How data center projects became a mainstay of the commercial construction sector.
I had to chuckle when an earlier “instructions to bid” (ITB) notice came to me attached to a non-disclosure agreement (NDA). I always josh with my “exactimator” colleagues that NDA means “never divulge all,” a reference to a typical architect’s penchant for coming up short on the details. That, and the irony of an NDA being shared with half the bidding world, i.e., all parties who will have to solicit material pricing, requests for information (RFIs), etc. The contradiction in terms is always good for a laugh.
But levity notwithstanding, an NDA notice along with an ITB had lately indicated an opportunity to price another of several potential data center projects, a consideration not to be laughed at. I had recently bid a data center job that had proven to be quite lucrative, and I was primed to garner a repeat performance.
The recent development of data centers was, at the time, rapidly becoming a meat-and-potatoes benefit to a post-pandemic recovery and a boon to the commercial construction economy. But who would have thought that within a few short years’ duration that an incidental uptick in activity would rocket into a full-blown, multi-megaton construction boom? Well, the answer is: we should have.
As some-time number crunchers and part-time prognosticators, we guesstimators should have seen a tsunami of tech developments coming on like a runaway semi screaming down a mountain grade. From the moment that paper plans gave way to on-the-screen takeoffs, throughout the struggles to archive an avalanche of captured data (remember all the program crashes, email purges, and overheated servers?), the revelations made themselves clear. We were approaching the eve of a data center revolution, and we were relatively clueless about the forthcoming acceleration.
Fast forward to the present, and the growth scenario, if anything, has been previously understated. Data usage has increased at least 100 times over the past five years. To put this in better perspective, it is useful to say that more data has been created in the past three years than in all of history. Clearly, with this explosion in data creation, there is a demand to store, process and deliver it. Data centers provide the physical infrastructure to meet this demand.
Construction activity regarding data centers has been nothing short of phenomenal. There are currently 3,561 data centers operating in the United States. Over the past five years, that number marked a factored increase of 17 times. Of course, actual construction activity must be assessed with the size and scope of the projects involved. A recent Microsoft design typically encompassed a footprint approximately the size of three football fields. This roughly translates into a couple hundred thousand square feet of drywall partition, plus consistent orders of magnitude involving beads, grid (ceilings and access floors), paint, plaster, and insulation per project. Data center construction has been (and will continue to be) a boon to the wall and ceiling segment of the construction industry.
What’s more, the outlook for the near future promises to outpace the present, perhaps exponentially. The evidence is clear that data center construction has overperformed to date. But if we can say that data center activity is in high gear, we would have to conclude that the advent of artificial intelligence as the next wave of tech development is anticipated to shift the current scenario into hyperdrive.
That anticipation has translated into billions in investment earmarked for the development of data infrastructure. Tech powerhouses like Microsoft, NVDIA and QTS have pooled together with investment giants like BlackRock to underwrite the expected avalanche of new data center construction related to artificial intelligence (AI). Yet the current capacity is falling woefully short of anticipated growth.
Today, 78 data center facilities are under construction. But global tech leaders say that the present levels of activity are vastly insufficient. The statistics bear this out. In 2024, data center construction increased 69% over 2023. Now consider that the nascent AI engine has not yet even begun to spring to life. Clearly, the United States desperately needs to build more data centers.
But as with any dynamic enterprise, obstacles will arise, and the projected expansion of data centers is no exception to that rule. To begin with, the space requirement to build a typical data center (remember the three football fields example?) has become a daunting snag. Aggressive deregulation and an easing of permitting requirements can help to relieve the constraints on the acreage-eating monster that AI portends.
But even more daunting than the space obstacle is a looming demand for electrification. Data generation requires huge amounts of power, and AI exponentially so. The United States has enjoyed a relatively flat power demand for the past 20 years. But the sudden surge for power-hungry data centers will most certainly double our power grid’s capacity over the next 12 to 13 years. Increased electrification of such a magnitude will require much innovation and a shift in emphasis on energy consumption (think, natural gas and nuclear).
But such potential roadblocks are not insurmountable, and national security demands that they be resolved so that the United States can continue to be the foremost leader in technological advancement. So, my advice to commercial construction contractors is to reserve a healthy spot in projected volume for data center construction—and for estimators to brush up on their technological terms so we don’t end up flat-footed again, like on the first tech wave.
