The Nuclear Fuel Cycle

Why the cost of the entire fuel cycle should choose the next reactor

Home • Opinion • Mine to Disposal: Pricing the Closed Cycle • Opinion — Jeff Hill

Abstract

In the author’s view, the most consequential choice in the nuclear fuel cycle is which kind of reactor gets built next, because it fixes what every plant upstream and downstream will do for the next sixty years. That choice belongs to the people and companies who carry the cost. Today they cannot make an informed decision, because regulation and policy hide the true cost of the residues and the repository: the disposal fee has been zero since 2014, and the government holds the liability.

The piece follows one tonne of mined uranium through the once-through cycle, sets out the material balance of light-water and fast reactors, and uses a simple rule to show when a fast reactor would pay its way in material terms. It argues that the pace of any transition is set by how fast plutonium is recovered from spent fuel, that recovery waits on prices government now holds down, and that the job of policy is to stop distorting them, and states the strongest objections alongside the author’s answers. It extends the core argument of The Nuclear Fuel Cycle: A Complete Primer: that the destination is a materially closed fuel cycle, fed by material already above ground, and that the constraint that governs it is not physics. The opinions are the author’s; the facts underneath are cited.

One Drum

Chapter 4 of The Nuclear Fuel Cycle ended at the mine gate, with a drum of yellowcake loaded for transport. Follow the uranium in it all the way to the end, and the shape of the problem this article is about comes into view.

Take one tonne of mined uranium, a round number that makes the arithmetic easy. At an ore grade of 0.1 percent, a conventional mine had to move and mill about a thousand tonnes of rock to produce it, and the tailings from that rock are now capped and monitored at the mill site. The tonne goes to a conversion plant and becomes uranium hexafluoride. It then goes to an enrichment plant, where Chapter 6’s worked example applies: about 10.2 kilograms of natural uranium for every kilogram of fuel enriched to 4.5 percent, at a tails assay of 0.3 percent. Just under 100 kilograms of our tonne become reactor fuel. The other 900 kilograms leave the enrichment plant as depleted uranium, sealed in steel cylinders [1].

The 100 kilograms of fuel spend four or five years in a light-water reactor. At a typical burnup of about 50 gigawatt-days per tonne, roughly one gram of heavy metal is fissioned for each megawatt-day of heat, so about 5 percent of the fuel, around 5 kilograms, actually splits [2]. That is all the energy the tonne will ever deliver in a once-through cycle: about half of one percent of the uranium that was mined. The fuel leaves the reactor still about 94 percent uranium, carrying roughly 1.3 kilograms of plutonium bred from the uranium-238 during its time in the core [2].

Everything else is a residue, and each residue has a home. The mill tailings stay at the mine. The depleted uranium joins the stock at the former enrichment sites at Paducah and Portsmouth, originally about 800,000 tonnes of uranium hexafluoride in some 67,000 cylinders [4]; more than 730,000 tonnes remain [3]. The Department of Energy is converting it to oxide on a schedule its own Inspector General estimates may run to 2074 [4]. The spent fuel goes to a pool and then to a dry cask. The United States holds about 95,000 tonnes of it [5]. Worldwide, reactors had discharged 447,758 tonnes of heavy metal by June 2026, according to the IAEA, and about 322,000 tonnes of that is still in storage [5]. Its final home, the geologic repository described in Chapter 16, does not yet exist in the United States [1].

This is the argument of the article in miniature. The once-through fuel cycle is a chain of plants, each optimized for its own step, each passing its residue to the next or to storage. Mining is made cheaper, enrichment more efficient, burnup higher. Every one of those improvements is real, and none of them changes the outcome at the end of the chain: of every hundred tonnes mined, less than one is used, and the other ninety-nine are stored in two different forms for a very long time.

A sustainable nuclear fuel cycle has to be judged across the whole chain, from the mine to the repository. Judged that way, the most consequential choice is not about any single plant. It is which kind of reactor gets built next, because that choice determines what every other plant in the chain will do for the next sixty years. In an efficient market, that choice would be made on price, and the price would include the whole chain.

The Material Balance

Every reactor draws on, and adds to, a small set of material balances: ledgers of every tonne in and every tonne out. A light-water reactor draws on mined uranium and adds plutonium to the stock of spent fuel. A fast reactor draws on depleted uranium, the material in those cylinders at Paducah and Portsmouth, and either adds plutonium or consumes it, depending on how it is configured. Set out per gigawatt-year of operation, the two look like this.

TABLE 1. The fissile material balance, per gigawatt-year.

	Light-water reactor	Sodium-cooled fast reactor
Uranium mined	about 160–205 tonnes	none
Fertile feed	—	about 1 tonne of depleted uranium
Plutonium	adds about 250–295 kg to spent fuel	needs about 7 tonnes to start (about 4.6 tonnes of it fissile); then adds or consumes, by design
Plants it depends on	conversion, enrichment, fuel fabrication	reprocessing, fast-reactor fuel fabrication

Sources: Goldston [2]; World Nuclear Association [6]; Till and Chang [7]; Kütt et al. on BN-800 [25]; Doligez et al. [8]. The fast-reactor start-up figure is per gigawatt and includes fuel outside the core in cooling, reprocessing and fabrication: Till and Chang assume 3.0 tonnes of fissile plutonium in the core and 1.6 tonnes outside it [7]; the total assumes light-water plutonium is about two-thirds fissile. Plutonium from fuel cooled for decades is nearer 55 to 60 percent fissile, because plutonium-241 decays with a 14-year half-life; the 12-tonne cautious case allows for this. Doligez et al. give the same order of magnitude [8].

The ledger has a consequence that the phrase “breeders make their own fuel” tends to hide. A fast reactor cannot start itself. Its first load of plutonium has to come from somewhere, and today the only large source is the spent fuel of light-water reactors. One gigawatt of fast reactor needs the plutonium discharged by roughly twenty-five to fifty gigawatt-years of light-water operation, depending on the design and how long its fuel spends outside the core [7] [8].

A farmer would recognize the arrangement. A harvest can be eaten, or part of it can be kept back as seed. Seed cannot be planted until there is a harvest to keep it from, and the size of next year’s planting depends on how much was set aside this year. The light-water reactors are the harvest. The fast reactors are the planting. Reprocessing is the threshing floor where seed is separated from straw. The two reactor types are not rivals. One supplies the other, and the right fleet is a mix whose composition shifts as seed accumulates.

What the Residues Become

Read the material balance from the end of the chain rather than the beginning, and its significance changes. In a once-through cycle, depleted uranium is a liability: a corrosive chemical stock that has to be converted, packaged and eventually disposed of. In a fast-reactor cycle it is fuel. A gigawatt of fast reactor needs about a tonne of it a year. The roughly 490,000 tonnes of uranium in the American stock would supply a large fast fleet for longer than any planning horizon anyone would care to state [3].

The same reversal applies to spent fuel. Its plutonium is the start-up fuel the fast reactors need, and its uranium, still about 94 percent of its mass, is fertile feed. Reprocessing separates those materials from the fission products, which are then vitrified. Chapter 16 described what that does to the hazard profile of the waste. Fission products dominate the first few centuries and decay toward the radiotoxicity of the ore the fuel came from in about 300 years [1]. The long-lived actinides are what extend the hazard to hundreds of thousands of years, and in a closed fast-reactor cycle they can be returned to the reactor again and again until most are fissioned; only the small fraction lost in each pass through reprocessing, and any not recycled, goes to the repository.

The opposing view deserves to be stated here, where it bears most directly. The US National Academies concluded in 2022 that advanced reactors and their fuel cycles will do little to reduce the need for a geologic repository. They also found that a repository’s long-term safety depends only weakly on how much actinide it contains, unless something breaches the geological barrier [9]. The reason is that long-term dose comes mainly from mobile fission products such as iodine-129 and technetium-99, which recycling does not remove. Both findings should be taken at face value. A closed cycle does not abolish the repository.

It changes three other things, and the mine-to-disposal view is what makes them visible. It changes what goes into the repository: fission products and process losses rather than spent fuel that is still about 94 percent uranium. It changes how much has to be mined, and therefore how many tonnes of rock are milled and how many tailings piles are capped, to fill the repository in the first place. And it turns the largest stock of nuclear material in the country, the depleted uranium, from a waste-management task into a fuel reserve. A fuel cycle that is judged only at the repository gate misses two of those three.

What the Studies Found

None of this is a new idea, and the question of how much it is worth has been modeled carefully. The OECD Nuclear Energy Agency did so in 2013, for a world whose nuclear output grows to about 19,000 terawatt-hours a year by 2100. A fleet of light-water reactors alone mined 19 million tonnes of uranium by 2100 and 62 million by 2200. Adding fast reactors that merely break even on plutonium cut those figures to 15 and 39 million. Adding true breeders, with doubling times of about 18 and 12 years, brought them to 10 or 11 million by 2100, and to between 11 and 14 million by 2200 [10]. Even in the breeder cases, light-water reactors still made up 29 to 42 percent of the fleet in 2100. The IAEA’s GAINS project reached the same shape: fast reactors at about half the fleet by 2100, and cumulative mining down by roughly a third [11].

The pattern in both studies is worth stating plainly, because it explains why no buyer has yet answered the question. The savings arrive slowly, and then decisively. Over the rest of this century they are moderate. Over the next, they are the difference between a fuel cycle that depends on continuous mining and one that largely does not.

France has drawn the conclusion in policy. On 12 March 2026 its Nuclear Policy Council set the aim of no longer importing natural uranium by 2100. It launched a four-year design phase for fast reactors, with a decision on building the first one around 2030, and confirmed the renewal of the La Hague recycling complex [12].

A Rule, Not a Plan

No one can forecast 2100 well enough to fix a target mix of reactors today, and no planner should try. Minimizing mining does not require a forecast. A simple rule, applied each time a reactor is built, shows what the material balance alone would favor at that moment.

The rule is simple. When new capacity is needed, look at the plutonium that is separated and ready, or that can be recovered from spent fuel within the year. If it is enough to start a fast reactor, the ledger favors one. If it is not, it favors a light-water reactor, which adds to the plutonium that will start the next. Plutonium is separated only as fast reactors are ready to receive it, so it is held as a working buffer and never accumulates as a stockpile. The mix is not chosen in advance. It emerges, decision by decision, from the state of the ledger, which is how a market would choose if every material in the chain carried a price.

Our model applies that rule to the world fleet under the Red Book’s high case, 916 gigawatts by 2050 and held there. Built from light-water reactors alone, that fleet mines 10.2 million tonnes of uranium by 2100, more than the 8.1 million tonnes of identified resources the Red Book counts at costs below $260 a kilogram [13]. Under the rule, the savings depend overwhelmingly on a single variable: how quickly spent fuel is reprocessed to recover its plutonium. With reprocessing held at today’s world rate of about 1,500 tonnes a year, cumulative mining falls by 15 to 32 percent by 2100, depending on how well the fast reactors breed; those runs are not in Table 2, which holds the breeding ratio at 1.16. Grow reprocessing to 5,000 tonnes a year by 2050, and the saving is 36 percent by 2100 and 62 percent by 2150. At 15,000 tonnes a year it is 54 and 74 percent.

TABLE 2. Uranium mined each year, world high case (thousand tonnes; author’s model).

Case	2025	2050	2075	2100	2125	2150
Light-water reactors only	64	148	148	148	148	148
Rule, reprocessing held at today’s level	64	129	103	64	35	26
Rule, reprocessing grown to 5,000 t a year by 2050	64	117	85	29	1	0
Rule, reprocessing grown, but fast reactors late and slow	64	143	121	86	60	49

The last row combines three handicaps: first fast reactors in 2045 rather than 2035, start-up inventories of 12 tonnes per gigawatt rather than 7, and an industrial ramp that takes 25 years to reach full pace. Breeding ratio 1.16 in all rule cases. Figures after 2050 hold the high case flat at 916 gigawatts and are extrapolation.

Two things follow from the runs, and they are the heart of the argument.

The first is that the plutonium exists. Spent fuel in storage holds thousands of tonnes of it, far more than the first generation of fast reactors could absorb. The pace of the transition is not set by geology, and it is not set by physics. It is set by how fast the plutonium is recovered, and that depends on whether anyone can earn a return recovering it.

The second is that the breeding ratio, the number that dominates most discussions of fast reactors, matters only while reprocessing is scarce. When recovery is ample, a fast reactor that merely breaks even does nearly as well as a high-performance breeder, because in a mixed fleet the light-water reactors supply most of the seed. That is fortunate, because the measured record of breeding is modest [14]. France’s Phénix measured a breeding ratio of 1.16 against a design value of 1.13, and about 40 percent of its fuel was made from its own recycled plutonium, some assemblies completing the loop three times. Russia’s BN-800 is configured as a plutonium burner: a 2014 calculation puts its breeding ratio at about 0.8 with blankets and 0.6 without [25]. The argument for the mix does not depend on breeders performing at the top of their design range. It depends on the plants that recover their fuel.

The Plants Behind the Reactors

This is where the mine-to-disposal view stops being an abstraction. Every reactor decision is also a decision about the plants upstream and downstream of it, and those plants take a decade to license and build and then run for half a century.

A world that builds only light-water reactors under the high case needs, in our model, conversion capacity for about 148,000 tonnes of uranium a year by 2050, against roughly 62,000 tonnes of licensed capacity operating today [26]. It needs, in our model, about 111 million separative work units of enrichment a year, against about 68 million of world capacity in 2025 [27]. Under the rule, conversion demand peaks near 2050 at about 117,000 tonnes and then declines, to about 78,000 tonnes by 2080 and close to nothing by 2125. A conversion plant built for the 2050 peak would be underused for most of its working life.

The back end moves the other way. Reprocessing of light-water fuel is needed in large volume during the transition, then gives way to the steady reprocessing of fast-reactor fuel in its own loop. The plants that today deconvert depleted uranium hexafluoride for storage, at Paducah and Portsmouth, could become the first step in making fast-reactor fuel. La Hague, which processed 1,221 tonnes of spent fuel in 2025, its highest figure since 1999 [15], is the only plant of commercial scale in the Western world, and France is now planning its successor [12].

Every plant built for the front end is, in effect, an assumption that the transition will not come soon. That assumption may often be right. It should be made by the people who carry its risk, at prices that include the material it leaves behind.

The American Case

The United States is the sharpest illustration, because it holds every material the transition needs and lacks the one plant that would release them.

Its spent fuel contained about 824 tonnes of plutonium at the end of 2018, according to an analysis by Robert Alvarez [16]. With the discharges since then, our estimate is about 980 tonnes today. At 7 to 12 tonnes of plutonium per gigawatt (total plutonium; 7 is the base case, 12 the cautious case), that is enough to start between 80 and 140 gigawatts of fast reactors, from most of the current American fleet to about one and a half times it. Its depleted uranium, about 490,000 tonnes of uranium, would feed such a fleet far longer than any plant would operate; our model’s fast fleets use less than 30,000 tonnes of it by 2150. What the United States does not have is a commercial reprocessing plant. Without one, no buyer can choose a fast reactor started on recycled plutonium, and the saving is zero. Fast reactors fueled with high-assay low-enriched uranium, such as TerraPower’s Natrium, which received an NRC construction permit in March 2026, are outside this argument and our model [40].

Add a single reprocessing plant of 800 tonnes a year from 2040, the size of one of La Hague’s two plants, and our model of today’s 97-gigawatt fleet mines 45 percent less uranium by 2100 and 67 percent less by 2150.

The national target makes the stakes plain. An executive order of May 2025 aims to expand American nuclear capacity from about 100 gigawatts to 400 gigawatts by 2050 [17]. Built from light-water reactors, that fleet mines about 4.3 million tonnes of uranium by 2100 in our model, more than half of the world’s identified resources below $260 a kilogram. Each gigawatt of light-water reactors built commits about 10,000 tonnes of uranium over a sixty-year life, along with the conversion, enrichment and disposal capacity to match. The reactors ordered in the next fifteen years will therefore set a floor on American uranium demand that lasts until about 2100.

Federal spending so far has followed the once-through chain. In January 2026 the Department of Energy awarded $2.7 billion for enrichment [18], and in September 2025 up to $187.5 million to expand conversion [19]. Both are needed for the fleet that exists. There has been no comparable federal commitment to reprocessing: no federally funded reprocessing plant or program, although an executive order, a proposed NRC rule and a DOE committee now address it [20] [39].

The reason is not mysterious. Almost every reprocessing plant and fast reactor in the world has been built or backed by a state [23], and at today’s uranium prices recycling does not pay [22], so no private buyer has carried the capital risk. The materials that would feed such plants are barely on the market. The Department of Energy owns the depleted uranium at Paducah and Portsmouth. It has reserved about 44 percent of the cylinders for other users, including a private buyer, but the Government Accountability Office found in 2022 that its authority to sell is doubtful under the 1996 law that governs its uranium [29] [30]. Spent fuel stays at the reactors, but under the Standard Contract the Department agreed to take title to it starting 31 January 1998, and it has not [31]. The government pays utilities damages for that instead, $11.1 billion from the Treasury’s Judgment Fund through fiscal 2024 [32].

Disposal is a federal liability. Utilities paid for it through a fee of 1.0 mill a kilowatt-hour, a tenth of a cent. In 2013 a federal court ordered the Department to propose cutting the fee to zero [38], and it has been zero since 16 May 2014 [28]. The fee was flat per kilowatt-hour, so it never told a buyer anything about one fuel cycle against another. The cost of throwing material away no longer reaches anyone who decides whether to throw it away. In company-reported results, a DOE-funded study by Argonne, Oklo and Deep Isolation found that deep-borehole disposal of waste from recycled fuel came in under ARPA-E’s target of 0.1 cent per kWh — the same 1 mill/kWh utilities paid into the Nuclear Waste Fund until 2014 — though the cost figures have not yet been published [37].

None of this requires new custodians. Title and custody already come apart in the uranium market. The Sprott Physical Uranium Trust owned about 37,000 tonnes of U₃O₈ on 30 September 2026, held for it at licensed plants run by Cameco, ConverDyn and Orano [33], and the Department of Energy itself took title to depleted uranium cylinders in 2002 while their custody stayed with the enricher [30]. Spent fuel already sits with NRC-licensed holders under 10 CFR Part 72 [31]. Making it tradable would take a change to the Standard Contract and the Nuclear Waste Policy Act, and a clear answer on who carries the disposal liability once title moves. Whoever takes title should pre-fund that liability at transfer, in a trust or escrow, so that no owner can walk away from it.

At least one company is betting it can close the cycle privately. In September 2025 Oklo announced a privately funded plant at Oak Ridge, Tennessee, to recycle used light-water fuel by electrorefining into metal fuel for its own fast reactors, and in March 2026 it finished pre-application talks with the NRC [34]. The NRC ended its reprocessing rulemaking in 2021, judging existing rules adequate, and in June 2026 proposed an explicit licensing route under Part 70 [35] [39]. Oklo’s own annual report warns that a market for recycled fuel “may never be established” [36].

My recommendation follows from the ledger, but it is not a plan. Let the market make the call, and stop distorting the prices it would use. Treat spent fuel and depleted uranium as resources that can be owned and sold, by clear title or by auction, so that their value shows up in the price of the next reactor. License reprocessing and fast-reactor fuel plants on predictable timelines, because their lead time is the longest in the chain and regulatory uncertainty is a cost no investor can price. And charge for disposal by what actually goes into the repository, its heat, volume and long-lived actinides, so that discarding material is never free. When the economics turn, the buyers will close the cycle themselves.

The Case Against

The objections to this argument are serious, and they deserve to be stated in their strongest form.

The first is abundance. MIT’s study of the nuclear fuel cycle concluded in 2010 that uranium supplies will not limit the expansion of nuclear power for the foreseeable future, and it recommended storing spent fuel for about a century to keep the choice open [21]. The second is cost. Bunn and colleagues at Harvard calculated in 2003 that recycling plutonium in existing reactors does not pay until uranium exceeds about $360 a kilogram in 2003 dollars [22]. The third is the record. The International Panel on Fissile Materials counted more than $50 billion spent on breeder programs worldwide without producing a reactor economically competitive with light-water plants. Its account includes Monju’s sodium fire in 1995, and 27 sodium leaks at BN-600 between 1980 and 1997, fourteen of which caught fire [23]. The fourth is proliferation. The National Academies concluded in 2022 that reprocessing poses greater proliferation and terrorism risks than the once-through cycle [9].

None of these is answered by enthusiasm. Each is answered, if it is answered at all, by the ledger.

The abundance argument rests on identified resources that a light-water-only expansion would exhaust within this century in the high case, and on a recommendation to wait that predates both today’s growth targets and the Red Book’s finding that existing and committed mines meet even its low case only through 2032 [13]. The cost argument compares prices that leave out the value of the residues and any credit for a smaller repository, for a decision whose consequences run for sixty years. At today’s prices the once-through cycle is cheaper, and this piece does not claim otherwise. The case rests on correcting those prices, and on value a price set today does not capture: uranium for the next century, and a fuel supply that depends less on mines abroad. The ledger shows that waiting has a price of its own, paid in tonnes of uranium committed by every reactor built in the meantime. The record is real, and it should temper every promise made about the next generation of fast reactors. Phénix nonetheless showed that the loop can be closed in practice. The proliferation concern is the strongest of the four, and it is the one place where government must set hard limits, not prices. Licenses should allow plutonium to be separated only as fast reactors are ready to receive it, held as a working buffer rather than a stockpile, and never accumulated in the quantities the past forty years have produced. Each such plant would need full IAEA safeguards, with its plutonium counted continuously and inspected, and a regulator, not the owner, would decide when a reactor is ready to receive it.

What This Means

Return, finally, to the drum at the mine gate. In the once-through cycle its uranium passes through the chain once. About half of one percent of it is fissioned, and the rest is managed for longer than any human institution has lasted. Every plant in the chain does its job well. The chain as a whole does not, because no one pays for it as a whole.

The fuel cycle the primer describes in Chapter 27 is not reached by a single decision, and it is not reached by waiting for uranium to become expensive. It is reached by a long sequence of ordinary decisions, made by many buyers, each looking at a price: which reactor to build next, which plant to build to serve it, and what each will leave behind. The material balance has been settled for decades. The materials are already above ground. What remains is to let the price see the whole chain, from the mine to the repository, because the chain is one system whether anyone prices it or not.

This is an opinion piece. It sets out the author’s argument on a question the evidence leaves open. The facts and figures in it are cited and held to the same standard as the rest of the site; the conclusions are the author’s own. The site does not endorse any company, product or investment. The author may, from time to time, hold positions in publicly traded nuclear companies, including companies named here.

Spot an error? Suggest a correction

Notes

1. J. Hill, The Nuclear Fuel Cycle: A Complete Primer, Chapters 4, 6, 16 and 27, and Appendix XII.

2. R. J. Goldston, “Climate Change, Nuclear Power, and Nuclear Proliferation: Magnitude Matters,” Science & Global Security 19 (2011); cited for plutonium and transuranics per gigawatt-year at 50 MWd/kg. https://scienceandglobalsecurity.org/archive/sgs19goldston.pdf

3. US Department of Energy, Office of Environmental Management, Portsmouth/Paducah Project Office, “Oxide Shipping” (more than 730,000 metric tons of DUF₆; close to 70,000 cylinders remain), accessed 21 October 2026. https://www.energy.gov/pppo/oxide-shipping

4. US Department of Energy, Office of Inspector General, Audit Report DOE-OIG-23-04, November 2022; conversion could take until 2074 at the 2019 rate. https://www.energy.gov/sites/default/files/2022-11/DOE-OIG-23-04.pdf

5. American Nuclear Society, Nuclear Newswire, “IAEA releases interactive tool on global spent nuclear fuel,” 25 June 2026; reports the IAEA totals (447,758 tHM discharged; about 322,000 stored; United States 94,926 tHM). https://www.ans.org/news/article-8149/iaea-releases-interactive-tool-on-global-spent-nuclear-fuel/

6. World Nuclear Association, “Russia’s Nuclear Fuel Cycle,” Information Library, updated 30 September 2026. https://world-nuclear.org/information-library/country-profiles/countries-o-s/russia-nuclear-fuel-cycle

7. C. E. Till and Y. I. Chang, “Some Comments on the Impact of Thermal Near Breeders on Uranium Utilization,” NEACRP-A-1976-0273, OECD Nuclear Energy Agency, 1976 (LMFBR assumed initial core inventory 3,000 kg fissile per GWe, external inventory 1,600 kg per GWe, compound doubling time 15 years). https://www.oecd-nea.org/upload/docs/application/pdf/2020-07/neacrp-a-1976-0273.pdf

8. X. Doligez, S. Bouneau, S. David et al., “Fundamentals of reactor physics with a view to the (possible) futures of nuclear energy,” Comptes Rendus Physique 18 (2017) 372–380. https://doi.org/10.1016/j.crhy.2017.10.004

9. National Academies of Sciences, Engineering, and Medicine, Merits and Viability of Different Nuclear Fuel Cycles and Technology Options and the Waste Aspects of Advanced Nuclear Reactors, 2022 (published 2023). https://doi.org/10.17226/26500

10. OECD Nuclear Energy Agency, Transition Towards a Sustainable Nuclear Fuel Cycle, NEA No. 7133, 2013. https://www.oecd-nea.org/upload/docs/application/pdf/2019-12/7133-transition-sustainable-fuel-cycle.pdf

11. IAEA, Framework for Assessing Dynamic Nuclear Energy Systems for Sustainability: Final Report of the INPRO Collaborative Project GAINS, Nuclear Energy Series NP-T-1.14, 2013. https://www-pub.iaea.org/MTCD/Publications/PDF/Pub1598_web.pdf

12. Présidence de la République, “Cinquième conseil de politique nucléaire,” 12 March 2026. https://www.elysee.fr/emmanuel-macron/2026/03/12/cinquieme-conseil-de-politique-nucleaire

13. OECD Nuclear Energy Agency and IAEA, Uranium 2026: Resources, Production and Demand (the Red Book), September 2026. https://oecd-nea.org/upload/docs/application/pdf/2026-09/uranium_2026_-_resources_production_and_demand.pdf

14. Generation IV International Forum, “Phénix and Superphénix Feedback Experience”; measured breeding ratio 1.16 against a design value of 1.13. https://www.gen-4.org/sites/default/files/2024-09/Phenix_Superphenix_final.pdf

15. Orano, “Rapport d’information du site Orano la Hague,” 2025 edition; 1,221 tonnes processed in 2025. https://cdn.orano.group/orano/docs/default-source/orano-doc/groupe/publications-reference/rapport-d-information-du-site-orano-la-hague-2025.pdf

16. R. Alvarez, “The long-term problem of ‘peaceful’ plutonium,” Bulletin of the Atomic Scientists, 8 March 2021. https://thebulletin.org/2021/03/the-long-term-problem-of-peaceful-plutonium/

17. The White House, Executive Order 14300, “Ordering the Reform of the Nuclear Regulatory Commission,” 23 May 2025 (400 GW by 2050). https://www.whitehouse.gov/presidential-actions/2025/05/ordering-the-reform-of-the-nuclear-regulatory-commission/

18. US Department of Energy, “U.S. Department of Energy Awards $2.7 Billion to Restore American Uranium Enrichment,” 5 January 2026. https://www.energy.gov/articles/us-department-energy-awards-27-billion-restore-american-uranium-enrichment

19. US Government Accountability Office, GAO-26-107385, “Nuclear Fuel: Actions Needed to Enhance Cost Reporting and Economic Analysis for Federal Uranium Supply Efforts”; reports the September 2025 agreement of up to $187.5 million with Solstice for Metropolis. https://www.gao.gov/assets/gao-26-107385.pdf

20. The Nuclear Fuel Cycle (thenuclearfuelcycle.com), “Policy Changes, July–October 2026” (forthcoming); cited for the DOE fuel-cycle committee’s recycling and reprocessing subcommittee (28 July 2026).

21. Massachusetts Institute of Technology, The Future of the Nuclear Fuel Cycle, MIT Energy Initiative, 2010 (2011 final report). https://energy.mit.edu/wp-content/uploads/2011/04/MITEI-The-Future-of-the-Nuclear-Fuel-Cycle.pdf

22. M. Bunn, S. Fetter, J. P. Holdren and B. van der Zwaan, The Economics of Reprocessing vs. Direct Disposal of Spent Nuclear Fuel, Harvard University, December 2003. https://www.belfercenter.org/publication/economics-reprocessing-vs-direct-disposal-spent-nuclear-fuel

23. International Panel on Fissile Materials, Fast Breeder Reactor Programs: History and Status, Research Report 8, February 2010. https://fissilematerials.org/library/rr08.pdf

24. International Panel on Fissile Materials, “2024 civilian plutonium and HEU declarations submitted to IAEA,” March 2026. https://fissilematerials.org/blog/2026/03/2024_civilian_plutonium_a.html

25. M. Kütt, F. Frieß and M. Englert, “Plutonium Disposition in the BN-800 Fast Reactor: An Assessment of Plutonium Isotopics and Breeding,” Science & Global Security 22 (2014) 188–208. https://doi.org/10.1080/08929882.2014.952578

26. World Nuclear Association, “Conversion and Deconversion,” Information Library, updated 14 October 2026 (world primary conversion licensed capacity 62,000 tU in 2024), accessed 21 October 2026. https://world-nuclear.org/information-library/nuclear-fuel-cycle/conversion-enrichment-and-fabrication/conversion-and-deconversion

27. World Nuclear Association, “Uranium Enrichment,” Information Library, updated 29 September 2026 (world enrichment capacity 68,076 thousand SWU in 2025), accessed 21 October 2026. https://world-nuclear.org/information-library/nuclear-fuel-cycle/conversion-enrichment-and-fabrication/uranium-enrichment

28. US Energy Information Administration, “Agency Information Collection Extension” (Form NWPA-830G), Federal Register 80 FR 36537, 25 June 2015, doc. 2015-15628: fee reduced from 1.0 mill/kWh to 0.0 effective 16 May 2014 after NARUC v. DOE (D.C. Cir., 19 November 2013); see also DOE, Nuclear Waste Fund FY2015 and FY2014 Financial Statements, OAI-FS-16-03. https://www.federalregister.gov/documents/2015/06/25/2015-15628/agency-information-collection-extension

29. US Government Accountability Office, Nuclear Waste Cleanup: DOE’s Efforts to Manage Depleted Uranium Would Benefit from Clearer Legal Authorities, GAO-22-105471, June 2022. https://www.gao.gov/products/gao-22-105471

30. US Department of Energy, Final Environmental Impact Statement for Construction and Operation of a Depleted Uranium Hexafluoride Conversion Facility at the Paducah, Kentucky, Site, DOE/EIS-0359, 2004 (DOE responsibility for the DUF₆ inventory; 2002 transfer of title from USEC with custody retained by USEC). https://www.energy.gov/sites/prod/files/EIS-0359-FEIS-01-2004.pdf

31. 10 CFR Part 961, Standard Contract for Disposal of Spent Nuclear Fuel and/or High-Level Radioactive Waste (DOE services to begin not later than 31 January 1998); storage by licensees under 10 CFR Part 72. https://www.ecfr.gov/current/title-10/chapter-III/part-961

32. US Department of Energy, Office of Nuclear Energy, “Spent Nuclear Fuel and High-Level Waste Disposition Program,” presentation to the Wyoming Legislature, 2025 (Judgment Fund payments $11.1 billion as of FY2024, citing DOE Nuclear Waste Fund audit reports). https://wyoleg.gov/InterimCommittee/2025/09-2025072914-02USDOEPresentation.pdf

33. Sprott Asset Management, Sprott Physical Uranium Trust fact sheet, 30 September 2026 (81,697,348 lb U₃O₈; storage at Cameco, ConverDyn and Orano); Annual Report 2025, note 1 (custodians). https://sprott.com/media/qztbbd4d/uranium-trust-factsheet.pdf

34. Oklo Inc., “Oklo Announces Fuel Recycling Facility as First Phase of up to $1.68 Billion Advanced Fuel Center in Tennessee,” 4 September 2025; Oklo, NRC pre-application engagement completed, 19 March 2026. https://www.oklo.com/newsroom/news-details/2025/Oklo-Announces-Fuel-Recycling-Facility-as-First-Phase-of-up-to-1-68-Billion-Advanced-Fuel-Center-in-Tennessee/default.aspx

35. US Nuclear Regulatory Commission, “Reprocessing” (SECY-21-0026; rulemaking discontinued 2021; applications would be reviewed under existing regulations). https://www.nrc.gov/materials/reprocessing

36. Oklo Inc., Annual Report on Form 10-K for fiscal year 2024, Risk Factors (“The market for recycled nuclear fuel in the United States may never be established”). https://s203.q4cdn.com/103172959/files/doc_financials/2024/q4/8bed5a51-dd20-49a3-a549-ebab52f03125.pdf

37. Deep Isolation, “Nuclear Waste Innovator Deep Isolation Demonstrates Safe, Cost-Effective Disposal Pathway for Waste from Recycled Fuel,” press release, 22 September 2026, https://www.deepisolation.com/press/nuclear-waste-innovator-deep-isolation-demonstrates-safe-cost-effective-disposal-pathway-for-waste-from-recycled-fuel/; ARPA-E, CURIE funding opportunity announcement DE-FOA-0002691, 15 March 2022, program metric 2 (disposal cost about 0.1 ¢/kWh). Results are company-reported and not peer reviewed; doses were compared against a 10 mrem/yr benchmark chosen for the study, not a US licensing standard.

38. National Association of Regulatory Utility Commissioners v. US Department of Energy, 736 F.3d 517 (D.C. Cir. 2013).

39. US Nuclear Regulatory Commission, “Modernizing Materials Licensing,” proposed rule, 91 FR 38124, 24 June 2026, FR Doc. 2026-12702 (explicit Part 70 route for spent fuel reprocessing facilities; comments due 10 August 2026). https://www.federalregister.gov/documents/2026/06/24/2026-12702/modernizing-materials-licensing

40. US Nuclear Regulatory Commission, “US SFR Owner, LLC; Kemmerer Power Station, Unit 1; Construction Permit and Record of Decision,” Federal Register, 16 March 2026, FR Doc. 2026-05067 (permit CPAR-1 issued 9 March 2026). https://www.federalregister.gov/documents/2026/03/16/2026-05067/us-sfr-owner-llc-kemmerer-power-station-unit-1-construction-permit-and-record-of-decision