The Nuclear Fuel Cycle

Part VII of The Nuclear Fuel Cycle: A Complete Primer

Chapters 26–27 close the book’s argument. Chapter 26 restricts itself to documented commitments (reactors under construction, licenses granted, capital committed, contracts signed) and asks whether conversion, enrichment, fabrication and workforce can keep pace with what is already in the field. Chapter 27 then states a conclusion: the destination is a materially closed fuel cycle, reached through a four-stage roadmap and fed by material already above ground. The Part moves from the 2026 pipeline through the 2035–2045 decision decade to the physical endpoint. The facts it leaves standing are about 795,000 metric tons of US U-238 feedstock, a fast breeder yield about 60 times the once-through yield, fission products that decay substantially over about 300 years, and a governing constraint of sequencing and political patience.

Chapter 26 — Next Twenty Years

Documented commitments only. The chapter limits its evidence to documented commitments. “Pipeline” means projects with legal, financial, regulatory or physical weight behind them. The central 2035 question is whether conversion, enrichment, fabrication and workforce capacity can keep pace with construction commitments already in place. In 2035–2045 the market begins to separate demonstrated from expected performance, and the closed-cycle decision becomes practical.

The construction pipeline. IAEA PRIS data showed about 60–65 reactors under construction as of late 2025. China targets about 10 new units approved a year, predominantly large PWRs of the Hualong One design. Its HTR-PM pebble-bed demonstration at Shidaowan was connected to the grid in December 2021 and declared in commercial operation in December 2023. India’s 500 MWe PFBR at Kalpakkam reached first criticality on April 6, 2026, making India the second country after Russia to do so at this scale in the contemporary period. Commercial generation is expected later in 2026, after power ascension testing and Atomic Energy Regulatory Board clearance. Vogtle 3 and 4, the first new large US light-water units in about three decades, documented in detail what large construction costs under current conditions; France and South Korea show unit costs falling on standardized designs.

Advanced reactors in the ground. Kairos Power’s Hermes 1 at Oak Ridge is the first non-light-water reactor to receive an NRC construction permit in more than 50 years. Hermes 2, the first power-producing Generation IV reactor so licensed, broke ground in April 2026 and will supply up to 50 MW to the TVA grid. TerraPower’s Natrium at Kemmerer, Wyoming received its permit on March 4, 2026 and began nuclear construction on April 23, 2026. GE-Hitachi’s BWRX-300 holds a CNSC construction license for Darlington, the first for a power-producing advanced reactor in North America. The UK signed its Rolls-Royce SMR contract on April 13, 2026, with Wylfa as first site. The fleet stood at about 440 reactors and 2,600 TWh a year in early 2026. “The pipeline is real; completion is not automatic.”

Conversion and enrichment. Uranium ore is not the near-term binding problem; the concern is geographic, with about 40% of production in Kazakhstan. Conversion is the chokepoint. ConverDyn’s Metropolis Works, the only operating US commercial plant, had about 15,000 tU a year of original nameplate, was reduced to 7,000 tU in 2017, restarted in July 2023, and projects more than 10,000 tons of UF₆ in 2026, against US demand of about 18,000 tU. Rosatom held about 40% of global SWU capacity in 2026 and supplied about 24% of US enriched uranium demand in 2023. On January 5, 2026 the DOE issued $2.7 billion in enrichment task orders, $900 million to each of three US producers.

HALEU and advanced fuel. HALEU, enriched to 10–20% U-235, is required by Natrium and other advanced designs. The only operating US source is Centrus Energy’s Piketon demonstration at about 950 kg a year. TRISO particle fuel is required for high-temperature gas reactors. The EBR-II and Natrium lineage needs metallic uranium-zirconium or metallic uranium-plutonium fuel alloyed with zirconium. Neither TRISO nor advanced metallic fuel is available at commercial scale in the West.

Regulation. The ADVANCE Act (July 2024) addresses licensing timelines, fees, allied regulators’ reviews and brownfield sites. Part 53, a technology-neutral, risk-informed performance-based licensing framework, was finalized March 25–26, 2026 and took effect April 29, 2026. It is the first new initial-licensing regulation since Part 52 in 1989. The NRC’s December 2023 rule lets an SMR’s emergency planning zone be set by consequence analysis, potentially at the site boundary instead of the traditional 10-mile radius. The safety physics is unchanged.

Workforce. Licensing can authorize projects, but it cannot supply trained people or qualified manufacturers, and the workforce rebuilds on timescales of decades. Operator licensing takes about 18 months of training plus an NRC examination. Darlington’s CANDU refurbishment captured about 8,000 lessons and outperformed the Bruce refurbishment before it.

2026–2035 and 2035–2045. The Crane Clean Energy Center, formerly TMI Unit 1, is undergoing restart under a 20-year Constellation–Microsoft agreement of September 2024, with Constellation investing about $1.6 billion, a $1 billion DOE loan closed in November 2025 and restart targeted for 2027. Natrium, targeted for the early 2030s, would give the first US sodium fast reactor data since EBR-II shut in 1994. Fuel supply, through HALEU, is the variable most likely to compress the schedule. The 2035–2045 analysis is scenario analysis. La Hague reprocesses about 1,700 tHM a year, but Astrid’s cancellation in 2019 left France’s cycle partially closed. Russia’s BN-800 has a breeding ratio of about 1.0. The BN-1200M, licensed in 2025, targets about 1.4, which would make it a net fissile material producer; a fleet at that ratio has a doubling time of 15–30 years.

What must be true. US enrichment fell from about 64% of the Western market in the mid-1980s to under 10%. Three failure modes are named: an advanced-reactor demonstration problem met by a disproportionate response, unresolved HALEU, and continued reprocessing deferral. Carter’s 1977 deferral was executive, Reagan modified it in 1981, and no statutory prohibition exists. Each decade produces what the next requires: “The sequencing is the argument.”

Chapter 27 — The Closed Cycle Endgame

A conclusion, stated. The preceding twenty-six chapters were analytical; this one synthesizes and says plainly what the primer thinks. The destination is a materially closed fuel cycle. Fertile U-238 becomes fuel, actinides are returned to the reactor and fissioned, the waste stream is dominated by fission products, and new mining matters far less because the principal fuel resource is already above ground. The endpoint is physically achievable. It is not commercially mature in the United States.

Partial and material closure. France separates uranium and plutonium and recycles the plutonium, roughly 1% of spent fuel by mass, as MOX in thermal reactors. That is a partially closed cycle: the minor actinides (neptunium, americium, curium) stay in the high-level waste. A materially closed cycle uses fast reactors to consume plutonium and minor actinides and to breed Pu-239 from U-238. The test is whether separated actinides are fissioned. “The once-through cycle stores actinides. The closed cycle uses them.”

The four-stage roadmap. Stage 1, from the present to about 2035, is demonstration: Natrium, the BN-1200M if its schedule holds, and China’s CFR-600, with operating data as the goal. Stage 2, about 2035–2045, is commercial selection and the decision whether to build reprocessing around successful fast reactors. Stage 3, about 2045–2060, is partial industrial closure, as depleted uranium enters as fertile feedstock. Stage 4, after about 2060, is a materially closed cycle in which the fuel constraint becomes the rate at which reactors and facilities can be built.

The once-through yield. A large LWR produces about 45 GWh of electricity per metric ton of uranium. About 160 metric tons of natural uranium feed one GWe-year, about 8,760 GWh at full capacity or about 7 TWh at 80%. That energy comes from the 0.7% that is U-235. Measured against the full energy content of the heavy metal, the once-through cycle is about 1.5–2% efficient. A mature fast breeder cycle yields about 60 times as much: about 2,700 GWh, or 2.7 TWh, per metric ton.

The stockpile. The chapter’s boxed arithmetic runs: about 700,000 metric tons of depleted uranium plus roughly 95,000 metric tons of U-238 in more than 95,000 metric tons of spent fuel gives about 795,000 metric tons. At 2.7 TWh per ton that is about 2,130,000 TWh. Against US consumption of about 4,000 TWh a year, it is about 500 years. The book presents this as an energy-content calculation to be read as scale, since the exact figure varies with breeder performance, recycling losses and demand. The resource is not waiting to be discovered; it has already been mined, enriched and stored.

The waste dimension. Spent fuel is about 94–96% U-238, 0.8–1% U-235, about 1% plutonium, 3–4% fission products and less than 0.1% minor actinides. Np-237 has a half-life of about 2.14 million years, and Am-241 decays into it. In a closed cycle these actinides are fissioned in a fast spectrum. Cs-137 and Sr-90 have half-lives of about 30 years, so after about 300 years their activity is about one one-thousandth of the initial level. The problem shifts from geological-timescale isolation to engineered isolation for centuries.

Three constraints. Reprocessing has been demonstrated industrially in France and Russia, pyroprocessing experimentally at Idaho National Laboratory, and sodium fast reactors have run in Russia, France, Japan, India and the United States. What governs the timeline is sequencing, regulatory capacity and political patience. A breeder fleet cannot start on depleted uranium alone, and the book argues that the 1977 Carter deferral broke the sequence. No US commercial reprocessing has operated since the 1970s, and Part 53 does not by itself license the back end. The roadmap spans roughly fifty years, and France, the book argues, shows that continuity must be built deliberately.

The endpoint. The book describes Stage 4 as a physical state of the fuel cycle. Fertile feed comes from depleted uranium. For the fission-product stream, deep geological disposal or deep borehole disposal, in boreholes several kilometers deep, are the primary permanent approaches, with dry cask storage as interim management. At about 2.7 TWh per ton, the binding limit on generation becomes reactor deployment capacity.

The argument and the question. Five findings carry the case: about 3,000 tons of coal per kilogram of uranium, the 0.7% U-235 split, about 18,000–19,000 reactor-years with mortality below any other major source, a waste problem manageable when stated accurately, and an industrial base being rebuilt. “The constraint that actually governs the situation is not physics.” The question left open is whether institutions can make and sustain the roadmap’s decisions over its timescale. The primer’s task is to make that question understood accurately. “That is the purpose of a primer.”

Supporting appendices

• Appendix XII — Physics and Mathematics. “The Breeding Ratio and Doubling Time” gives BR ≈ η − 1 − L − A, with a sodium fast reactor example (η ≈ 2.45, L ≈ 0.03, A ≈ 0.10) giving BR ≈ 1.32. The doubling-time formula DT ≈ (fuel cycle time) / (BR − 1) gives 30 years at BR 1.2 and 15 years at BR 1.4 on a 6-year cycle, bracketing Chapter 26’s range. The resource estimate, 785,000 MT × 2.7 TWh/MT ≈ 2,119,500 TWh, gives about 530 years (range 400–600).
• Appendix XIII — Optional Problem Set. Problem 3-S, The Closed Cycle Case, is the full synthesis. One 10 MT cask holds about 9.0 MT of U-238, worth 24.3 TWh. BR 1.3 on a 6-year cycle gives a 20-year doubling time, and a single breeder seeded in 2035 reaches 16 reactors around 2115. Problem 3-1, The Efficiency Ceiling, puts the BN-1200M’s expected actual efficiency at about 44%. Problem 2-1, The 300-Year Threshold, shows Cs-137 at about 0.1% after 300 years while 99.99% of Np-237 remains.
• Appendix X — Industry and Landscape. Its Reprocessing and Recycling entry covers La Hague, Mayak, China’s emerging capacity and Rokkasho; other entries cover nuclear construction labor and advanced reactor developers.
• Appendix VIII — Glossary of Terms gives working definitions, with the chapter of first use, for every term marked here. Appendix IX — Regulatory and Policy Timeline places the 1977 deferral and 1981 Reagan reversal beside the 2025–2026 entries (BN-1200M license, task orders, Part 53, PFBR criticality, Wylfa) and the pending horizon: the Crane restart in 2027 and the full Russian import ban on January 1, 2028.

Longer technical notes and Top Stories tagged to Part VII will appear on this page as they are published.