San Antonio Councilman Says Data Centers Went From Bland Offices to Hyperscale Eyesores
Ric Galvan, who has watched data centers move into San Antonio since the 2000s, says the newest hyperscale builds are far harder for residents to accept.
San Antonio City Councilmember Ric Galvan has a long view of the data center boom, and he does not like where it has landed. When the first facilities arrived in his city in the 2000s, he recalled, they looked like ordinary office buildings and drew little complaint. The newest generation is a different animal: hyperscale campuses covering millions of square feet and packed with the physical hardware that runs AI workloads.
That shift in scale is at the center of a growing fight over how much land, power and public patience the AI buildout can consume. Galvan's district has become a test case because the city sits on cheap land and an increasingly strained grid, and operators keep coming. Local officials across Texas and the Southwest are fielding the same complaints about noise, water use and property values that San Antonio is now hearing.
The numbers explain the change. A single hyperscale site can dwarf the office-park data centers of twenty years ago, and the largest campuses now rank among the biggest power customers in their regions. Utility planners in Texas have warned that data center demand is a leading driver of new generation and transmission spending, costs that eventually reach ratepayers. For residents, that means the building next door is no longer a quiet tenant with a few server racks.
Camouflage has become one answer. Operators have tried screening campuses with tree buffers, earth berms and muted facades, and some have pitched projects as low-profile neighbors rather than industrial plants. The approach can soften a zoning hearing, but it does not change the load the facility places on the grid or the water it pulls for cooling. Galvan's point is that the disguise only works until the scale becomes obvious.
What happens in San Antonio matters beyond the city limits because it previews the politics of the next wave of AI infrastructure. If residents and council members start rejecting projects over size and resource use, operators will face longer permitting fights and higher costs, and some may shift to jurisdictions with friendlier terms. The open question is whether design, siting and community payments can keep pace with the buildout, or whether the industry has already outgrown the goodwill that greeted its first, unassuming buildings.