What Data Centers Are Doing to the Housing Market Next Door

Long before anyone had heard of a data center, there was already a well documented pattern in American land use. Landfills, power plants, and prisons rarely land in the neighborhoods with the most capacity to fight them. I think a new report from Realtor.com just showed data centers falling into that exact same historical pattern, quietly, without anyone necessarily setting out to make it happen that way.

What the Report Actually Found

In 2015, only 12 US ZIP codes contained a large data center, defined as 50 megawatts or more.By June of this year, that number had grown to 108, on pace to reach 125 by year end, roughly a tenfold increase in just over a decade. The share of US home sales happening within five miles of one of these facilities has more than doubled since 2018, from 0.67 percent to roughly 1.5 percent today, and is projected to approach 2.3 percent by 2027 based on what is already under construction.

Here is the finding I did not expect, and I think it is the most important one in the entire report. Realtor.com modeled what would have happened to that proximity share if the industry had simply stopped building new facilities in 2018. Under that scenario, the number would sit at roughly 0.6 percent today, essentially flat. The entire increase traces back to brand new facilities landing in communities that never had one before, not to homebuyers choosing to move closer to existing ones. This is not a story about people voting with their feet. It is a story about where the industry itself is choosing to go.

Why the Property Value Story Is Not the Real Story

I expected this report to show housing stock turning over faster near large data centers, some sign of residents trying to leave once a facility landed near them. It does not show that. Housing turnover in ZIP codes near large data centers tracked almost identically to broader metro areas throughout the entire period studied, with the gap never exceeding two tenths of a percentage point. Whatever this buildout is doing to nearby communities, it is not visibly driving people out of their homes at a measurably different rate.

What the report shows instead is a shift in who is on the receiving end of new facilities in the first place. The average large data center that opened in 2018 drew about 24 megawatts of power. The average one opening this year draws about 60. The median 2026 opening sits in an area with roughly 70 percent fewer residential housing units per square mile than the median facility built in 2017, and the typical facility expected to open in 2027 sits about 34 miles from the nearest major city. Newer facilities are also increasingly landing in ZIP codes with below median household income, a real reversal from the pattern that used to define this industry, when data centers clustered near wealthier suburbs with existing fiber and power infrastructure.

Why This Is the Question Most Site Selection Models Never Ask

Every site selection model I have seen treats a community’s economic and political capacity as background noise, something that shows up later in the approvals process rather than something worth pricing into the site decision itself. This report suggests the industry may already be doing exactly that pricing, just implicitly, by consistently choosing sites in communities with less density, less wealth, and by extension, historically, less organized capacity to resist. That is not necessarily a conspiracy. It is very likely just the accumulated result of thousands of individual site decisions, each one reasonably prioritizing land cost and power availability, that happen to land in the same kind of place again and again.

I do not think that makes the pattern any less worth naming. A community’s ability to negotiate real terms, real benefits, real accountability, is not evenly distributed, and an industry that keeps landing in the places with the least of that capacity should expect the political reckoning to eventually catch up with the pattern, the way it has in Virginia, in Texas, and now increasingly everywhere else.

What This Means for Anyone Building Community Engagement or Site Strategy

Price a community’s actual capacity to negotiate into a site’s real risk profile, not just its power availability and its zoning. A site in a lower income, lower density area may look easier to move through quickly, and it may genuinely be easier in the short term. That same lack of organized capacity is exactly the condition that tends to produce the sharpest political reaction once a community does organize, often after the deal terms are already locked in and much harder to renegotiate.

This is the kind of cross disciplinary read, real estate economics meeting infrastructure siting, that belongs in a direct conversation rather than a general takeaway. If you are building community engagement strategy or site selection criteria and want to talk through what this pattern actually means for your specific markets, reply and tell me what you’re working on.