfolio. While I am just a dumb engineer, I wonder what the monetary value is purchasing your number one competitor. Guess we’ll leave that up to the professionals.
The other deal that happened in Q2, 2017 was the Equinix purchase of Verizon’s 29 data centers in 15 markets. The acquisition included the NAP of the Americas in Miami, the most connected data center to South America. Over the last four to five years, Equinix has purchased data centers in Brazil and other regions and has found that the investments have been profitable … a good method of testing the Latin America market. The combined data center locations are over 179 locations in 22 countries. One of the advantages of Equinix is their IBX platform (network) as well as catering to vertical markets. Out of the 29 sites Equinix acquired, 11 were a part of Terramark Worldwide (previously owned by Manny Medina, the new partner of Cyxtera).
These two data center acquisitions aren’t the only purchases that are in play. Silver Lake recently sold its interest in Vantage Data Centers to a small group of private investors. This combined with Cyrus One, QTS, Carter Validus, and others even further the total value of acquisitions in 2017.
THE RACE FOR HYPERSCALE
Last issue I wrote about the five hurdles collocation/wholesale providers need to overcome when considering designing for hyperscale tenants. DuPont Fabros historically has large footprints in a wholesale environment. With this said, the acquisition by Digital Realty further positions them to support additional hyperscale tenants. The focus is on cloud. Recent expansions at DuPont Fabros allow for a footprint that can easily accommodate the hyperscale tenant. Equinix’s model however is different in that its primary model is to support collocation racks. The quarterly quota for Equinix per site is said to be around 175 KVA, a far different model than wholesale.
NOW COMES THE DUMB ENGINEER
Between 8% and 18% of today’s data center portfolios contain infrastructure that was built circa 2000-2008. During that era, many of the data centers were designed at 50 to 75 Watts/sq ft, which limits their bus size when considering expansion. It’s difficult to update an existing bus in a live data center, therefore complete new systems need to be installed. This creates two problems. The first is that many of the data centers that are from that era do not have enough space for complete new mechanical, electrical, and plumbing (MEP) infrastructures to be installed. The second issue becomes the heightened risk of upgrading infrastructure in a live data center. Also when upgrading live, due to the method of procedures (MOP) and standard operating procedures (SOP), the cost of labor and coordination sky rocket. As you know, today’s requirements are between 150 to 200 W/sq ft.
Above and beyond upgrading the bus system, a static UPS life cycle is between 12 to 15 years (well maintained). Replacement too is subject to risk. Chiller system life cycle is also just a few years beyond the UPS, typically requiring replacement at 20 years.
CAPEX REPLACEMENT AND THE SELLER
During every acquisition the seller typically submits a capital expenditure (CAPEX) upgrade program identifying equipment upgrades over a 10-year period. The issue becomes that it is the objective of the seller to identify a low cost of CAPEX, thus making the sale more profitable. All too often we find that the seller of the CAPEX program lacks detail, and the final CAPEX is much larger than originally anticipated or identified by the offering memorandum. Any additional CAPEX identified during acquisition due diligence is typically deducted from the sale offer.
A true pro forma identifies the revenue flow, absorption rates, cost of capital, and 10-year CAPEX. It’s only until you get the full picture that the true business model/acquisition can occur.
SOLUTION: TECHNOLOGY MIGRATION/CONSOLIDATION OVER CAPEX UPGRADES
One solution other than upgrading old colocation assets is technology migration/consolidation of multiple data centers within a single market. In many cases the first data center in that market will have the oldest infrastructure and the newer data centers that were built has access capacity. Fading out old assets need to be a part of the overall program.
One of the latest migration strategies that we’ve developed include rack swing space (rented for the migration) in which you can transfer processing to the swing racks, while you install and populate the new racks during migration. This approach is much better than a “push-pull” scenario in which you are physically decommissioning and relating existing racks. A colocation provider can strategically identify when leases are up, and get out front of the tenant with a migration strategy, other than losing an existing lease/customer. It is far less costly to migrate technology than to upgrade major infrastructure.
Next issue we will discuss the three major components (disciplines) required in a successful migration/consolidation for colocation providers.