Data center site selection has an expensive back half. Transmission studies, geotechnical work, environmental review, and entitlement counsel are all necessary, and all of them cost real money per site.
The front half is where that money gets pointed, and it is also where most candidate sites should fall away. A campus needs power at a scale few parcels can reach, land that survives setbacks and floodplain, a cooling plan, fiber, and a jurisdiction willing to say yes. Much of that is measurable before anyone is hired.
1. Start from the facility profile, not the map
A screen is only as good as the profile it screens against. Write it down first: the load you are planning for, the acreage the campus needs after setbacks, the cooling approach, and how far from a population center your latency budget lets you go.
An edge facility and a large campus share almost no constraints. Screening a region without a profile produces a list of big parcels, which is the list you already had.
2. Measure power as distance and voltage, not as a yes
Public transmission and substation datasets show where high-voltage lines and substations run, often with their voltage class. A screen can measure each candidate's distance to lines of the class the load needs, and flag the sites that would need a long new line or a new substation.
What public data cannot show is what is spare. A high-voltage line at the fence tells you nothing about whether the utility can serve another large load from it, or when. That is an interconnection question, and it belongs to the utility.
3. Net the acreage down to what can be built
Gross parcel acreage flatters almost every site. Subtract wetlands mapped in the National Wetlands Inventory, FEMA floodway and floodplain, slopes the pad cannot absorb, and visible easements — pipeline and transmission corridors cross a surprising number of large parcels.
What is left is net buildable acreage, the number that should be compared across candidates. It is also where assemblage shows up: a footprint spread across six owners is a different project from one parcel with one deed.
The same netting decides utility-scale solar sites, and solar farm land suitability analysis walks through it layer by layer.
4. Place each site on the fiber and water maps
- Fiber: published long-haul routes, carrier points of presence and exchange locations, and the path from the site to them. Latency-sensitive loads care about distance; most others care about route diversity.
- Water: utility and district service-area boundaries, distance to mains, and treatment capacity where it is published. A site outside any service area needs a different cooling plan or a well-and-permit conversation.
- Sewer and discharge: whether the site is served, because blowdown from evaporative cooling has to go somewhere.
At the screening stage each of these is a distance and a boundary check. None of them establishes what a provider will actually commit to serve.
5. Read the zoning path before the terrain
County and municipal zoning layers, the comprehensive plan's future land use, and overlay districts together indicate whether a data center is a permitted use, a conditional one, or a rezoning. That difference is often measured in months, and sometimes in whether the project happens at all.
Neighbors matter too. A campus beside a subdivision faces a different public hearing from one surrounded by industrial land, and the parcel map shows which kind of neighbor each site has.
6. Check access and earthwork
Road functional class, weight limits, and the distance to an interchange decide how construction traffic and heavy equipment reach the site. LiDAR-derived slope and relief give a first estimate of the cut and fill a level pad would need.
Neither usually rules a site out on its own. Between two otherwise close candidates, they often decide it.
7. Rank the survivors on the combination
After the hard constraints, a regional list is usually far shorter. The survivors are ranked on the combination — power distance, net acreage, fiber position, water service, entitlement path, and earthwork — with the weighting written down in advance.
Each ranked site should carry its evidence and its open questions, so the next conversation with a utility or a consultant starts from a specific site and a specific question. That is what our industrial and data-center screening delivers; the records behind it are set out in property data sources for commercial analysis.
What a desktop screen does not settle
- Available transmission capacity, interconnection timing, and upgrade cost — utility answers, usually through a formal study.
- Water allocation, and whether a provider will serve a large new industrial load at all.
- Geotechnical conditions, contamination below the surface, and what a Phase I environmental site assessment finds.
- The entitlement outcome, incentives, and the local politics of a large new load.
The screen decides which handful of sites deserves that work. On the right handful the back half gets cheaper; on the wrong one it gets expensive fast.
