Nuclear Waste Solutions

Nuclear waste solutions means the permanent disposition of the material that comes out of a reactor and cannot be used again, and the first thing to get right is that this is not one material. Spent nuclear fuel is intact irradiated fuel assemblies discharged from a power reactor, still containing uranium, plutonium and fission products; it is the dominant civil inventory. High-level waste proper is the vitrified residue left after spent fuel has been reprocessed — chemically different, smaller in volume, and legally distinct in most jurisdictions. Long-lived intermediate-level waste is hulls, hardware, sludges and resins: not thermally hot, but radiologically long-lived enough that surface disposal is not an option. Transuranic waste is a mostly American category covering plutonium-contaminated material from weapons production. And low-level waste is disposed near the surface almost everywhere and is not the subject of this brief. Conflating these is the classic error in writing on the topic, and it is what makes the record look either better or worse than it is: the world's one operating deep repository takes only defence transuranic waste, and the world's most expensive retrieval operation is recovering low- and intermediate-level drums.

Established The inventory that has to go somewhere is roughly 448,000 tonnes of heavy metal. That is the figure in the second edition of the IAEA's Global Spent Nuclear Fuel Inventory, drawn from the 2025 Joint Convention reporting cycle and released on 22 June 2026: about 448,000 tHM discharged from power reactors worldwide, of which roughly 75% sits in storage and roughly 25% — about 126,000 tHM — has been reprocessed. Of the stored fraction, about 41% is in pools and about 31% in dry casks. The United States alone holds something like 95,000 tonnes as of 2026 and adds about 2,000 tonnes a year.

The technical answer the field converged on four decades ago is a deep geological repository: several hundred metres down in a rock formation stable over geological time, with the waste inside engineered barriers designed so that no future generation has to do anything. The alternatives that appear in the discourse — partitioning and transmutation, deep boreholes, indefinite interim storage — are treated in this brief not as rival solutions but as claims about how much of the repository problem can be avoided, and the honest answer in every case is: less than advertised.

The slot offers two ready-made framings and neither survives the numbers. One says nuclear waste is an unsolved civilisational problem. The other says it is technically solved and politically blocked. The evidence supports a third reading — technically solved, institutionally slow — but that reading needs a correction before it is honest, and the correction is where the interesting work is. It is set out in the position section below.

Scope, because two sibling briefs touch this one. Advanced fission owns the reactor and the fuel cycle — what is burned, at what enrichment, in what spectrum, and therefore what comes out. This brief owns what happens to the material after discharge and does not re-argue fuel cycles, except where a fuel-cycle claim is made about disposal, which is why partitioning and transmutation appear here at all. Small modular reactors matter here for one reason: they multiply the number of sites at which waste is generated and stored, which changes the transport and interim-storage problem without changing the disposal problem at the far end. Neither sibling brief has a disposal route that exists today in more than one country, and stating that plainly is part of this brief's job.

Established Start with the base rate, because everything else in this subject is an argument about why it is what it is. The number of countries operating a deep geological repository for high-level waste or spent fuel is zero. Not one, anywhere, after four decades of national programmes. By the end of 2027 it will be one. The number operating a deep geological repository for any long-lived radioactive waste is one — the United States, at the Waste Isolation Pilot Plant, and only for defence transuranic waste, which is neither spent fuel nor high-level waste. The Finnish Government's own framing of what Posiva is about to do is that it would be the “first in the world,” and it is right.

Established Finland is the success case and it took 43 years. National screening of Finnish territory began in 1983. Four candidate sites were investigated from 1993 to 2000. Eurajoki's municipal council voted to host on 24 January 2000, by 20 votes to 7. The Government issued its Decision-in-Principle on 21 December 2000 and Parliament ratified it on 18 May 2001 by 159 votes to 3. Excavation of ONKALO began in 2004; the construction licence was granted on 12 November 2015; the operating licence application was filed on 30 December 2021; the full-system trial run, five dummy canisters emplaced at 430 m, ran from 30 August 2024 to 4 March 2025; and on 4 August 2026 the Radiation and Nuclear Safety Authority issued a favourable statement to the Ministry of Economic Affairs and Employment finding that the Nuclear Energy Act's conditions for an operating licence are fulfilled.

Established It is not licensed and it is not operating, and this is the single most misreported fact in the subject. The favourable statement is a regulatory opinion, not a licence. The Government grants the licence, and the Ministry said only that a decision proposal “could be prepared and submitted to the Government in the autumn.” Even after a grant, Posiva may not begin operating before a further inspection under §20 of the Nuclear Energy Act. The licence is sought to the end of 2070, with disposal continuing into the late 2120s. Anyone writing that Onkalo is licensed, or operating, is wrong today, and the misconceptions section names the widely-syndicated source that got it wrong.

Established And here is the number that reframes the whole debate: of those 43 years, four years and seven months was regulatory review of a finished, trial-run-complete facility. The application was filed on 30 December 2021 and the favourable statement came on 4 August 2026, with the regulator's own deadline extended repeatedly, including a missed deadline in June 2026. This happened in a jurisdiction with unopposed municipal consent, no litigation, no organised opposition, a fully funded liability, and a completed hole in the ground with dummy canisters already emplaced and retrieved. Strip out every political obstacle the standard narrative blames and half a decade of review remains. The slowness is not primarily political. It is institutional, and it persists after the politics is solved.

Established Four major nuclear nations have no site at all. The United States is 44 years from the Nuclear Waste Policy Act of 1982 with zero high-level waste emplaced; the statute as amended in 1987 designates Yucca Mountain as the only site that may be considered, so no alternative can be pursued without Congress amending the Act, and the Yucca adjudication itself remains suspended. Japan's implementing organisation began work in 2002 and is 24 years in, at stage one of three. Germany spent 43 years and €1.9 billion on Gorleben and formally excluded it in 2020. The United Kingdom has two candidate communities left, a date that has slipped from 2040 to the late 2050s, a whole-life estimate of £20 billion to £54 billion, and an official delivery-confidence rating of red: successful delivery of the project appears to be unachievable.

Established Where siting has worked it has worked slowly and by consent. Canada selected a site on 28 November 2024 — Ignace, jointly with the Wabigoon Lake Ojibway Nation — after a 14-year process that began in 2010 with 22 interested communities, with regulatory approval expected around 2032, operation around 2040–2045, and a stated project life of 175 years. Sweden's government approved Forsmark on 27 January 2022, the Land and Environment Court issued the environmental permit in October 2024, and construction began in January 2025 — more than 6,000 canisters and about 60 km of tunnels, with emplacement expected around 2035 and full development into the 2080s. France has a favourable safety opinion from its regulator dated 4 December 2025 and a public inquiry in the second half of 2026, with operations around 2040–2050. Switzerland filed its general licence application on 19 November 2024, expects a Federal Council decision around 2029–30, may face a national referendum around 2031, and does not expect emplacement before 2050.

Established The cost asymmetry is the sharpest set of numbers in the brief, and it runs the opposite way to intuition. Finland's National Nuclear Waste Management Fund held €3.0 billion at the end of 2025, and the Finnish Government states that this covers the costs of the remaining waste management of all nuclear waste generated in Finland. That is a government statement about its own liability, not a vendor claim. Against it: Germany's Asse II retrieval alone is estimated at €4.7 billion ±30% — roughly one and a half times Finland's entire national liability — to dig 126,000 drums of low- and intermediate-level waste back out of a mine that was never a licensed repository. Gorleben cost €1.9 billion for 43 years and zero emplacement. Yucca Mountain cost about $15 billion through 2009 for zero emplacement. Sellafield's cleanup is estimated at £136 billion undiscounted in 2023-24 prices, about 45 times Finland's total, and it rose 18.8% in real terms in five years. Finland's entire solution costs roughly 2% of Sellafield's bill and less than Germany's single retrieval mistake. The multiples are this brief's arithmetic; the inputs are sourced.

Established Now the correction that makes the “technically solved” reading honest: it mislocates the residual risk. The risk is not in the geology. No repository failure anywhere in this evidence base has been geological. Every actual failure was operational or administrative. At WIPP, a drum breached underground on 14 February 2014 because a specification line reading “kitty litter/zeolite clay” became “kitty litter (clay)” in internal communications and an organic wheat-based absorbent was substituted for an inorganic one. At Asse II, 126,000 drums were emplaced between 1967 and 1978 into an unstable, unlicensed former salt mine that is now taking on brine. At San Onofre on 3 August 2018, a loaded canister hung up on a divider shell while the rigging went slack and nobody noticed until a radiation protection technician's readings came back unexpectedly high; had it slipped it would have been an 18-foot load drop. At Gorleben, 43 years of exploration ended at a site rated unfavourable on all three indicators of the overburden criterion.

Frontier There is one genuine open technical question and the “solved” framing papers over it: the copper canister. Sweden's Land and Environment Court declined in January 2018 to endorse SKB's application until it demonstrated that degrading processes affecting the copper canisters are not a problem for long-term safety. The Government approved the licence anyway in January 2022. Construction began in January 2025. The dispute was closed administratively rather than empirically, and the regulator's substantive defence is not that the critics are wrong but that it does not matter: SSM presented hypothetical calculations showing that even if all roughly 6,000 canisters failed immediately, doses would remain near background, because the bentonite buffer and the host rock carry the safety case. That is a sound multi-barrier argument. It is also a quiet concession that the engineered copper barrier is not what the licence rests on, which is not how KBS-3 was originally sold.

Established The finding, in one sentence. Nuclear waste disposal is a solved geological problem, an unresolved materials-science problem at the canister scale that regulators have decided does not matter, a repeatedly demonstrated operational failure mode, and above all an institutional process so slow that it takes four decades in the best case, in the friendliest jurisdiction, with the money already banked and nobody objecting. The slowness is not the price of solving the problem; it is what the institutions do instead of solving it — and the American case proves it, because the technical review passed and the programme died anyway. The Nuclear Regulatory Commission's staff completed the Safety Evaluation Report in January 2015, finding that the Department of Energy's application met the regulatory requirements with two exceptions, both legal rather than technical: the Department had not obtained land withdrawal or water rights. The Government Accountability Office's verdict on the termination is verbatim and unhedged: “DOE's decision to terminate the Yucca Mountain repository program was made for policy reasons, not technical or safety reasons.”

The open questions in this subject are unusually easy to separate from the ones that merely sound open, because the field has been running long enough that most of the sounding-open ones have answers nobody likes. What follows is what is genuinely unresolved.

Frontier Copper corrosion in anoxic water is the one live materials-science dispute, and it is specific enough to state precisely. The question is whether copper corrodes in oxygen-free water at rates that matter over a hundred-thousand-year design life. Two pathways are proposed. Cuprite, Cu₂O, forms only under conditions thermodynamically limited to about 10⁻⁹ bar of hydrogen — which is to say, essentially never in a repository. A CuOH-type phase, if it exists, would imply a limit nearer 10⁻³ bar and therefore corrosion continuing indefinitely in anoxic groundwater. The CuOH phase has never been isolated and identified despite repeated attempts. That is the strongest single fact in the regulator's favour and it is not a proof of absence.

Frontier Three further copper questions are open and are not resolved by the CuOH argument. Localised sulphide corrosion — pitting and stress-corrosion cracking rather than uniform thinning — is disputed; measured groundwater sulphide concentrations of 10⁻⁷ to 10⁻⁵ M plus slow diffusion through the bentonite buffer are the regulator's mass-transfer argument against it. Radiation-induced degradation of copper is disputed. So is strain hardening of the cast-iron insert. The critics of record are Peter Szakálos and Christofer Leygraf of KTH Royal Institute of Technology, whose 2007 hydrogen-evolution experiments underlie the case, and who hold that canister lifetimes are several orders of magnitude shorter than SKB projects. This brief did not obtain the 2007 paper directly and knows it only through the descriptions given by the regulator and by an opposed non-governmental organisation, which is stated here rather than glossed.

Frontier The LOT experiment is a live empirical disagreement about a specific dataset, which is rarer than it sounds. LOT was a twenty-year copper-and-clay test at the Äspö Hard Rock Laboratory. Two packages were retrieved in autumn 2019 and SKB reported on them in June 2020. The opposed NGO reads the results as showing corrosion under anoxic conditions inconsistent with SKB's model, and criticises the reporting for missing metallographic cross-sections of the most corroded areas, low-resolution imagery, non-traceable sample identification and possible selection of images. SKB attributes the observed corrosion to oxygen trapped inside the experimental packages. Both readings are of the same metal. This brief does not resolve it, and notes that it has no unmediated statement of the applicant's case: SKB's own technical reports were not obtained, so the applicant's position here is reconstructed from the regulator's account and from a hostile summary. That is a real asymmetry and it is recorded rather than hidden.

Frontier Chloride-induced stress corrosion cracking of dry-storage canisters is open, and the honest statement is not the one usually made. The independent United States Nuclear Waste Technical Review Board states that “no evidence of CISCC in SNF canisters has been found, although very few canister inspections have been conducted” — and, critically, “none with rigorous inspection criteria and mature inspection techniques.” Two horizontally stored welded canisters at Calvert Cliffs, inspected in 2012 after more than fifteen years, showed no signs of gross degradation. Three conditions must coexist for the mechanism to run: susceptible 304 or 316 stainless steel, tensile stress near welds, and chloride salt deliquescence in the right temperature and humidity band. The correct summary is therefore not “no degradation has occurred” but “almost nobody has looked properly.”

Handwave Any claim that partitioning and transmutation removes the need for a repository. The United States National Academies ruled on this in 2023 and the language is flat: implementing a fully closed fuel cycle that includes reprocessing “does not eliminate the need for a geologic repository because fission products will still require disposal, material losses will inevitably occur in reprocessing and in other parts of the fuel cycle.” Even in the limit of 100% partitioning-and-transmutation efficiency — which the committee calls “a highly unlikely situation because processes are not perfect” — the remaining high-level waste would still be radioactive fission products requiring disposal. The committee found advanced fuel cycles with P&T costly and challenging to implement, requiring operation for many decades to achieve permanent benefits, a commitment to nuclear power for several centuries, and significant financial investment, and it recommended that the United States continue the once-through cycle as policy, with any alternative required to demonstrate advantages over that baseline.

Frontier What remains open about transmutation is narrower and more interesting than the claim it is usually made to support. Europe's flagship accelerator-driven system, MYRRHA in Belgium, broke ground on its first phase on 25 June 2024. That phase, MINERVA, is a 300 m accelerator delivering 100 MeV protons and operating at 2 K, and its two target stations are for medical and research radioisotope production and for fusion-materials research. The implementer's own page for Phase 1 does not describe minor-actinide transmutation at all. Transmutation demonstration requires Phase 3, the subcritical lead-bismuth-cooled fast reactor. Full operation was targeted for 2034 when Belgium's Council of Ministers approved €558 million in 2018; as of the 2024 groundbreaking it is 2038 — a four-year slip in six years. So the sceptical synthesis is this: transmutation will not be demonstrated in Europe before about 2038, by which time Onkalo on current schedule will have been emplacing spent fuel for roughly eleven years and Sweden will be about three years in. Transmutation is not on the critical path to disposal. Disposal is on the critical path to transmutation.

Frontier Deep borehole disposal is a modelling result, not a disposal result. The most recent announcement in this space, dated 26 March 2026 and described as validating borehole disposal for waste from an electrorefining process, was a computer-based study with no physical demonstration and no waste emplaced. It modelled generic shale and granitic host rocks and reported doses several orders of magnitude below a stringent radiological exposure standard, under a federal research programme with two national laboratories. The vendor's own qualifier is the tell: the technology could offer an option should federal law be amended to authorise the disposal of high-level waste in a borehole repository. In the United States that means amending the Nuclear Waste Policy Act — the same statutory lock that immobilises Yucca Mountain. There is no licensing framework, no application and no site. The same vendor has previously reported emplacing and retrieving a test canister in a borehole; this brief did not verify that and does not rely on it.

Frontier The genuinely unresolved question about interim storage is what “indefinite” means, and this brief cannot answer it. What can be stated from the regulator: dry casks are licensed or certified for up to 40 years with possible renewals of up to 40 years; “NRC regulations do not specify a maximum time for storing spent fuel in pool or cask”; the waste-confidence position holds that storage is safe for at least 60 years beyond a reactor's licensed life, which at current licensing terms amounts to at least 120 years; and ageing management programmes are required. What cannot be stated is the content of the Continued Storage Rule of 2014, which replaced the vacated Waste Confidence Decision and which analyses short-term, long-term and indefinite storage timeframes. The Federal Register notice returned a server error and was not read. That is the most material gap in the evidence behind this brief, because it is precisely where the regulator's assumptions about institutional control over a century or more are written down. This brief therefore does not assert what the NRC assumes about indefinite storage, and a reader should treat any secondary summary of it, including sympathetic ones, as unverified here.

Established The first bottleneck is institutional throughput, and Finland is the proof because Finland has removed every other obstacle. Consent, funding, geology, a completed facility, a completed trial: and still four years and seven months of review, with the regulator's deadline extended more than once, including a missed deadline in June 2026. This is the floor. Every other national programme adds litigation, siting failure or statutory deadlock on top of that floor. Any estimate of how long a repository takes that does not include half a decade of review on a finished facility is not an estimate of this process.

Established The second is statutory lock-in, and the American case is the cleanest example of a bottleneck that no amount of technical work can clear. The Nuclear Waste Policy Act as amended in 1987 designates Yucca Mountain as the only site that may be considered for permanent disposal of commercial spent fuel. No alternative site can be pursued without Congress amending the Act, and the Act has not been amended. The licensing history is a study in institutional stalemate: application 3 June 2008, docketed 8 September 2008, the Department of Energy moved to withdraw on 3 March 2010, the Atomic Safety and Licensing Board denied the withdrawal on 29 June 2010, the Commission deadlocked two-two on appeal, Congress zeroed review funding from FY2012, the Board suspended the adjudication on 30 September 2011, a federal appeals court ordered the Commission in August 2013 to resume review using existing funds, staff completed the Safety Evaluation Report in January 2015 finding the application met the requirements, an environmental supplement followed in early 2016 — and the adjudicatory proceeding remains suspended.

Established The third is the funding architecture, which in the United States is a bottleneck constructed out of money. The Nuclear Waste Fund held $51.4 billion at the end of FY2025, having collected $24.6 billion in utility contributions between 1983 and 2025 and grown the rest on Treasury interest; the one-mill-per-kilowatt-hour fee was set to zero on 16 May 2014 after a court ordered the Secretary to propose a zero fee. The Fund is statutorily spendable only on Yucca Mountain, which Congress will neither fund nor delicense. Meanwhile the Judgment Fund, a different federal pot, had paid out about $11.1 billion by the end of 2024 in damages and settlements for the Department's failure to take title to spent fuel — the very failure the first fund exists to prevent. The Department's own FY2024 audit put its total remaining liability at $37.6 billion to $44.5 billion, up from $34.1 billion to $41 billion a year earlier, a rise of about 10% in twelve months. The Fund now holds more than the high end of the liability it cannot be spent to discharge.

Established The fourth is the cost of reversal, which is the bottleneck nobody budgets for. Asse II is the case: 126,000 drums and about 47,000 m³ of low- and intermediate-level waste emplaced between 1967 and 1978 into a former potash and rock-salt mine bought by the federal government in 1965 as a research facility. The mine is unstable, cracking and taking on brine, currently collected and removed to keep it off the waste. The Bundestag legislated that the waste must be retrieved, in-place decommissioning having been abandoned on long-term-safety grounds. Retrieval is scheduled to begin in 2033 at an estimated €4.7 billion ±30%. Eleven years of emplacement in the 1970s has generated a reversal bill larger than an entire national disposal programme.

Established The fifth is delivery capability, and the United Kingdom has been unusually candid about lacking it. Sellafield's cleanup is estimated at £136 billion undiscounted, up 18.8% in real terms since March 2019 — £21.4 billion higher — over roughly a hundred years, at £2.7 billion in 2023-24, about two-thirds of the decommissioning authority's spending across all seventeen of its sites. Legacy pond and silo emptying milestones have moved back by six to thirteen years against 2018 estimates. Four major projects begun before 2018 are collectively £1.15 billion over cost and between 58 and 129 months late. A replacement sample-analysis facility was started in 2016, paused after seven years and £265 million, with a further £127 million assessed as wasted on the fallback laboratory refurbishment that was itself then paused. The Magnox Swarf Storage Silo, described in Parliament as the most hazardous building in the United Kingdom, has been leaking radioactive water since 2018 and is now expected to keep leaking into the late 2050s, up to thirteen years longer than previously thought, at a volume sufficient to fill an Olympic swimming pool roughly every three years. The national auditor's conclusion is that he cannot conclude Sellafield is achieving value for money yet.

Frontier The sixth is that consent, the mechanism everybody now recommends, has a poor record outside the place it was invented. It worked in Finland and in Canada. It failed in Cumbria in 2013, at Theddlethorpe in June 2025, and at Suttsu and Kamoenai in Japan where two municipalities cleared stage one in November 2022 and the prefectural governor opposed advancing. Japan has now effectively abandoned volunteerism: the literature survey begun on Minamitorishima on 20 May 2026 is the first initiated by central government rather than requested by a locality. Whether an imposed process can succeed where consent-based ones have not is an open institutional question, and the record so far answers it in the negative.

Established Disposal depends on nothing on this map, and that is an unusual and important sentence in a frontier-research corpus. Deep geological disposal uses conventional deep mining, canister fabrication, bentonite emplacement and remote handling, all demonstrated at full scale and all trialled end-to-end with dummy canisters in Finland before the operating licence was sought. There is no pending physics result, no unbuilt machine and no unproven material on the critical path. What the field waits on is a consent process capable of siting the facility, a funded liability behind a project measured in centuries, and a regulator with the throughput to review a licence in less than half a decade — and those three are institutional, financial and administrative capabilities rather than discoveries.

Frontier One partial exception, stated honestly. The copper canister is the single engineered component whose long-term behaviour is genuinely disputed. It is not a dependency in the usual sense, because the regulatory case is explicitly constructed to survive its total failure, but a brief that claimed no open materials question here would be overstating.

Established What depends on this brief is the more consequential direction. Every argument for expanding fission assumes a disposal route that at present exists, operating, in no country and, by the end of 2027, in one. Advanced fission depends on it directly and in a specific way: fuel-cycle choices determine the composition and volume of what arrives here, and the National Academies' finding that a closed fuel cycle does not eliminate the need for a repository means that no reactor design decision removes this dependency. Small modular reactors depend on it in a different way — by multiplying the number of sites at which spent fuel is generated and stored, they multiply the interim-storage and transport problem while leaving the disposal problem at the far end unchanged in kind and larger in count.

Established Interim storage is the dependency that is currently doing the load-bearing, and it was not designed to. Roughly 75% of the world's discharged fuel sits in storage rather than in disposal, in a regulatory regime with 40-year certificates renewable for up to 40 years and no specified maximum. Whatever the disposal programmes do, storage is the system on which the entire inventory depends in the meantime, and its licence architecture assumes a destination that does not yet exist.

Established The most instructive experiment in this field was an accident, and its cause was a procurement document. On 14 February 2014 a drum breached in Panel 7, Room 7 of the Waste Isolation Pilot Plant. The Department of Energy's Accident Investigation Board, reporting in Phase 2 on 16 April 2015, identified the cause as “an exothermic reaction of incompatible materials in LANL waste drum 68660 that led to thermal runaway, which resulted in over-pressurization of the drum, breach of the drum, and release.” The chemistry was metal nitrate salts, a wheat-based organic cat litter, and a neutralisation reagent, undergoing hydrolysis, oxidation and nitration of the organics. The administrative root cause is the finding. The absorbent specification changed from inorganic zeolite clay to organic cat litter when “kitty litter/zeolite clay” became “kitty litter (clay)” in internal communications, with inadequate technical review. The Board's local root cause was the laboratory's use of an organic wheat-based absorbent instead of the directed inorganic one; its systemic root cause was the failure of two federal field offices to ensure that the laboratory had adequately developed and implemented repackaging and treatment procedures. Twelve contributing causes were identified.

Frontier The negative result inside that accident is the part almost nobody records. The laboratory produced roughly 700 similar drums with the same mixture. Only one failed, and investigators never determined why. That is an unresolved question about a mechanism that has already caused a three-year repository closure, and it means the population of drums that could in principle do this again was characterised by outcome rather than by understanding.

Established The recovery is the experiment in institutional response time. Operations were suspended in February 2014; the facility was authorised to resume in December 2016; emplacement from above-ground storage resumed in January 2017 and incoming shipments in April 2017. That is three years. The tail is longer: the replacement Safety Significant Confinement Ventilation System — needed because, in the operator's own framing, the facility was never designed to operate in a contaminated state — completed commissioning only in spring 2025, eleven years after the drum breached.

Established Finland ran the experiment the rest of the field should copy and largely has not. From 30 August 2024 to 4 March 2025 Posiva emplaced five dummy canisters at 430 m in a full-system trial run, with no spent fuel involved, before applying to operate. The value of that trial is visible in what it made possible: a regulator reviewing a facility it had watched perform, rather than a design.

Frontier LOT is the twenty-year experiment whose readout is disputed. Copper and clay packages at the Äspö Hard Rock Laboratory, two retrieved in autumn 2019, reported by the implementer in June 2020. What would settle it is not more argument but a specified re-analysis: metallographic cross-sections through the most corroded regions, traceable sample identification, and imagery at a resolution that supports the claim being made. Both sides agree those are the missing items; they disagree about whose omission it is.

Frontier Canister inspection is the experiment that has barely been run. Two horizontally stored canisters inspected at Calvert Cliffs in 2012 after more than fifteen years is close to the whole of the rigorous field record for chloride-induced stress corrosion cracking, on an American dry-storage population spread across 37 states at 68 general-licence and 17 specific-licence sites. The independent review board's phrasing — very few inspections, none with rigorous criteria and mature techniques — describes an experiment that is available, affordable and not being done at scale. It is the cheapest decisive experiment in this brief.

Frontier Two natural experiments are already running and will report without anyone building anything for the purpose. The first is Japan's Minamitorishima survey, begun 20 May 2026 and expected to take about two years: the first literature survey in the Japanese programme initiated by central government rather than requested by a locality, and therefore a direct test of whether imposition can substitute for volunteerism. The second is Switzerland's possible national referendum around 2031, which would be the first time a general public rather than a host municipality voted directly on a repository. The consent literature has no case of either.

Frontier And the transmutation question has a scheduled readout that is further away than the disposal it is offered as an alternative to. MYRRHA Phase 3 is the transmutation demonstration; full operation is now targeted for 2038, slipped from 2034. On current schedules Onkalo will have been emplacing spent fuel for about eleven years by then. Whether Phase 3 is funded as opposed to authorised in principle could not be established for this brief.

Established The national programme record is short enough to enumerate in full, and enumerating it is more informative than any summary of it. Each row below was verified from national or regulatory sources individually rather than taken from a single count, which matters because the aggregate claims in this field are usually assembled by interested parties.

Established The engineering itself is unremarkable, and that is the point. The Finnish and Swedish designs both use the KBS-3 method, which Posiva licenses from SKB: spent fuel assemblies inside a nodular cast-iron insert, inside a copper overpack, lowered into a vertical deposition hole, packed around with compacted bentonite clay that swells on contact with groundwater, in crystalline bedrock roughly two billion years old. Onkalo's emplacement level is at 430–433 m, with 8 m deposition tunnels, about 3,250 canisters, and an initially licensed capacity of 6,500 tU. Sweden's Forsmark is the same concept at larger scale: more than 6,000 canisters and about 60 km of tunnels. Each barrier is independently conservative and the system is designed to tolerate the failure of any one of them — which, as the position section notes, is exactly the argument the Swedish regulator fell back on when the copper barrier was challenged.

Established The French variant shows that the concept is not tied to one geology. Cigéo is sited at about 500 m in a claystone formation roughly 160 million years old, sized for approximately 10,000 m³ of high-level waste and 73,000 m³ of long-lived intermediate-level waste, with an operating period of about a century. Note the volume ratio: the intermediate-level stream is seven times the high-level stream by volume. A brief that talks only about spent fuel is describing a minority of what has to be emplaced.

Established A correction on a widely repeated figure. Onkalo is frequently described as 520 m deep. That number appears to describe the deepest point of the ONKALO characterisation tunnel rather than the emplacement level. The Finnish Government gives about 430 m; the trial run placed canisters at 430 m; press reporting from the site gives 433 m; the industry association gives a range of 400–450 m. Use 430–433 m.

Frontier The engineering question with the longest tail is retrievability, and it is genuinely unresolved rather than merely unfashionable. Designs increasingly promise that emplaced waste can be recovered for some period, which is politically valuable and physically awkward: a repository optimised for permanent passive isolation and one optimised for retrieval are not the same facility, because the features that make retrieval easy — open access routes, un-backfilled tunnels, deferred sealing — are the features that make the passive case weaker. Finland's answer is to make the facility passively safe once sealed, requiring no future action; sealing is not expected until around 2120. The trade is rarely quantified in public, and no source consulted for this brief quantifies it.

The nearest neighbour is advanced fission, and the relationship is not the one usually asserted. Advanced fuel cycles are frequently offered as a way of shrinking or eliminating this problem; the National Academies' 2023 consensus study rules that out explicitly, finding that a closed cycle does not eliminate the need for a geologic repository because fission products still require disposal and material losses are inevitable. So the relation is: fuel-cycle choices determine what arrives here and in what form, and none of them determines whether a repository is needed.

Small modular reactors are adjacent in a quieter and more practical way. Their effect on this brief is arithmetic rather than conceptual: more sites generating and storing spent fuel means more dry-storage installations, more transport movements and more canister-handling operations — and canister handling is where San Onofre nearly dropped a loaded multipurpose canister eighteen feet in 2018. The failure modes documented in this brief are operational, and operational failure rates scale with the number of operations.

Underground cities are adjacent through deep excavation engineering and through very little else; the shared content is rock mechanics and ventilation, and the WIPP ventilation replacement — commissioned eleven years after it was needed, because the facility was never designed to operate contaminated — is the one transferable lesson.

Megaproject governance is adjacent because a 175-year Canadian project and a Finnish facility sealing around 2120 under a licence expiring in 2070 are the limiting cases of every accountability question that field studies. So is civilisation resilience planning, for institutional memory: Finland's design answer is passive safety after sealing, requiring no future action, which is correct and does not address the century of active operation before sealing.

Established The defining institutional fact is that consent-based siting has succeeded exactly twice, and both successes took more than a decade. Finland's Nuclear Energy Act gives the host municipality an absolute veto: acceptance by the site municipality is a precondition before the national government can accept a nuclear project. A 1994 amendment banned both import and export of nuclear waste, which removed the option of exporting the problem and thereby forced the domestic negotiation to conclude. Canada reached a site the same way, over 14 years from 2010, ending in a joint decision with the Wabigoon Lake Ojibway Nation on 28 November 2024. Both are real successes and both are slow.

Frontier But the mechanism is more fragile than its reputation, and the Finnish detail is the evidence. Eurajoki's first council vote was 13–13 in 1993, resolved by the chairman's casting vote; then 15–10 in 1994; then 20–7 in 2000. Seven years of drift, in a community already hosting an operating nuclear plant on the same island. The veto did not manufacture consent so much as guarantee a long negotiation with a pre-selected partner. Since then the same mechanism has failed in Cumbria in 2013, at Theddlethorpe in June 2025 after a community partnership formed only in June 2022 and more than £2 million in community funding had been paid, and in Japan at Suttsu and Kamoenai, which cleared stage one in November 2022 only for the prefectural governor to oppose advancing. Two successes, four failures, and the successes both had a pre-existing nuclear relationship.

Frontier Japan has now abandoned volunteerism, and this is the institutional experiment to watch. The literature survey begun at Minamitorishima in Ogasawara village on 20 May 2026 is the first in the Japanese programme initiated by central government rather than requested by a locality. It is the fourth location nationally and is expected to take about two years. Japan's own target of a repository by 2033–2037 is described in national reporting as considered unlikely; the country holds about 19,000 t of spent fuel in temporary storage and will eventually need capacity for more than 40,000 canisters. Whether imposition can succeed where consent has not is the open question, and the record so far answers it in the negative.

Established The institution that does not exist, in the country with the largest inventory, is a siting authority. The United States has no legal ability to consider any site but Yucca Mountain, a fund it cannot spend on anything else, and an adjudication it has suspended. It also has no interim-storage provision in current executive policy: the executive order of 23 May 2025 on the nuclear industrial base directs evaluation of reprocessing and recycling and asks for recommendations on the efficient disposal of wastes generated by recycling through a permanent disposal pathway — and contains no mention of the Nuclear Waste Policy Act, of Yucca Mountain, or of geologic repositories by name, and no interim-storage provisions at all. The Department's own one-year anniversary summary a year later listed $19 million for recycling research and development, requests for information on innovation campuses, and a surplus-plutonium programme — and no repository action whatsoever. The policy instrument that would unlock disposal is an amendment to a 1982 statute, and no branch of government is currently pursuing one.

Established The regulator is the institution whose throughput actually sets the schedule, and nobody budgets for it. Finland's review of a finished, trialled, uncontested facility took four years and seven months, with deadlines extended repeatedly including a missed one in June 2026. Sweden's took from application to government approval to court permit to construction across more than a decade. Switzerland's general licence application of November 2024 is under review until 2027 with a Federal Council decision around 2029–30 and a possible referendum around 2031 — seven years from application to political finality before construction begins around 2034. Regulatory throughput is the one input to this process that no programme has ever improved, and it is the one that binds after everything else has been solved.

Established Delivery assurance is the last institutional layer, and the British case shows what it looks like when a government grades its own work honestly. The national infrastructure assurance body rated the geological disposal facility red in its annual report of 11 August 2025: successful delivery of the project appears to be unachievable. The date has moved from 2040, stated in 2015, to the late 2050s. The whole-life cost estimate is £20 billion to £54 billion — a range so wide it is itself a statement about institutional confidence. Sellafield meanwhile signed 16 non-disclosure agreements in three years and paid £377,200 in employment settlements in 2023-24, which a parliamentary committee raised as a governance concern in its own right. A self-assessment this negative is unusually strong evidence, and it is the sort of evidence that only exists where an institution has been built to produce it.

Evidence quality is the first ethical question here and it has an unusually sharp instance. Frontier The best peer-reviewed synthesis of the copper-corrosion dispute is a 2023 paper in Advances in Geosciences — refereed, with a named editor and a named reviewer plus one anonymous referee — written by four employees of the Swedish regulator whose own approval decision is the subject of the dispute. It is simultaneously peer-reviewed and an interested party's account. That is not a reason to discount it; its factual content on corrosion mechanisms is the best available and this brief relies on it. It is a reason to say so in the open, and to note that its own conclusion concedes residual openness. The opposing case is carried in this brief largely by a non-governmental organisation explicitly opposed to the programme, and this brief has no unmediated statement of the applicant's own case: SKB's technical reports on the LOT experiment were not obtained. A reader should treat section by section: the sequence of decisions is established, the interpretations are contested, and one party is missing from the record here.

Established Public money and public accountability produce the second, and the American arrangement is close to indefensible on its own terms. Utilities collected a fee from ratepayers and paid $24.6 billion into a fund between 1983 and 2025. The fee was set to zero in 2014 after litigation. The fund now holds $51.4 billion, exceeding the high end of the Department's own remaining liability estimate, and is statutorily spendable only on a site Congress will neither fund nor delicense. A separate federal pot has paid $11.1 billion in damages to the utilities for the resulting failure to take title. The public therefore pays twice: once into a fund that cannot be spent, and again in judgments for the consequence of not spending it. That is not a design; it is what a statutory deadlock looks like after four decades.

Frontier A 2026 development sharpens the accountability question and is flagged as unsettled. An academic policy centre argued in March 2026 that the Department quietly amended its standard contract with plant owners from the 2000s onward without publishing the changes in the Federal Register, substantially weakening new owners' claims, potentially imposing a wait of up to 70 years before spent fuel removal, and materially reducing federal exposure — with the conclusion that there is not much incentive for the Department to do anything in the near term. This is an academic policy centre rather than peer-reviewed work, and its direction aligns with the industry's interest in stronger contract rights. But the Federal Register point is a checkable factual claim, and if it holds it means the government reduced its own liability by administrative means without the notice procedure that exists to make such changes visible.

Established Opportunity cost is measurable here in a way it rarely is. Germany spent €1.9 billion over 43 years exploring a site that failed the overburden criterion on all three indicators, and will spend €4.7 billion undoing eleven years of emplacement at a different site that was never a licensed repository. The United Kingdom assessed £127 million as wasted on a single paused laboratory refurbishment and £265 million on the replacement facility it was standing in for. Each decade of British repository delay requires another interim store at £500 million to £760 million. These are not the costs of solving the problem; they are the costs of the interval before solving it, and they are being incurred now.

Frontier The intergenerational question is genuine and cuts both ways, and this brief declines to resolve it. Deep disposal makes an irreversible decision for people who cannot be consulted. Indefinite storage transfers an unbroken maintenance obligation to the same people, with the added feature that they did not choose it either and the revenue stream funding it may have stopped. Retrievability is offered as the reconciliation and is not free: the design features that make retrieval easy weaken the passive safety case, and no source consulted quantifies the trade.

Established Finally, the obligation this brief takes on itself: to state what it could not verify. It could not read the 2014 Continued Storage Rule and therefore makes no claim about the regulator's assumptions for indefinite storage. Its $51.4 billion fund balance rests on an industry association because the government audit's financial statements did not render. Its global inventory figures reach it through industry-aligned trade press reporting the IAEA rather than from the IAEA directly. Its $2 billion figure for the WIPP recovery reaches it second-hand through an opposed campaigner relaying a newspaper analysis, which is why the range half a billion to two billion is given rather than a point estimate. It has no post-2009 figure for Yucca Mountain spend, no cost estimate for the French repository, no German target date for selecting a replacement site, and no confirmation of whether the Belgian transmutation reactor's third phase is funded as opposed to authorised. And it cannot confirm whether the Finnish Government granted the operating licence in the weeks after 4 August 2026. These are the brief's gaps, and stating them is cheaper than being caught by them.

Established The civilisational reading is that this is a solved engineering problem attached to an unsolved institutional one, and the ratio between them is the finding. Finland's entire national solution costs roughly 2% of Sellafield's cleanup bill and less than Germany's single retrieval mistake at Asse II. Doing it properly is dramatically cheaper than not doing it, and not doing it is what most nuclear nations have chosen by default rather than by decision. The general principle this case illustrates is uncomfortable and recurs across this corpus: the cost of institutional paralysis is not zero, it is not deferred, and it is usually larger than the cost of the action being avoided. The United States has $51.4 billion it cannot spend and has paid $11.1 billion for not spending it.

Frontier The timescales raise a resilience question this map takes seriously elsewhere. A 175-year Canadian project, a Swedish programme developing into the 2080s, and a Finnish facility sealing around 2120 under an operating licence that expires in 2070 all assume institutional continuity longer than most states have managed. Finland's design answer is passive safety after sealing, requiring no future action, and that is the right answer to the post-closure problem. It does not address the century of active operation before closure, during which the facility depends on a licensing authority, a funded operator and a competent regulator continuing to exist in recognisable form.

Established For the wider energy argument the implication is narrow and firm, and it cuts against both camps. Waste disposal is not a reason to avoid fission: the geology works, the engineering is conventional, and one country has built the thing. But the near-total absence of operating repositories is also not evidence that disposal does not work — it is evidence that almost nobody has finished doing it, and every case of not finishing has an institutional rather than a technical cause. Both the anti-nuclear reading and the pro-nuclear reading of the base rate are wrong, in symmetrical ways.

Speculative The deepest civilisational content is a claim about intergenerational obligation that no source in this brief settles. Deep disposal asks the present to make an effectively irreversible decision on behalf of a future that cannot be consulted; indefinite interim storage asks that future to keep maintaining a facility it did not choose, funded by revenues that stopped in 2014 in the American case. The disposal option is conventionally treated as the one requiring justification. The storage option is the one requiring an unbroken institutional chain across centuries, and nobody has been asked to consent to it.

Established What already happened, because this timeline usually starts far too late. Finland began screening national territory in 1983, three years before Chernobyl. Germany designated Gorleben provisionally in 1977 and abandoned it in 2020. The United States passed the Nuclear Waste Policy Act in 1982 and amended it in 1987 to name a single permissible site. Waste was emplaced at Asse II from 1967 to 1978 and will begin coming back out in 2033. Every date that follows sits on top of half a century of prior programme.

Frontier Late 2026 to early 2027: the first emplacement, if it happens. The Finnish Government's licence decision proposal was expected in the autumn of 2026, with a further regulatory inspection required before operation even after a grant. The chief executive of the utility co-owner said in June 2026 that first emplacement would come “either at the end of this year or most probably at the beginning of next year.” That is an owner's schedule claim rather than an outturn, and is flagged accordingly. When it happens, the count of operating high-level repositories on Earth goes from zero to one.

Established 2033: Asse II retrieval begins, at an estimated €4.7 billion ±30%, to recover 126,000 drums emplaced between 1967 and 1978.

Frontier ~2035: Sweden begins emplacement at Forsmark, about ten years after construction started in January 2025, with full development continuing into the 2080s.

Frontier ~2038: MYRRHA reaches full operation, or slips again. The target was 2034 when funding was approved in 2018 and is 2038 as of the 2024 groundbreaking. Only Phase 3 demonstrates transmutation.

Frontier 2040 to 2050: France and Canada. Cigéo operations are expected around 2040–2050 following a public inquiry in the second half of 2026; Canada expects regulatory approval around 2032 and operation around 2040–2045.

Frontier ~2050: Switzerland, with a Federal Council decision around 2029–30, a possible referendum around 2031, and construction from about 2034.

Handwave Any date for a repository in the United States, the United Kingdom, Japan or Germany. None of the four has a site. The American programme is statutorily locked to a site whose adjudication is suspended and whose fund cannot be spent elsewhere. The British programme is officially rated as appearing unachievable and has slipped from 2040 to the late 2050s. Japan's own target of 2033–2037 is described in national reporting as considered unlikely, and the country is 24 years into a three-stage process at stage one. Germany restarted its search in 2020 and this brief could not verify a current target date for selecting a site. Dates in this category are not forecasts and none is offered.

Established Late 2120s: Finland finishes. Disposal continues into the late 2120s and sealing is expected around 2120. The operating licence now being sought runs to 2070, which is to say that the facility will need at least one more licence from a government that does not yet exist.

• Depends on Nothing on this map. Deep disposal uses conventional mining, canister fabrication and bentonite emplacement, all demonstrated at full scale and trialled end-to-end with dummy canisters before licensing, so nothing it waits on is a result another brief produces. The copper-corrosion dispute is the one open materials question, and the regulatory case is constructed to survive the canister's total failure, so it does not gate the capability.
• Requires (not on this map) A consent process capable of siting a repository, which has succeeded twice and failed in Cumbria, Theddlethorpe, Suttsu and Kamoenai. A regulator with the throughput to review a licence in under half a decade — the Finnish review of a finished, trialled facility took four years and seven months, and that is the floor rather than the average. A statutory siting authority that permits an alternative site to be considered at all, which the United States has not had since 1987. A funded liability behind a project measured in centuries, which Finland has at €3.0 billion and the United States has $51.4 billion of and cannot spend. And deep mining with remote handling, the only genuinely industrial requirement in the list and the only one nobody worries about.
• Enables Advanced fission and small modular reactors, both of which assume a disposal route in their expansion cases. The enabling relation is real and is asymmetric: no fuel-cycle or reactor choice removes the need for a repository, on the National Academies' finding, so the dependency runs one way only.
• Adjacent Underground cities, sharing deep excavation engineering and nothing else; megaproject governance, since a 175-year project is the limiting case of the accountability gap that brief describes; and civilisation resilience planning, for the question of institutional memory across a facility that seals around 2120 under a licence that expires in 2070.

“Onkalo is operational.” Established It is constructed and trial-run-complete but not licensed to operate, and none of the three descriptions usually offered — operational, in trial runs, licensed — is correct as of this writing. The regulator issued a favourable statement on 4 August 2026; the Government, not the regulator, grants the licence; a decision proposal was expected in the autumn; and even after a grant Posiva may not start operating before a further inspection under §20 of the Nuclear Energy Act. A widely-syndicated state news agency reported on 4 August 2025 that the Finnish regulator “has issued a license for the operation” of the facility. That report is wrong on two counts: the regulator does not grant operating licences, and no licence had been granted. It is contradicted by European press reporting in April 2026 that a licence was still expected within months, by further reporting on 1 June 2026 describing the licence as pending, and by the Finnish Government's own release of 4 August 2026. This brief cites the erroneous report only to document the error.

“Onkalo is 520 metres deep.” Established Use 430–433 m. The 520 m figure appears to describe the deepest point of the characterisation tunnel rather than the emplacement level. The Government says about 430 m, the trial run placed canisters at 430 m, press reporting from the site says 433 m, and the industry association gives 400–450 m. This is a small correction that recurs in almost every secondary account.

“Nuclear waste is a solved problem that politics blocks.” Established For the United States specifically there is a stronger and much better-sourced claim available, and it should be used instead: the technical review passed. Nuclear Regulatory Commission staff completed the Safety Evaluation Report in January 2015 finding that the Department of Energy's application met the regulatory requirements, with two exceptions that are legal rather than technical — the Department had not obtained land withdrawal or water rights. And the Government Accountability Office found that the decision to terminate the programme “was made for policy reasons, not technical or safety reasons,” with the licensing board confirming that no new technical or safety issues had arisen since the 2008 application. Those two citations are much harder to dismiss than the general claim, and they say something more precise: the programme did not lose a technical argument, it was never allowed to finish one.

“Yucca Mountain was proved unsafe.” Established Also unsupportable, in the other direction. The adjudication was suspended on 30 September 2011 and remains suspended; the environmental supplement was completed in early 2016; no hearing on the contentions was ever held. Neither “proved safe” nor “proved unsafe” is available. What is available is that it cost about $15 billion through 2009 — $9.5 billion from the Nuclear Waste Fund, $5 billion from other appropriations and $956 million from the Treasury Judgment Fund for industry lawsuits — and was estimated to need a further $41 billion to $67 billion to complete. This brief has no verified figure for spend after 2009 and does not offer one.

“The Finnish municipal veto is the transferable mechanism.” Frontier This is the most consequential correction in the brief, because consent-based siting is now recommended almost everywhere on the strength of the Finnish case. The Nuclear Energy Act does give the host municipality an absolute veto — acceptance by the site municipality is a precondition before the national government can accept a nuclear project — and a 1994 amendment banned both import and export of nuclear waste, foreclosing the option of making it someone else's problem. But the consent history is not the triumphal story usually told. Eurajoki's council voted 13–13 in 1993, decided by the chairman's casting vote; then 15–10 in 1994; then 20–7 in 2000. Consent was narrow, incremental, accumulated over seven years, and granted by a community already hosting an operating nuclear plant on the same island. The veto did not create consent; it created a seventeen-year negotiation with a pre-selected nuclear community. Whether that is transferable to a non-nuclear community is genuinely unresolved, and the records in Cumbria, Theddlethorpe, Suttsu and Kamoenai suggest not.

“The world has 494,000 tonnes of spent fuel.” Established It has about 448,000 tonnes of heavy metal. The 494,000 figure circulating in popular coverage is the same number expressed in United States short tons. These are not competing estimates and should never be presented as a range.

“Six workers were hospitalised in the WIPP release.” Established That belongs to a different event — the salt-truck fire of 5 February 2014, nine days earlier. The Accident Investigation Board states that no personnel were underground at the time of the 14 February radiological release. The commonly cited figure of twenty-one workers with positive bioassay results relates to above-ground personnel and could not be verified for this brief; no worker count is used here.

“The Swedish corrosion concerns were resolved.” Frontier They were closed administratively, which is a different thing, and the sequence should be stated rather than summarised. In January 2018 the Land and Environment Court declined to endorse the application until SKB demonstrated that degrading processes affecting the copper canisters are not a problem for long-term safety. In spring 2019 SKB submitted supplementary information and the regulator accepted it. In December 2021 the Swedish Council for Nuclear Waste recommended attaching continued copper-research requirements to any approval. On 27 January 2022 the Government approved the licence. The opposed NGO's characterisation is that approval came without the industry having been able to show that the copper canisters will work as intended; the sequence is established, the evaluative framing is an interested party's. And the regulator's own peer-reviewed synthesis concedes residual openness, recommending additional testing on representative oxygen-free phosphorus-doped copper and acknowledging that new information related to these processes may still emerge.

“Transmutation will eliminate the waste problem.” Handwave The National Academies say otherwise in terms, and the correct response is not to hedge. A fully closed fuel cycle including reprocessing “does not eliminate the need for a geologic repository.” Even at 100% efficiency — “a highly unlikely situation because processes are not perfect” — the residue is still radioactive fission products requiring disposal. Europe's flagship accelerator-driven system will not demonstrate transmutation before about 2038, slipped from 2034, and its Phase 1 target stations are for radioisotopes and fusion materials, not transmutation at all.

“Deep boreholes have been demonstrated.” Handwave Distinguish modelling from disposal. The March 2026 announcement described as validating borehole disposal was a computer study of generic shale and granitic host rocks with no waste emplaced, and the vendor's own caveat is that the option would require federal law to be amended to authorise high-level waste disposal in a borehole repository. Promotional framing of federal executive orders as vindication of the concept is not evidence of a regulatory pathway, because there is no regulatory pathway.

“The risk is that the geology fails.” Established No repository failure anywhere in this evidence base has been geological. The documented failures are a mis-specified absorbent at Los Alamos, an unlicensed mine used for disposal at Asse II, a canister nearly dropped eighteen feet at San Onofre, an overburden criterion failed on all three indicators at Gorleben, and a statutory deadlock in the United States. The residual risk is administrative and operational, and those are precisely what the reassuring version of the story treats as the easy part.

“Interim storage is fine indefinitely.” Established Dry casks are licensed or certified for up to 40 years with renewals of up to 40 years, ageing management programmes are required, and the regulator's own position is that its regulations do not specify a maximum storage time. That is not the same as an affirmative finding that indefinite storage is safe; it is the absence of a limit. The waste-confidence position of at least 120 years of safe storage is a bounded claim about a bridge, not a destination. And this brief could not read the 2014 rule in which the regulator sets out its assumptions for the indefinite case, so it makes no claim about them.