SMR uranium demand by design

Methodology and assumptions

Per-unit fuel intensity

Enriched product per year = MWth x capacity factor x 365 / (1,000 x burnup in GWd/t), the WNA fuel-cycle method. Natural uranium = product x feed factor, where the feed factor comes from the site's enrichment mass balance at the design's assay, 0.711% natural feed and the tails assay (default 0.20%, adjustable). tU converts to Mlb U3O8 at 2.5998 Mlb per 1,000 tU (2.59979 lb per kgU). Thermal power is the developer's figure where published, otherwise MWe divided by the class efficiency.

Class defaults (modeling assumptions)

Light-water oxide, LEU: 33% thermal efficiency, 45 GWd/t, 90% capacity factor, first core 75 tU product per 1,000 MWe. WNA fuel-cycle overview reference reactor: 33% thermal efficiency, 45 GWd/t burnup, 90% capacity factor, and a first core of about 75 tU of enriched uranium per 1,000 MWe, scaled by MWe for smaller units (the scaling is our assumption). sourceTRISO, HALEU: 40% thermal efficiency, 150 GWd/t, 90% capacity factor, no first-core rule. Site assumptions: 40% thermal efficiency (the ratio X-energy publishes for the Xe-100, 80 MWe from 200 MWth, applied to the class), 150 GWd/t burnup (no developer in our sources publishes one for TRISO pebbles or compacts), 90% capacity factor. No first-core rule: TRISO core masses are not published.Metal fuel, HALEU: 40% thermal efficiency, 100 GWd/t, 90% capacity factor, no first-core rule. Site assumptions: 40% thermal efficiency, 100 GWd/t burnup and 90% capacity factor for sodium-cooled metal-fuel designs; none of these is published by TerraPower or Oklo in the sources we cite. First cores are used only where a developer has stated one (Natrium).Liquid salt fuel, LEU: 44% thermal efficiency, 45 GWd/t, 90% capacity factor, no first-core rule. Site assumptions: 44% thermal efficiency and a light-water-like 45 GWd/t burnup for liquid-fuel molten-salt designs on standard-assay LEU; Terrestrial Energy publishes neither in the sources we cite.

Scenarios

Low: firm projects only: Only operating, under-construction and licensing projects count, at the demand model’s weights for those statuses; announced, design-stage and planning rows count for nothing.Base: demand-model weights: The site demand model’s realisation probability by status (operational 1.0 down to announced 0.35), the same weights the forward demand curve on /reactors uses.High: every announced project: Every non-cancelled project in the tracker is built on its target year, fleet frameworks included (TVA/ENTRA1 6 GW, Meta/Oklo, Google/Kairos).Each project ramps linearly from its target year over 2 years, the same shape as the forward demand model on /reactors; a project's units are its tracker MW divided by the design's MWe.

First cores

A first core is added once, the year before the target year, weighted by the project’s probability, only where a developer has published a core mass or the WNA light-water reference applies; it is never modeled as a multiple of annual reloads.

What is excluded

Large-reactor restarts and existing plants appear in the pipeline tracker because they are AI offtake deals; their fuel is the operating fleet’s and is already in the demand model, so they are not SMR demand here. The cancelled Carbon Free Power Project and the undesigned Oracle framework are excluded too. The IMSR400 is listed for the design comparison but has no project in the tracker.

How to read it

The figures are natural uranium (U3O8 equivalent) that enrichers would have to buy, not what utilities will contract; compare them with the roughly 180 Mlb per year the operating fleet needs. HALEU product tonnes are the enrichment-side constraint: compare them with the HALEU supply guide’s delivered and planned capacity.

Design parameters last reviewed 2026-10-19. See the site-wide methodology for provenance labels and the other datasets, and the first-core and reload guide for the worked light-water example the method is checked against.