The facility sits forty minutes outside a town whose name I’ve agreed not to print, on land that was, until 2011, alfalfa fields. It cost $340 million to build. It operated for six years. It has now been empty for four, which means the building has already spent more of its existence disused than in service — a ratio that will only get worse with time, since nobody is paying to tear it down, and the desert is in no hurry to reclaim it.
A Year With Ruiz
I spent a year getting access, and then eleven months actually walking its halls with a decommissioning contractor named Ruiz, who had catalogued nearly nine hundred data centres over a twenty-year career and told me, on our first afternoon, that this was the biggest one he’d ever watched die. The scale is the first thing that undoes you: two hundred and forty thousand square feet of raised flooring, built to hold racks that hummed at a constant 68 degrees, now silent except for wind finding gaps in the loading bay doors.
Engineered for a Permanence It Never Achieved
What struck me, walking those halls, was how much of the building had been engineered for a permanence it never achieved. The concrete was rated for a hundred years. The cooling systems, redundant three times over, were designed against failure modes that assumed decades of continuous operation. Nobody who poured that concrete was thinking about the building’s obsolescence, and yet obsolescence arrived years before the warranty on the generators expired.
Nobody who poured that concrete was thinking about the building’s obsolescence.
Ruiz has a theory he’s reluctant to say out loud to clients, which is that we have gotten extremely good at building for permanence and extremely bad at building for the actual, much shorter lifespan our technology grants things. A cathedral and a server farm can require similar tolerances of concrete and steel; only one of them is designed with any expectation of outliving the people who use it.
The Same Vertigo, Provisionally
I don’t think the essay this became is really about data centres. I think it’s about the particular vertigo of standing inside something built with total confidence in its own future, watching that confidence turn out to have been miscalculated by a matter of degree rather than kind — the same speed at which we’re all, provisionally, building everything else.
- Graham, S. (2016). Vertical: The City from Satellites to Bunkers.
- Easterling, K. (2014). Extrastatecraft: The Power of Infrastructure Space.
