Monthly Archives: October 2013

Is your landlord an asset manager or a data center company?

I just read an article about how Sears has become a zombie retailer. I am guessing the reference to zombie is designed to tie into a Halloween theme while also compare it to an actual walking dead person that eats brains. I haven’t been in a Sears in at least 5 years – for any reason. The one thing in the article posted here that caught my attention was a comment about the CEO:

“He’s not a retail executive, he’s an asset manager,” said Mary Ross Gilbert, managing director at Imperial Capital. “What we’re seeing reflects what an asset manager would do, which is find ways to realize value.” 

Then it hit me – there are a lot of data center companies who are asset managers vs. data center operators. The differences can be subtle if you don’t know much about the industry or they can be pretty obvious if you do. For instance, an asset manager will find a building – any building like a strip mall, office building downtown, or brewery and think its a great deal so he buys it. Then he spends 2-3x the amount of money to improve the asset to turn it into a data center because data centers are a hot asset class right now. The asset manager is now in the data center business.

To maximize revenue things like over subscription to the power system is implemented, buying used generators from farmers for expansion is acceptable, and running cables and piping (which is really expensive) into nooks and crannies of buildings to try to monetize stranded power that came about as part of a bad design happens, as well as other things that if you don’t know, you find out quickly. Our personal experience was this – one firm who easily had the highest reputation produced a design we paid $25,000 US for that got us a $46M design for an 8MW Tier III facility. It was a rock solid design, used generally accepted principles across the board, and was $21M more than the better design we selected. Why? Piping, conduit, and cables are extremely expensive and if you don’t employ a scalable solution that allows you to utilize your power footprint in chunks, you will overspend. By a LOT.

What was the difference? We are data center guys. We are not asset managers, we know what questions to ask, which buildings make good investments and how to deploy a data center for less than $4M/mw until you run out of power. We like using our brains, not eating them.

The Financial Impact of a Data Center Upgrade

The Upgrade Problem.

We do a fair amount of consulting with investment firms. They usually contact us to correlate company data with market data and validate trends and get a pulse on what risks may be out there that a 10K won’t identify. Investors don’t like risk, especially the risk that they don’t know about, but can live with and hedge against risk they do know about. We are not registered investment advisors, nor do we own positions in any company we write about so keep that in mind. We do know a little about data centers and what they cost, and our clients enjoy our knowledge base.

The financial risks that are looming on the horizon are real, they are expensive, and ultimately don’t solve a problem, but may prevent one. The elephant in the room is aging infrastructure. UPS equipment, cabling, batteries, generators, conduits, ceiling tiles, steel, the very components that make a data center what it is.

The financial risk is seen at different layers:

  1. Service levels – if you perform unscheduled UPS maintenance, you can lose your entire facility and every customer will feel it. Ask Alchemy Communications. Great company that performed an unscheduled UPS maintenance work order and dropped the facility. If the outage lasts any length of time, then the penalties will kick in, meaning lost revenue. Depending how many tenants you have this can mean a lot of money. You may get sued and that costs time and money. You will lose customers – whether in a knee jerk reaction or because they don’t want it to happen again. If you’re lucky they have a short memory of bad things.
  2. Upgrade costs – the third largest chunk of expense for a data center after power and taxes is equipment. Many data center buildings were built 25 years ago or more. One Wilshire was completed in 1966. Since then it has morphed from a run of the mill office building into the 5th most interconnected building on the planet thanks to the internet. Converting an office building to a data center is common practice, however up against the useful life of some key equipment creates a huge capex liability when it’s time to freshen up at several million dollars per megawatt.
  3. Competition – Blunt Hammer built a business plan for a group of investors that put the development cost of a new facility at less than $4M/mw. Much less. This allows a competitor to enter the market at half the cost of today’s facilities’ rent, and build a new facility for less than a capex bill for a renewal on an old building. Leaves REITS holding the bag on old, expensive real estate that is maybe 2% of the cost of a project anyway.

So we talk about these issues a lot. We see Equinix sell their facilities to a REIT and in our opinion a smart move. Why? Because they aren’t saddled with the upgrade burden of what we would estimate as a $2.5M – $3M per MW in the not so distant future. That’s a lot of capex to absorb on its own, but when combined with the operational risk – it’s like changing the tire on a moving car because customers’ uptime is involved.

Smart companies are hedging against this by locking in debt. Most of the press releases we have seen say that the debt is for expansion, but we believe a significant portion of that debt is to cover upgrades – especially those that coincide with customer renewals so that they don’t incur the expense of upgrades without the revenue from leases.

What do we tell our clients about these risks, besides engage us to mitigate them? We tell them to ask questions, look at ages of things, and talk to other customers. And if they need help, reach out –