Voltane is a screening tool. It narrows a country of possible sites down to a short list worth a closer look. It does not pick sites, and it does not replace engineering, permitting or due diligence.
It has a nuclear part and a Texas part. Voltane Nuclear screens 504 candidate sites in 49 states for three reactor classes: microreactors, small modular reactors and large reactors. Voltane Texas finds land for utility-scale solar, storage and hybrid projects across 234 Texas counties in ERCOT.
The data
Everything starts from public data. Each source keeps its own date, and the app shows it next to every number.
- Power plants and coal retirements: EIA-860M (August 2026 inventory, preliminary) and EIA-860 (2025 final).
- Reactor status and early site permits: the U.S. Nuclear Regulatory Commission, checked October 4, 2026.
- Transmission: the LBNL harmonised dataset and the final public HIFLD release (data to 2024-09-30); in Texas, OpenStreetMap lines reconciled with HIFLD.
- People: 2020 Census blocks and urban areas.
- Hazards: USGS seismic hazard and Quaternary faults, the USACE dam inventory, FAA airports, EIA gas pipelines, and the JRC / Copernicus modelled river flood map (not a FEMA zone).
- Land (Texas): TxGIO statewide parcels, USGS 3DEP elevation, NLCD land cover, PAD-US 4.1 protected land and the National Wetlands Inventory.
- Data centers: OpenStreetMap (741 mapped campuses) and Epoch AI (71 US data centers), merged so each counts once: 784 in all. Neither list is complete.
How a score is built
Each site is measured against 11 criteria: distance to transmission, generation already at the site, distance to major water, people nearby, distance to a population center of 25,000 or more, seismic hazard, faults, flood, other hazards, distance to data centers, and what the site already is.
Each measurement becomes a component between zero and one on a straight-line ramp. A line on the site earns full marks; a line far away earns none. Each reactor class weights the components differently, and the score is the weighted sum on a scale to 100.
| Criterion | Micro | SMR | Large |
|---|---|---|---|
| Transmission access | 8 | 14 | 18 |
| Existing interconnection at the site | 10 | 10 | 8 |
| Major surface water nearby | 3 | 10 | 18 |
| Population exposure | 8 | 10 | 14 |
| Distance to a population center | 4 | 6 | 8 |
| Seismic hazard | 8 | 10 | 10 |
| Distance to a Quaternary fault | 5 | 6 | 6 |
| Flood hazard | 8 | 8 | 6 |
| Nearby external hazards | 6 | 6 | 4 |
| Load proximity (data centers) | 22 | 10 | 2 |
| Reusable industrial site | 18 | 10 | 6 |
Microreactors lean on nearby load and reusable industrial sites; large reactors on grid, water and low population. Weights are product assumptions. Data (JSON)
Weights are product assumptions, not regulation. No regulator sets them. Users can change them in the app, and those custom weights are assumptions too.
There is no fixed exclusion radius anywhere in Voltane. A population center nearby is scored and flagged, never used to exclude a site. Seismic hazard and faults are risk signals, never pass or fail. Water is proximity only: it does not establish water rights or cooling capacity.
Screening only — not a site suitability determination.
Proxies, not capacity
Some signals are easy to misread. Existing interconnection in Nuclear is the generation that sits, or sat, at the site: a sign that a switchyard and a transmission corridor exist. Queue competition in Texas is the MW already in the ERCOT interconnection queue nearby. Both are proxies. Neither is available grid capacity, and Voltane never reports available MW at a substation.
In Texas, project size is "indicative": the largest flat, unconstrained block of land divided by 4.25 acres per MW. Real layouts depend on setbacks, roads and equipment.
Missing data is unknown, not zero
When a criterion has no data, it earns no points and the site's completeness goes down. It is never scored as favourable, and it never means "no risk". For SMR scores, 79% of candidate sites have every criterion measured.
What Voltane cannot tell you
- Water rights or cooling capacity. Being near a river is not a right to use it.
- Real interconnection capacity. Only a utility or grid operator study can say how much a substation can take.
- Landowner willingness. Voltane publishes no landowner names, and a good parcel is not a willing seller.
- Permitting outcomes. Screening is not licensing. Voltane never predicts what a regulator or utility will decide.
- Site-specific engineering. Seismic, geotechnical and flood studies happen on the ground, not on a map.