Trusence Every claim has a source
Last updated 21 November 2026 Search Türkçe
Infrastructure

Data centers are about twice as likely to sit in majority-Black neighborhoods, study finds

Smaller co-location sites are the kind most often placed in communities of color, while the largest facilities tend to go to white rural areas.

A new study from researchers at George Washington University and American University's CSINT center finds that data centers are sited in majority-Black, Hispanic and Asian neighborhoods at markedly higher rates than in white ones. Across 81,000 US census tracts and 5,291 data centers, majority-Black neighborhoods had 2.2 times the odds of hosting a facility compared with white neighborhoods in the same state, rising to 2.6 times for majority-Hispanic and 4.2 times for majority-Asian areas. The authors report the disparities hold after controlling for population, income, education, land costs, transmission line proximity and state effects, and they attribute part of the pattern to the legacy of redlining, since transmission lines were often routed through Black neighborhoods. The researchers say hyperscale operators like Meta's facilities tend to be built in white rural areas, while smaller co-location sites are more often placed in communities of color. The study recommends data-center-specific zoning, racial equity in planning, and racial impact statements before construction.

Why it matters

The study suggests that where data centers end up is not only a matter of power lines and land prices, since the disparities persist after those factors are accounted for. Residents of majority-Black, Hispanic and Asian neighborhoods are more likely to live near these facilities than residents of white neighborhoods in the same states, and the authors link part of the pattern to redlining, when transmission lines were routed through Black neighborhoods. The authors' proposals of data-center-specific zoning, racial equity in planning and racial impact statements before construction point to how local decision-making could change if they were adopted.

Sources · 1