first part

GlobeSt.com: What do you consider the single biggest difference between data centers and other types of property?

Kerrigan: The amount of power they consume. Which is also what distinguishes one data center from another. Power is so important, in fact, that some developers are pricing per kilowatt hour rather than per square foot. It's a whole different model. What's the real estate worth, the amount of space or amount of power? Many of our [tenant] clients argue it's the amount of power you can get to the space. You might have a lot of space, but if no more power is available, the space is useless. For existing data center tenants, the real question becomes, as these leases expire, are you better served having 100,000 square feet with two or three megawatts, or are you better finding a facility built post 2005 that might be only 30,000 square feet and have the same amount of power.

GlobeSt.com: Could you elaborate on that?

Kerrigan: Google is a good example of a tenant, where they may be consuming all the power in a building, even though they're using only part of the space. You could have 70% of the building vacant, but you can't lease any more space because you don't have any additional power to deliver to other tenants. It's like a retail shopping center. It's good to have an anchor tenant, but if they're not paying enough rent but using up all the parking spaces, the anchor becomes counter-productive. So it's not really how much space they take, but how much power they use.

GlobeSt.com: What impact does that have on development?

Kerrigan: First, it means vacancy isn't a reliable measure of demand. You could have a lot of space empty that can't realistically be rented out. So you can't look at vacancy rates and automatically know if the market is healthy or not. In places like Silicon Valley, where there's a lot of really large power users, you could have a lot of empty space and still be short. But in another market where there's mostly smaller users, you could have a lower vacancy rate and still have too much space. The number of large users is growing, and to the extent they're still renting, they're using more and more of the available power. And things are changing so fast, it's hard to keep up. Facebook just announced they're taking 90,000 square feet from Digital Realty Trust (at Turn-Key Datacenter in Santa Clara, CA). They didn't even exist three or four years ago. Google uses so much power that they're rolling out their own centers. They used to be a data center tenant, but now they're an owner-occupant.

GlobeSt.com: Does it require special expertise to develop data centers?

Kerrigan: Generally, data center developers operate as well as build the centers. They develop everything for you. Then you just move in. You just bring your own servers. They'll have everything from tanks and fuel to power density units. But the level can vary a lot. I compare data centers to car dealers. Not everyone needs to have the Maserati. A Lexus will typically suffice, and some will say that's an overstretch. There are four large developers. There's Dupont Fabros (Technology Inc.), which generally does ground-up development. There's Digital Realty Trust, whose model is to take an older building and retrofit it. They used to be just a real estate company, but now they do spec suites, kind of like executive suite offices. There's also Quality Group of Cos. and 365 Main (Inc.), which do a little bit of both. There's lots of other guys out there that are going out and trying to do this. The number of data center developers is in its infancy, but the guys who tried to enter the market in the past 18 months can't get in because of the financing situation. That's why we think supply on a national basis will be hard to find 12 months from now.

GlobeSt.com: What about cost?

Kerrigan: Data centers are very expensive. For an office building, you may have $80 a square foot for tenant improvements, but you may have to spend $1,000 a square foot for a data center. Then there's the operating cost, which is why location is so important. The price of electricity is a big factor. You're always going to have to have local data centers, where utility costs can't be changed, but for large companies like Google the cost of power is going to be a major factor in where they locate their data centers. What's driven most of the expansion on a national basis is the cost of power. But a lot is also driven by client demand: I need a dot on the map. It needs to be Chicago or it needs to be Milwaukee. What it comes down to is finding the cheapest location for power in the Chicago or Milwaukee area.

GlobeSt.com: How big is the market?

Kerrigan: It's hard to come up with the exact size of the market. Everyone keeps their information pretty close to the vest. The challenge is, you could have a building that's 100% leased by seven tenants, but each tenant could be marketing their space for sublet to other companies. You have tenants who lease space to use as their own data centers, and you have tenants who provide data center services for other companies. Co-location providers may lease 350,000 square feet with the intention of subleasing most of it to smaller companies who can't afford their own data center. You'll never be able to tell with any certainty how much of leased space is actually being uses and how much is available for subtenants. It's hard to keep track of where they're at capacity wise.

Want to continue reading?
Become a Free ALM Digital Reader.

Once you are an ALM Digital Member, you’ll receive:

  • Breaking commercial real estate news and analysis, on-site and via our newsletters and custom alerts
  • Educational webcasts, white papers, and ebooks from industry thought leaders
  • Critical coverage of the property casualty insurance and financial advisory markets on our other ALM sites, PropertyCasualty360 and ThinkAdvisor
NOT FOR REPRINT

© 2024 ALM Global, LLC, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to [email protected]. For more information visit Asset & Logo Licensing.