The Data Center Gold Rush: Why the Concrete Will Outlast the Hype
On the Fashion of Concrete
There is a certain kind of American enthusiasm that arrives every decade or so dressed in new clothes, and it is worth a moment's sober attention before we all agree, once again, to be dazzled by it.
A generation ago it wore the costume of real estate. Men in good suits filled hotel ballrooms to explain, with the fervor of revivalists, that the true fortune was not in the property itself but in getting there first, in believing loudly enough and early enough that the belief became its own justification. The buildings were almost incidental. What was being sold was momentum.
Today the ballroom has moved, and the subject is data centers. The vocabulary is different, but the tune is the one our grandfathers would have recognized instantly. "Supercycle." "Once in a generation." Vast sums are being committed, not on the strength of demonstrated returns, but on the conviction that whoever pours the most concrete first shall inherit the earth.
It is instructive, as it always is, to notice who is doing the talking and who is doing the paying. The men urging haste are rarely the ones who will be left holding the deed when the enthusiasm cools. This has always been so. It was so with railroads, so with radio, so with the fiber optic cable laid across half a continent in 1999 for traffic that would not arrive for a decade. The financiers, the developers, the analysts who issue their breathless notes each quarter, these are the men who profit from the telling of the story, whether or not the story turns out to be true. That the very suppliers building this infrastructure are privately arranging to protect themselves against its collapse tells a careful observer a great deal more than any keynote address.
But there is a peculiarity to this particular fashion that deserves its own paragraph, and it is this: the machines inside these buildings are growing obsolete at a pace that would have astonished even the most excitable speculator of an earlier age. A structure raised this year to house the finest computing equipment available will, within a few short seasons, be housing equipment that a serious engineer would consider quaint. The scale being built is a monument to a moment, and the moment does not hold still long enough to justify the monument.
The land will remain. The power lines will remain. The concrete, indifferent to fashion, will remain standing long after the particular machines it was built to shelter have gone the way of the vacuum tube. What will not remain is the premise, so confidently asserted today, that such enormous scale was ever truly required. A fraction of the space, filled with better machines, will do the job that whole campuses are being built for now. This is not prophecy. It is simply what has happened every other time men have built for the technology of the moment as though it were the technology of forever.
None of this is cause for alarm, only for memory. Every such enthusiasm overreaches itself, and the overreach is always defended, right up until the moment it is abandoned, by men who are certain this time is different. The land survives. The story does not.
In five years the circuit will have moved on to some new word, quantum or robotics or whatever comes next, and the salesmen will be just as certain about that as they are certain about this. And a good many of these vast, single-purpose monuments to a passing moment will be standing half empty, waiting for someone with a different sort of imagination to decide what else so much power and so much concrete might be good for.
The wise were never fooled. They simply declined to say so out loud until it was safe.
