Data Center Demand for Building Materials

The mechanism

A data center is, before anything else, a building. It needs insulation for thermal management in an environment generating massive heat loads, waterproofing that matters more per square foot than in a typical office building because water and electronics do not mix, a roof, and lumber and other structural materials for the construction itself.

Brad Jacobs, discussing QXO's acquisition of Top Build, put it plainly: "Data centers need roofs too. Data centers need waterproofing very much so. Data centers often need lumber-related products. Data centers are big consumers of building products."¹ At the time of the deal, data center revenue was a single-digit percentage of Top Build's total revenue, but described as very fast growing, with the expectation that demand for these building products grows in proportion to the data center buildout driven by AI training and inference infrastructure.

Why it matters strategically

The demand this creates is durable and not disruptable by AI itself: insulation, roofing, and waterproofing will not be replaced by AI models. "It's not going to be disrupted by AI or LLMs," in Jacobs's words. "It's just not going to go away." The buildout that AI drives is exactly what creates demand for these materials rather than substituting for them. This complicates the common framing of an old economy set against a new one: as Tracy Alloway and Joe Weisenthal put it at the top of the episode, "you couldn't have the new economy without cooling and heating... the actual business of making things and moving things, heating and cooling, is very much linked with the new economy."¹ The physical substrate of the new economy is traditional construction. For QXO, which holds the number one position in installation, the number one position in waterproofing, and the number two position in roofing, the data center buildout functions as an embedded upside scenario layered on top of the core residential and commercial business, in a segment, commercial and industrial construction, that is already the fastest-growing part of the category.

Open questions

How durable data center construction spending proves to be depends on whether AI infrastructure spending decelerates or consolidates geographically, either of which would reduce this tailwind. The single-digit-percentage, fast-growing disclosure is an operator's estimate rather than a segment separately broken out in filings. And because large-scale data centers cluster in specific markets, a distributor's actual exposure to this demand depends heavily on its geographic footprint.