Nuclear Waste From SMRs: How Much, How Dangerous, and Where Does It Go?
The nuclear industry's smallest reactors may have an outsized waste problem — and the honest answer is more complicated than either side admits.
Nobody builds a nuclear reactor and then thinks, “Great, now what do we do with all the radioactive garbage?” And yet here we are, watching the SMR industry sprint toward deployment while the waste question jogs behind, waving its arms. It deserves a closer look. Not a panicked one, and not a dismissive one — but an honest, clear-eyed examination of what SMRs actually produce, how hazardous it really is, and whether the world has any credible plan to deal with it.
The short answer: SMRs produce broadly comparable waste to traditional reactors, with some important caveats that the industry would rather you didn’t read about. The longer answer takes a bit more unpacking.
What nuclear waste actually is
Before diving into SMR-specific concerns, it helps to understand what “nuclear waste” means in practice. The term covers a surprisingly wide spectrum of material.
Low-level waste (LLW) is by far the most common by volume — contaminated tools, gloves, filters, and protective clothing. This stuff loses most of its radioactivity within decades and is handled at near-surface disposal facilities.
Intermediate-level waste (ILW) includes reactor components, resins, and chemical sludge. More radioactive than LLW but not the nightmare fuel of public imagination.
High-level waste (HLW), which includes spent nuclear fuel (SNF), is the category that keeps regulators up at night. ☢️ It remains dangerously radioactive for tens of thousands of years. A single assembly of spent fuel from a pressurized water reactor is intensely radioactive, thermally hot, and chemically reactive. This is the waste category where SMRs face their most pointed scrutiny.
The widely-cited PNAS study from Stanford and University of British Columbia — probably the most rigorous independent analysis to date — found that SMRs will produce more voluminous waste than large reactors by a factor of 2 to 30, depending on design. That is not a typo. The reason is physics, not engineering sloppiness: smaller reactor cores inevitably leak more neutrons outward, which irradiate surrounding structures and create additional radioactive material.
The key waste categories nuclear professionals watch most closely are:
Spent nuclear fuel (SNF), the primary HLW concern
Tritiated water, particularly in heavy-water or gas-cooled designs
Greater-Than-Class-C (GTCC) waste, material too radioactive for near-surface disposal
Activated metals — reactor components bombarded with neutrons until they themselves become radioactive 🔬
Low-level waste streams generated in larger quantities than conventional plants
Does a higher volume automatically mean a more dangerous problem? Not necessarily. Volume is actually not the most critical metric for repository performance. What matters more is decay heat and the radiochemical composition of the waste. On those two counts, SMRs offer no particular advantage over large reactors.
The neutron leakage problem nobody talks about
Here is the physics that the promotional brochures never quite get around to explaining. 🔑
Every nuclear reactor produces energy by sustaining a controlled fission chain reaction. In a large gigawatt-scale pressurized water reactor, the core is big enough that most neutrons stay inside the fuel zone, doing useful work. In a smaller core, a higher fraction of those neutrons escape the fuel region entirely and slam into the surrounding reactor vessel, reflectors, and structural components. This process is called neutron activation, and it turns otherwise inert steel and concrete into radioactive waste.
This is not a design flaw that engineers can simply engineer away — it is an unavoidable consequence of geometry. Smaller cores have a higher surface-area-to-volume ratio, and physics doesn’t negotiate.
The Union of Concerned Scientists noted that per unit of electricity generated, SMRs produce essentially the same quantity of highly radioactive fission products as large reactors. That framing matters. When the industry says SMRs generate “less waste,” they usually mean less total mass or volume from a single unit — which is true but trivially so, since a single 300 MW SMR also generates far less electricity than a 1,200 MW conventional plant. 📊
The waste-per-megawatt-hour comparison is the one that actually counts. And that number, for most SMR designs, does not beat conventional large reactors. Some advanced designs — particularly non-light-water reactors using higher-density fuels — do slightly better on fission product mass, perhaps 10 to 30% less. But those same designs often introduce chemically reactive coolants (liquid sodium, molten salt) that create their own uniquely messy waste streams, some of which have no existing disposal pathway.
Specific radionuclides to pay attention to:
Iodine-129 — half-life of 15.7 million years, mobile in groundwater, relevant to repository leakage scenarios
Technetium-99 — half-life of 211,000 years, similarly mobile, geologically concerning
Selenium-79 — another long-lived fission product that travels easily through water
Tritium — produced in larger quantities by some SMR designs, and difficult to contain 🌊
The PNAS researchers concluded that the “intrinsically higher neutron leakage associated with SMRs suggests that most designs are inferior to large LWRs with respect to the generation, management, and final disposal of key radionuclides.” That is a serious finding from a peer-reviewed source, and the industry has not produced a compelling rebuttal. It is worth sitting with.
Are you factoring waste volumes into your assessment of different SMR technologies? Because the reactor salespeople almost certainly aren’t volunteering this information unprompted.
What Argonne National Laboratory actually says
To be fair — and this debate deserves fairness — not every authoritative voice reaches gloomy conclusions. 🔬
Argonne National Laboratory, a U.S. Department of Energy facility, published its own analysis comparing three SMR designs to conventional large reactors. Senior nuclear engineer Taek Kyum Kim and colleagues concluded: “All told, when it comes to nuclear waste, SMRs are roughly comparable with conventional pressurized water reactors, with potential benefits and weaknesses depending on which aspects you are trying to design for.”
Crucially, the Argonne team found: “There appear to be no additional major challenges to the management of SMR nuclear wastes compared to the commercial-scale large LWR wastes.”
That’s a notably different conclusion from the PNAS study, and the discrepancy is real. The two analyses use different metrics and study different designs, which explains much of the gap. The Argonne study focused on NuScale’s light-water VOYGR design, which is closest to conventional reactor technology. The PNAS study deliberately examined a wider range, including molten-salt and sodium-cooled designs, where waste profiles get considerably messier.
The honest synthesis: light-water SMRs produce waste that is broadly comparable to — and manageable alongside — conventional reactor waste. Advanced non-light-water SMRs are where the picture gets genuinely murkier, and where the regulatory frameworks are least developed.
NuScale itself, in its public documentation, says that a full 60-year-lifespan 12-module plant would generate spent fuel storable in just 89 dry casks, fitting on a concrete pad of less than one acre. That is a genuinely compact footprint for 60 years of clean electricity. The question is whether “compact footprint” translates to “solved problem” — and it doesn’t, quite. Those dry casks still need somewhere permanent to go. 📦
Key design-by-design differences that affect waste profiles:
Light-water SMRs (NuScale, GE-Hitachi BWRX-300) — conventional waste streams, best understood
Molten salt reactors (various designs) — chemically reactive waste, novel disposal challenges
Sodium-cooled fast reactors (TerraPower Natrium) — HALEU fuel required, which demands far more raw uranium to produce per kilowatt-hour than conventional fuel
High-temperature gas-cooled reactors (X-energy Xe-100) — graphite waste streams, long-lived activation products
Where the waste actually goes — and the uncomfortable truth
Here is where the conversation becomes genuinely sobering, regardless of where you stand on nuclear power.
The United States currently has over 90,000 metric tons of spent nuclear fuel sitting at reactor sites around the country, according to the U.S. Government Accountability Office. That stockpile grows by roughly 2,000 metric tons every year. The DOE has been legally required to build a permanent repository since the Nuclear Waste Policy Act of 1982. It has not built one. The proposed Yucca Mountain site in Nevada was effectively killed politically in 2010. No alternative permanent site exists.
That is a stunning policy failure spanning four decades and multiple administrations. 🏛️
The global picture is only marginally more encouraging. The world has exactly one licensed deep geological repository for nuclear waste — the Waste Isolation Pilot Plant (WIPP) near Carlsbad, New Mexico, and it only accepts defense transuranic waste, not commercial spent fuel. The countries closest to having a real long-term solution are:
Finland, whose Onkalo repository at 430 meters depth in crystalline bedrock is the world’s most advanced project, nearing completion
Sweden, which broke ground in January 2025 on a repository at Forsmark, 500 meters deep in 1.9-billion-year-old rock, designed to hold 12,000 tonnes of spent fuel in 6,000 copper canisters
Canada, which selected a site in the Wabigoon Lake Ojibway Nation-Ignace area in November 2024, after a consent-based process that took 14 years
France, making progress on its Cigéo project in clay formations
Switzerland, which applied for a general construction license for a repository at Nördlich Lägern in November 2024 ⛏️
The Swedish and Finnish models are particularly worth watching. Both use the KBS-3 method — spent fuel encapsulated in copper canisters surrounded by bentonite clay, buried deep in stable bedrock. Finland’s Onkalo will be sealed around 100 years from now and is designed to isolate waste for at least 100,000 years. The natural analogy that gives geologists confidence: the Oklo natural fission reactors in Gabon operated for millions of years, producing fission products that remained immobile in surrounding rock for roughly 2 billion years, even with groundwater access. Nature, in other words, has already run this experiment.
The United States has none of this. Communities hosting SMRs, particularly the data centers and tech facilities that are the industry’s hottest current customer segment, should plan to store their spent fuel on-site indefinitely — because, as the Union of Concerned Scientists pointedly observed, vendor promises to remove used reactor modules “are simply not credible, as there are no realistic prospects for licensing centralized sites where the used reactors could be taken.”
Any community signing an SMR contract today is, functionally, signing up to become a long-term nuclear waste management site. That deserves to be stated plainly.
What a responsible path forward looks like
None of this is a reason to reflexively oppose SMRs. Nuclear energy remains one of the lowest-carbon, lowest-air-pollution electricity sources ever devised, and the climate math is simply too urgent to wave away any low-carbon technology without good reason. ♻️
But the waste question demands honest answers, not cheerful minimization. Here is what a serious approach looks like:
Standardize light-water designs first — these have the best-understood waste profiles and the most existing regulatory infrastructure. Build the advanced non-LWR designs only once waste management pathways are actually certified, not just theorized. 🏗️
Create a statutory deadline for U.S. consent-based siting of a consolidated interim storage facility. The GAO has repeatedly recommended this. Congress has repeatedly ignored it. At some point that becomes a policy choice, not an oversight.
Require waste management plans before licensing, not after. The Information Technology & Innovation Foundation’s 2025 report noted that NRC is only beginning to address SMR-specific waste issues in transportation and disposal — which is late in the game for technologies targeting deployment this decade.
Fund genuinely independent, multi-design waste assessments — studies that include the full range of SMR designs, not just the ones closest to conventional reactors. The gap between the Argonne and PNAS findings needs to be resolved with better data, not partisan cherry-picking.
Be honest with host communities — about what “on-site interim storage” means in practice, about how long the waste remains hazardous, and about the current absence of a permanent solution in most countries. 🤝
The nuclear industry spent decades insisting that waste was a solved problem waiting on political will. That was partly true and partly self-serving. The specific case of SMRs — particularly advanced designs — introduces genuine new complications that the industry’s standard “comparable to conventional reactors” talking points don’t fully address. And the world’s policy infrastructure for handling that waste, outside of Finland and Sweden, remains embarrassingly underdeveloped.
So: how much waste do SMRs produce? Broadly comparable to large reactors per unit of energy, with important exceptions for advanced designs. How dangerous? Very — the same long-lived isotopes, the same multi-thousand-year hazard timelines, the same need for geological-scale isolation. Where does it go? Right now, mostly into pools and dry casks at reactor sites, waiting for governments to find the political courage to build the permanent repositories they promised decades ago.
That last part is the one worth demanding answers on — from policymakers, from vendors, and from the industry bodies that keep calling waste a “solved” problem while the spent fuel inventory grows by 2,000 tonnes a year with nowhere permanent to go.