The United States has 94 operating commercial reactors at 54 sites. It has dozens of additional sites that once hosted reactors and still carry the transmission, water, and zoning lineage of thermal generation. It has eight federal nuclear complexes that already host research reactors, isotope facilities, or weapons-grade fuel handling under NRC and DOE jurisdiction. Across that combined footprint, the number of large contiguous private parcels within 5 to 30 miles of any of those nodes that have been optioned, tied up, or systematically acquired by a developer with a nuclear thesis is close to zero.
That is the trade. Nuclear-adjacent industrial real estate is the single most mispriced category in US infrastructure today. The capital chasing reactors has not yet noticed the land that sits next to them.
What Got Arbitraged Already
The first wave of nuclear repricing happened on operating plants. Amazon Web Services signed a 1.92 GW behind-the-meter PPA with Talen Energy at Susquehanna in Pennsylvania, anchoring a campus directly adjacent to a running reactor. Microsoft signed a 20-year offtake with Constellation to restart Three Mile Island Unit 1 at 837 MW, with first power targeted for 2028. Vistra is in active negotiations to dedicate Comanche Peak output to hyperscale load in Texas. Energy Harbor's Beaver Valley has been pulled into the same orbit through Vistra's portfolio. Each of these transactions priced the reactor, the interconnection, and the fenceline real estate as a single coupled asset.
The repricing is dramatic. A megawatt of base-load nuclear behind a hyperscaler PPA now clears at numbers that imply land basis multiples of 20 to 50 times the agricultural alternative. That arbitrage is finished. The operating fleet has 94 reactors and 54 sites. The mathematics of bilateral negotiation with utilities that already own the dirt means the spread is captured by Constellation, Talen, Vistra, Duke, Southern, and Dominion. New entrants do not get to underwrite that vintage.
The second wave moved to SMR vendors. Oklo went public via SPAC. NuScale carries a multi-billion dollar market cap on a single NRC design certification. X-energy has raised over $700 million with Amazon as anchor investor. TerraPower's Natrium project at Kemmerer, Wyoming sits on a $2 billion DOE cost-share. Holtec is repurposing Palisades in Michigan. Rolls-Royce SMR, BWXT, Kairos Power, Westinghouse AP300: each is competing for the same scarce siting bandwidth. Vendor equity is where most retail and venture capital has gone. It is also where the bottleneck is least binding.
What The Vendor Race Hides
The structural constraint on SMR deployment is not which vendor wins the engineering race. It is site acceptability under the NRC framework. The Early Site Permit (ESP) and Combined License (COL) processes take 24 to 60 months once an application is docketed, and applications cannot be docketed without a site that clears seismic, hydrologic, demographic, transmission, emergency planning, and environmental thresholds. The ADVANCE Act of 2024 streamlined fee structures and review milestones, but it did not change the underlying geology, hydrology, or population density of the country.
Run the NRC filters across the lower 48. Seismic exclusions under 10 CFR 100.23 take out most of California, large portions of the intermountain west, and the New Madrid corridor through Missouri and Tennessee. Cooling water requirements eliminate the high desert and most of the Great Basin. Population density limits under Regulatory Guide 4.7 strip out the corridors where load actually wants to be. Thirteen states maintain active moratoria or restrictive frameworks on new nuclear construction. What survives is a constrained inventory measured in low thousands of eligible parcels, not millions of acres.
Vendor technology is fungible across that constrained set. A site that clears NRC siting filters today will host whichever SMR design reaches commercial readiness first, then host a second design for the next build-out. The site has option value across the entire vendor field. A vendor has option value only on sites it can actually license.
That asymmetry is the mispricing. Capital is pouring into the side of the trade with the lower option value.
The Adjacency Map Nobody Has Optioned
The eligible inventory clusters in patterns that almost no commercial developer is tracking systematically.
The federal nuclear complex is the deepest pool. Hanford in Washington, Idaho National Laboratory in eastern Idaho, Oak Ridge in Tennessee, Savannah River in South Carolina and Georgia, Los Alamos and Sandia in New Mexico, the Nevada National Security Site outside Las Vegas, and the Pantex plant in Texas. Each of these complexes carries decades of NRC and DOE precedent on seismic characterization, radiological baselines, emergency planning zones, and transmission interconnection. The federal land itself is not for sale. The private parcels within 5 to 30 miles of the perimeter inherit a regulatory tailwind that no greenfield site can replicate. NRC site characterization data on those regions is already in the docket. Counties around those complexes have grown up around nuclear employment and carry political acceptance that is otherwise impossible to manufacture.
The announced SMR deployment regions are the second pool. TVA's Clinch River site in Oak Ridge, Tennessee, where TVA filed the first SMR ESP application in the country. Dow's Seadrift facility on the Texas Gulf Coast, paired with X-energy for industrial heat and power. Energy Northwest's Columbia Generating Station in Washington, which has filed for SMR deployment adjacent to the existing reactor. Utah Associated Municipal Power Systems in Carbon County, Utah, originally tied to NuScale and now in flux. Kemmerer, Wyoming for Natrium. Each of these announcements anchors a regional acceptance corridor. The land that sits within commuting distance of each announced site will be the support inventory for the build-out economy, and eventually for follow-on reactor deployment by the same operator or a successor.
The retired reactor footprint is the third pool. Rancho Seco in California, Trojan in Oregon, Maine Yankee, Connecticut Yankee, Yankee Rowe, Big Rock Point in Michigan, Shoreham on Long Island, Zion in Illinois, Kewaunee in Wisconsin, Crystal River in Florida, San Onofre in California. The reactors are decommissioned, but the transmission rights of way, the cooling water permits, the industrial zoning, and in most cases the NRC site files remain. Several of these sites are being actively evaluated for SMR repower. The private land adjacent to each carries a residual nuclear acceptability that the local jurisdiction has already absorbed.
The combined inventory across federal complexes, announced SMR regions, and retired reactor footprints is large in absolute acreage but small relative to the capital that will need to deploy against it once the first commercial SMR comes online. Almost none of it has been optioned. The acquisition window is open because the institutional capital pools that will need this inventory in 2028 to 2032 do not yet have a mandate to underwrite land outside an existing utility relationship.
Why Transmission and Water Rights Travel With The Land
A nuclear-adjacent parcel does not just carry zoning and political acceptance. It carries infrastructure lineage that is functionally impossible to recreate.
Thermal generation, including every reactor ever licensed in the United States, requires substantial cooling water. The water rights that were adjudicated to support those reactors in the 1960s, 1970s, and 1980s priced water at agricultural and municipal baselines. Those rights, where they remain attached to surrounding parcels or to the broader watershed allocation, travel through chain of title. In western states under prior appropriation doctrine, senior water rights established for cooling purposes carry priority dates that no new application can match. A parcel that inherits a 1968 priority date on 10,000 acre-feet per year of cooling water carries a balance sheet asset that the current spot market for industrial water in the same basin would value at multiples of the dirt itself.
Transmission rights of way operate the same way. A 345 kV or 500 kV interconnection that was built to evacuate a reactor's output sits in place after decommissioning. The right of way, the substation footprint, and in many cases the interconnection queue position remain associated with the host parcel and adjacent landowners. New transmission requires CAISO, MISO, PJM, ERCOT, SPP, or NYISO interconnection queues that now extend 5 to 8 years. A parcel that inherits an existing high-voltage connection compresses that timeline to near zero on the queue side.
Foreign Trade Zone designations, where they exist around nuclear-adjacent industrial corridors, add a tariff and customs layer that compounds the underlying advantage. Several of the federal complexes sit within or adjacent to active FTZ subzones. The combination of NRC-relevant precedent, senior water rights, existing transmission, FTZ status, and local political acceptance is the package that the next decade of base-load industrial development will pay a premium for. None of those attributes can be manufactured. They can only be acquired with the dirt that already carries them.
The Forward Spread
The repricing has not started in the third category. It will.
By 2030, the first wave of merchant SMR deployments will be coupled to existing industrial campuses with nuclear-relevant infrastructure inheritance, not greenfield NRC license applications on raw land. The first commercial SMR to reach commercial operation in the United States will sit on a site with an existing thermal generation history, an existing interconnection, and existing local acceptance. The economics of any other approach will not close on the financing timeline. Land that fits this profile will reprice from agricultural and light-industrial baselines to power-coupled industrial baselines through this window, with the strongest moves on parcels that pair federal complex adjacency with retired reactor infrastructure.
By 2032, anchor tenants on nuclear-ready campuses will sign 30-year base-load PPAs with credit profiles that look closer to defense contracts than commodity power purchase agreements. Hyperscalers, defense primes, advanced manufacturing tenants, and electrochemical producers will underwrite ground leases and offtake agreements against the same set of constrained sites. The PPA tenor will extend to match the reactor economic life. The credit support behind those PPAs will be investment grade. The land underlying each PPA will be valued as a function of the entitled power, not the underlying acreage. Per-acre values on nuclear-coupled industrial real estate will be 3 to 5 times comparable grid-tied industrial parcels in the same region.
By 2035, the spread between nuclear-coupled industrial real estate and grid-tied industrial real estate will widen to 8 to 15 times on a per-acre basis. The driver will not be a generic nuclear renaissance. It will be the binding scarcity of sites that can actually host serial SMR deployment under the regulatory framework that exists. The capital that will need to clear that constraint will be measured in hundreds of billions. The supply of sites that clear it will not have grown.
The investors who hold paper on those parcels in 2028 will sell into a market that has no substitute inventory. The investors who try to buy in 2030 will be negotiating against a deployment schedule that has already started.
The land does not care which reactor wins. It cares only that one of them does, on schedule, on a site that someone else already controls.