Key takeaways
- Define the target project type first; it sets every threshold that follows. One hyperscale search set minimums of 200 acres and 400 MW.2
- Knockout filters remove most parcels cheaply, using national public layers for transmission, substations and flood zones.34
- Score power on likely capacity and timing, not just distance to a line; average waits for large grid connections were reported at about four years in 2026.5
- Treat mapped constraints as screening flags: federal wetland maps are not regulatory boundaries.6
- Adjacent parcels in the same portfolio can combine into a stronger site.
- Verify the top tier with real conversations before calling anything a data center site.
01Who screens portfolios, and why
Large landowners such as timber companies, agricultural holders, energy companies, railroads, industrial owners and public agencies often hold hundreds or thousands of parcels. Some of that land may be worth far more as a data center site than in its current use. The challenge is finding which parcels, without commissioning a study on each one.
Developers and investors also screen portfolios when evaluating land partnerships or acquisitions. In both cases, the goal is the same: a ranked, evidence-based list of parcels that deserve a closer look, and a clear reason why the rest do not.
The method is not new. A 1998 GIS study that sited a national data center facility examined more than 50 data layers, applied hard screens such as flood-zone share and seismic risk first, and planned on-site checks afterward to verify the results.1 Today’s data are far richer, but the logic of cheap screens first and expensive verification last still holds.
Fig. 1The portfolio screening sequence
- 01
Define the target
Project type, load, acreage and timing.
- 02
Knockout filters
Mapped, conservative pass or fail tests.
- 03
Score survivors
Weighted criteria traced to data points.
- 04
Tier
Pursue now, hold, or not a fit.
- 05
Verify the top tier
Utility, county and field checks.
02Start by defining the target
A parcel can be a strong edge site and useless for a hyperscale campus. Before running any filters, decide which project types you are screening for. Our guide to site needs by data center type covers the differences in detail.
| Project type | Typical load | Typical land |
|---|---|---|
| Edge | 1–10 MW | 1–10 acres |
| Enterprise and colocation | 10–60 MW | 10–75 acres |
| Hyperscale campus | 100–500 MW | 150–600 acres |
| AI and gigawatt campus | 500 MW – 1 GW+ | 500–2,000+ acres |
Real searches bear out the scale. One broker’s request on behalf of an unnamed hyperscaler sought sites of at least 200 acres with 400 MW available by mid-2027 and potential to grow toward 1 GW.2 A 2026 data center FAQ posted by the City of Gibraltar, Michigan, notes that about 10 acres suits many projects, but hyperscale developments can need hundreds of acres, often 200 or more, for phased buildings, on-site electrical infrastructure, generator yards, internal roads, construction staging and stormwater management.7
These ranges set the minimum contiguous acreage and the transmission voltage that matter for each screen. A portfolio might be screened for two or three types at once, with separate thresholds for each.
03Stage one: knockout filters
Knockout filters remove parcels that clearly cannot work. They should be simple, mapped and conservative, so good sites are not lost by accident.
- Minimum contiguous acreage for the target type, counting adjacent parcels in the same ownership together.
- Distance to transmission at a voltage suited to the target load, such as 115 kV or 138 kV and above for larger campuses, and 230 kV to 500 kV or higher for the largest.
- Share of the parcel in a FEMA floodway or Zone A/AE.4
- Share of the parcel mapped as wetlands or open water.6
- Severe slope across most of the parcel.
- Protected land, conservation easements or land uses that rule out development.
Public national layers make this stage cheap. The federal HIFLD transmission dataset maps lines from 69 kV to 765 kV,3 and its companion substation layer covers facilities at 69 kV and above, with lower-voltage coverage described as incomplete.8 An Oak Ridge National Laboratory effort to build national electric infrastructure data from open sources reported about 573,000 transmission miles and 57,000 substations, updated annually.9 FEMA’s National Flood Hazard Layer shows flood zone, base flood elevation and floodway status for a location,10 but its digital coverage is not nationwide; where it is missing, users need the effective Flood Insurance Rate Map from FEMA’s Map Service Center.4 Our guide to GIS data sources for data center site selection lists more.
Fig. 2From portfolio to shortlist
Illustrative- Parcels in portfolio1,200
- Below minimum acreage-720
- Too far from transmission-260
- Flood and wetland share-95
- Slope or protected land-45
- Below scoring cutoff-60
- Top-tier parcels20
parcels
Distance thresholds should be set by the target load and the local grid, not by a fixed rule. A parcel slightly beyond a threshold but near a planned substation may deserve a second look. The guide to transmission voltage for data centers explains how voltage relates to deliverable capacity.
Data quality matters at this stage. Mapped wetlands, flood zones and transmission data can be outdated or imprecise at the parcel level; the Fish and Wildlife Service, for example, says wetland boundaries may have changed since the source imagery was captured and that on-the-ground inspection may revise them.6 Treat knockout results as strong indications, and keep a short list of borderline parcels that failed narrowly so they can be checked by hand rather than dropped.
04Stage two: scoring the survivors
Parcels that pass the knockouts get scored across weighted criteria. Weights depend on the target type: power dominates for large campuses, while fiber and proximity to users matter more for edge and colocation.
| Criterion | What is measured | Data source |
|---|---|---|
| Power proximity | Distance to suitable lines and substations, voltage | Transmission and substation mapping3 |
| Power outlook | Signals of capacity or congestion in the area | Public utility and ISO/RTO information, outreach |
| Buildable acreage | Acres left after constraints, and shape | Constraint mapping |
| Zoning | Current zoning and plan designation | County and municipal GIS, codes |
| Fiber | Distance to long-haul and metro routes | Carrier route data |
| Water | Proximity to municipal service or groundwater | Utility and state data |
| Neighbors | Homes and sensitive uses near the parcel | Parcel and land use data |
| Access | Road frontage and heavy-haul access | Road network data |
Most GIS scoring follows the weighted overlay method: each criterion is reclassified onto a common suitability scale, such as 1 to 10, each layer gets a percentage weight, and the weighted layers are summed.11 Our guide to site scoring and weighting methods covers the choices in more depth.
Fig. 3Sample weights for a large-campus screen
Illustrative- Power outlook30
- Power proximity15
- Buildable acreage15
- Zoning10
- Neighbors10
- Fiber8
- Water7
- Access5
% of total score
Power outlook earns heavy weight because timing, not distance, usually decides: as of 2026, the average wait for a large grid connection was reported at about four years nationally.5 Neighbors matter more than they once did, too. One private tracker reported that local opposition disrupted about $130 billion of U.S. projects in the first quarter of 2026.12
Keep scores transparent. Every score should trace to a specific data point, so a reviewer can see why a parcel ranked where it did.
Avoid letting a high total hide a critical weakness. A parcel with excellent fiber, zoning and access but no realistic path to power is not a strong large-campus site. Many teams cap the overall score when a critical criterion falls below a minimum, so weak power cannot be offset by strengths elsewhere.
05Stage three: tiering and verification
Scored parcels are sorted into tiers. A common approach is three groups: parcels worth immediate outreach, parcels to hold for future conditions, and parcels that do not fit current targets.
Fig. 4What each tier means
Few parcels
Tier 1: pursue now
- Passes all knockouts
- Strong power outlook
- Workable zoning path
- Goes to utility outreach and feasibility
Tier 2: hold
- Passes knockouts
- Waits on a substation, line or rule change
- Rerun when regional plans change
Tier 3: not a fit
- Fails a knockout or scores low
- Documented reason recorded
- May suit a smaller project type
The top tier then gets verification. That means informal conversations with the utility about capacity in the area, a check of local zoning and attitudes, and a review of anything the mapped data might have missed, such as recent land use changes or new transmission plans. Verification should also cover the utility’s commercial terms: AEP Ohio’s data center tariff, for example, generally bills new loads above 25 MW for up to 85% of contracted capacity.13 Some top-tier parcels drop out here, which is normal.
Screens also age. Utilities build new substations, transmission plans change and zoning gets updated, sometimes abruptly: Washington County, Maryland, adopted a 12-month moratorium on data center applications in June 2026.14 A parcel that falls into the hold tier today may move up in a few years, and a top-tier parcel can drop, so it is worth rerunning the screen when major infrastructure plans or local rules change in the region.
06What the screen should produce
- 01A ranked list of parcels by target project type, with scores and the reasons behind them.
- 02Maps of the top-tier parcels showing constraints and nearby infrastructure.
- 03A short profile of each top-tier parcel: acreage, buildable estimate, power context, zoning and open questions.
- 04Notes on parcel combinations that create larger sites.
- 05A recommendation on next steps: feasibility studies, utility outreach or market testing.
Keep the profiles short enough to share and specific enough to act on. Each open question should name who can answer it, whether that is the utility, the county planner or a surveyor, so the next step is obvious.
For a landowner, the top-tier profiles are the starting point for deciding whether to market the land, wait or pursue powered land positioning with the utility; the landowner’s path guide covers what comes next. For a developer, they feed directly into feasibility and early desktop and field assessment. If you want a second view on a portfolio, you can get a site reviewed.
07Common screening mistakes
- Treating distance to any power line as a power score, without regard to voltage or capacity.
- Scoring parcels individually and missing adjacent parcels in the same ownership.
- Relying on gross acreage instead of estimating buildable acreage.
- Treating mapped wetlands or flood zones as final answers rather than flags for field review.6
- Setting knockout thresholds so tight that good parcels are dropped on data errors.
- Using one set of thresholds for every project type, so edge-scale parcels are judged against hyperscale needs.
- Ignoring local rule changes: a new moratorium or ordinance can move a top-tier parcel down quickly.
- Skipping verification and presenting desktop scores as conclusions.
Most of these mistakes come from the same root: asking a desktop screen to answer a question only the utility, the county or a field crew can answer. The screen’s job is to decide where those conversations happen first, and in what order.
Common questions
How many parcels typically pass a data center screen?
It varies with the portfolio and the target. A portfolio of rural tracts near major transmission may produce many candidates, while a portfolio of small or scattered parcels may produce few or none. Knockout filters on acreage and power proximity usually remove the majority. What matters is that the parcels that survive are documented well enough to justify the next step.
What data is needed to screen a land portfolio?
At minimum, parcel boundaries and ownership. The screen then layers in transmission lines and substations,3 FEMA flood zones,4 wetlands mapping,6 topography, zoning, fiber routes, roads and nearby land uses. Much of this is available from public and commercial GIS sources. Utility capacity information is the hardest to get from mapped data and usually needs direct outreach.
How large does a parcel need to be for a hyperscale data center?
There is no fixed minimum, but hyperscale searches commonly start at a couple of hundred acres. One hyperscaler search sought at least 200 acres with 400 MW available,2 and a 2026 local-government FAQ notes that large projects are frequently 200 acres or more, with room needed for substations, generator yards, roads and stormwater.7 Buildable acreage, not gross acreage, is what counts.
Can a portfolio screen tell me what my land is worth to a data center developer?
A screen tells you which parcels have the attributes developers look for and how they compare, which is the foundation for any value discussion. It does not set a price. Value depends on power availability and timing, buildable acreage, local market demand and deal terms, and it is ultimately tested in negotiation with buyers or tenants.
Should a landowner screen a portfolio before talking to developers?
It usually helps. A screen shows which parcels are likely to interest developers and why, so the owner can focus on the strongest sites and answer early questions about power, acreage and zoning with evidence. It also reveals parcel combinations the owner may not have considered. Owners who arrive with a clear picture tend to have more productive conversations and avoid tying up weak parcels in long options.
Notes
- 1.Esri, “Siting a National Data Center Facility Using ArcView GIS,” 1998. proceedings.esri.com
- 2.Data Center Dynamics, “Hyperscaler seeks 200-acre site for 400MW data center in US,” n.d. datacenterdynamics.com
- 3.Data.gov (Homeland Infrastructure Foundation-Level Data), “Electric Power Transmission Lines,” n.d. catalog.data.gov
- 4.Federal Emergency Management Agency, “National Flood Hazard Layer,” n.d. hazards.fema.gov
- 5.Electrical Contractor Magazine, “How Long Does It Take to Power a Data Center?,” 2026. ecmag.com
- 6.U.S. Fish and Wildlife Service, “Wetlands Data Limitations, Exclusions and Precautions,” n.d. fws.gov
- 7.City of Gibraltar, Michigan, “Wayne Forward Data Centers FAQ,” 2026. cityofgibraltarmi.gov
- 8.Nicholas Institute, Duke University, “Electric Power Substations,” n.d. nicholasinstitute.duke.edu
- 9.U.S. Department of Energy, “Creating Foundation Electric Energy Infrastructure Data from Open Sources,” 2023. energy.gov
- 10.FEMA Emergency Management Institute, “National Flood Hazard Layer (NFHL),” n.d. emilms.fema.gov
- 11.Esri, “How Weighted Overlay works,” n.d. doc.esri.com
- 12.Data Center Watch, “Q1 2026: Data Center Watch Report,” 2026. datacenterwatch.org
- 13.Power Engineering, “Ohio utility regulators approve AEP’s contested data center tariff proposal,” 2025. power-eng.com
- 14.Washington County, Maryland, “Washington County Board of County Commissioners Approves Temporary Moratorium on Data Center Applications,” 2026. washco-md.net
Have a site in mind?
Get a straight answer on your land.
Send a parcel number, an address, a map pin or a target load. We’ll tell you what it can support and what it would take.
This guide is general information about data center site selection. It is not engineering, legal, tax or investment advice. Requirements vary by state, utility and county, so confirm the specifics for any site with the relevant authorities and advisors.
Terms in this guide
Related guides
More in Feasibility, diligence & deals
- What Does a Data Center Feasibility Study Cover?
- Data Center Site Due Diligence Checklist
- Selling or Leasing Land to a Data Center Developer
- How Long Does Data Center Site Selection Take?
- Desktop Screening vs. Field Assessment for Data Center Sites
- How to Read a Data Center Feasibility Report
- Letters of Intent for Data Center Land Deals: Binding Terms, Exclusivity and Pitfalls
- Utility Will-Serve Letters: What They Mean and What They Don’t
- How Developers Underwrite Data Center Sites
- Data Center Construction Costs per MW and the Site Factors That Drive Them
- Due Diligence Periods and Purchase Contingencies for Data Center Land
- Confidentiality, NDAs and Code Names in Data Center Site Searches
- What to Check on a Data Center Site Visit: A Field Checklist
- Public Data Sources for Power Capacity: EIA, FERC, OASIS, Queues and Hosting Maps
Guide 4 of 18 in Feasibility, diligence & deals · all 242 guides · glossary · calculators
