The nuclear ban and the Greens veto

Steve has spent twenty-two years on the tools at Eraring, the coal-fired power station on the shore of Lake Macquarie that, on a hot February afternoon, can still supply close to a quarter of New South Wales' electricity (approx.). He was told in 2022 that Eraring would close in 2025. Then, after the New South Wales government struck a deal with the plant's owner, Origin Energy, to keep it running longer because the grid was not yet ready to lose it, he was told the closure date had slipped again (approx.; NSW Government–Origin Energy agreement, 2024–25). Steve is not fighting the transition. He has watched wind turbines rise on the ridgelines west of Newcastle and knows a fifty-year-old coal plant cannot run forever. What puzzles him is narrower than that. When Eraring finally does close, the site will still have the transmission lines, the cooling water and the grid connection that took decades to build, and that any large power station needs. In France, Canada, South Korea, Britain and the United States, a retiring coal or gas plant with exactly that infrastructure is often where a nuclear reactor goes next. In Australia, it cannot be, because building one there would break the law — not fail an economic test, break the law. A specific law, passed in Canberra when Steve was a teenager, that nobody has to keep defending on the facts to keep in force.

That law is what this topic is about. Nuclear power is not absent from Australia's grid because engineers and economists weighed it against wind, solar, batteries and gas, project by project, and found it wanting. It is absent because two Commonwealth Acts prohibit it outright, in blanket terms, before any such assessment can happen. Whether that blanket prohibition still makes sense as Australia chases a legislated, fast, deep decarbonisation target — and whether the Greens, its most committed defenders, are right to keep it locked — is the question this article puts to the evidence, after first stating the government's and the Greens' case for the ban at its strongest.

The law that closed the door before the debate started

Australia's nuclear prohibition is not a policy the current government invented, and it does not sit in one place. It rests on two federal statutes. Section 10 of the Australian Radiation Protection and Nuclear Safety Act 1998 (the ARPANS Act) bars the national regulator, ARPANSA, from ever issuing a licence to construct or operate a nuclear fuel fabrication plant, a nuclear power plant, an enrichment plant or a reprocessing facility, anywhere in Australia, regardless of size or design. Section 140A of the Environment Protection and Biodiversity Conservation Act 1999 (the EPBC Act) does the same job on the environmental-approval side: it prevents the federal Environment Minister from approving any of those same facilities, however clean an individual project's environmental assessment might otherwise be. Between them, the two sections mean a nuclear power proposal in Australia cannot clear either of the two federal gates — safety licensing or environmental approval — that every other large generation project has to pass through on its merits. It is refused entry before the merits are even examined.

Both provisions date to the late 1990s, under the Howard Coalition government — not a Labor or Greens creation. The historical record, as recounted in Parliamentary Library research and Hansard from the period, links the ban's origin to a Senate negotiation: the government needed the support of the Australian Democrats, who then held the balance of power, to legislate approval for a new research reactor at Lucas Heights in Sydney's south (approx., 1997–98). That reactor, now called OPAL and operated by the Australian Nuclear Science and Technology Organisation, still runs today — producing medical isotopes used in cancer diagnosis and treatment across the country, not electricity. Locking in a prohibition on power reactors was part of the price of getting the research reactor through the Senate.

It helps to be precise about what kind of ban this is. In energy policy, a moratorium usually means a suspension a government could lift by changing policy alone. Australia's is not that. It is a statutory prohibition, meaning only an Act of Parliament — not a minister's decision, not a change of government policy — can remove it. And the federal layer is not the only one. New South Wales, Victoria and Queensland each maintain their own state-level nuclear prohibition laws, some dating back to the 1980s (approx.), independent of the Commonwealth Acts. That means repealing the ARPANS Act and EPBC Act provisions alone would not be enough to let a reactor be built in Australia's three most populous states; the relevant state parliament would also have to act. The ban a worker at a retiring coal site runs into is not one law. It is a stack of them, federal and state, each requiring its own parliamentary majority to remove.

For a quarter of a century after the Acts were passed, the federal ban was maintained without serious challenge by governments of both colours — Coalition and Labor alike. The Australian Greens, formed nationally in the early 1990s out of the state-based environmental and anti-uranium movements of the 1970s and 1980s, have opposed nuclear power as a founding policy position for as long as the party has existed, and have used their growing Senate presence since the mid-2000s to argue against ever revisiting it. That long bipartisan quiet ended in June 2024, when then Opposition Leader Peter Dutton announced a Coalition policy to build seven nuclear power stations — a mix of large-scale reactors and small modular reactors — at the sites of retiring coal plants, under government ownership, with the first units targeted to begin generating from the mid-2030s (approx.). The policy went to the May 2025 federal election as a live choice between the two major parties, and the Coalition lost that election, returning Labor to government (Australian Electoral Commission, 2025 federal election results).

That result did not settle the legal question, because an election loss is not a repeal. What actually keeps the ban in force is a shared, and in the Greens' case unconditional, political consensus against it. Labor's own platform also opposes nuclear power, citing cost and time rather than principle — a position this article tests directly below. The Greens' opposition runs deeper and is not conditional on any particular cost or safety finding changing; it has been a fixed plank of the party's identity since its formation. That distinction matters for what happens next. Any future attempt to repeal the ARPANS Act and EPBC Act prohibitions would need to pass the Senate, where the Coalition has never on its own held a majority and where Labor and the Greens have, between them, reliably supplied the numbers to block it. The Greens do not need to act alone to keep the ban standing — but they are the party for whom no future evidence is likely to change the vote.

The evidence: what the rest of the world runs on, and what it would cost and take here

What the rest of the G20 actually runs on

Start with the comparison the ban makes unusual. Nuclear power is not rare among Australia's peer economies — the opposite is true. Of the G20's member economies, most either operate substantial nuclear fleets today or are actively building new reactors: China, India, Russia, France, the United States, Canada, the United Kingdom, South Korea, Japan and others all generate meaningful shares of their electricity from nuclear power, and several are expanding it (International Atomic Energy Agency, Power Reactor Information System, current data). Australia is not unusual for lacking nuclear power — several G20 economies, including Italy and Indonesia, also currently generate none. What makes Australia unusual is that it is one of the few advanced economies to make nuclear power illegal outright, by statute, rather than simply choosing, so far, not to build it. Italy is the closest comparison: it shut its last reactors after a 1987 referendum, but that decision sits in energy policy and regulation, not in a standing legal prohibition of the kind the ARPANS Act and EPBC Act impose, and Italy has since reopened a formal debate about returning to nuclear power (approx.). Australia's ban is harder to unwind than a policy preference, because unwinding it means repealing an Act of Parliament, not simply changing a government's mind.

Look at what nuclear actually contributes where it is allowed. France generates roughly two-thirds of its electricity from nuclear power, the highest share of any major economy, and combined with hydro and other renewables draws around four-fifths of its electricity from zero-emissions sources (approx.; International Energy Agency, Electricity data, most recent year). Sweden, similarly, draws around 95 per cent of its electricity from a mix of nuclear and renewables, mostly hydro (approx.). Canada gets close to four-fifths of its electricity from zero-emissions sources too, split between nuclear — concentrated in Ontario — and hydro (approx.). South Korea, the United Kingdom and the United States each draw a smaller but still substantial share, in the range of a third to a bit over half, from nuclear and renewables combined (approx.; IEA Electricity data). Every one of those countries reached a high zero-emissions share using both technologies together, not renewables alone. Australia, drawing roughly 40 per cent of its electricity from renewables and none from nuclear (approx.; AEMO, National Electricity Market generation data, 2023–24), is attempting the same destination using only one of the two tools its peers have used to get there — despite holding around 28 to 30 per cent of the world's known recoverable uranium reserves and exporting it to the very countries that do (approx.; Geoscience Australia; World Nuclear Association).

France still generates roughly two-thirds of its electricity from a technology Australian law does not permit anyone to build.

What it would cost and how long it would take here

Peer-economy comparisons answer whether nuclear power works elsewhere. They do not answer whether building it in Australia, starting now, from zero, would be a good use of the next decade's money and attention — and that is the harder, more specific question the government's own advisers have actually tried to answer. CSIRO and the Australian Energy Market Operator jointly publish an annual assessment called GenCost, built to give planners a consistent, independently reviewed estimate of the levelised cost of electricity, or LCOE — a standard measure that spreads a technology's full lifetime construction, financing and operating costs over the electricity it is expected to produce, so different technologies can be compared like-for-like. Because the report is produced by a national science agency using an open methodology that outside experts, including nuclear advocates, are invited to challenge each year, it is the most credible public reference point either side of the nuclear debate has. Across its recent editions, GenCost has consistently placed nuclear power — both large-scale reactors and small modular reactors — among the most expensive new-build technologies assessed for the Australian grid, well above the cost of wind and solar even after adding the batteries, transmission and other firming infrastructure needed to make renewables reliable around the clock (CSIRO/AEMO, GenCost report, most recent edition, approx.). The gap is not because nuclear is inherently more expensive everywhere — established nuclear nations build reactors more cheaply than Australia's starting position would allow. It is the same first-of-a-kind premium described in the steelman above, priced into a specific national estimate rather than argued in the abstract.

That premium bites hardest on small modular reactors, the technology the 2024 Coalition policy leaned on most heavily for several of its proposed sites. Small modular reactors are factory-built units, designed to be assembled on-site faster and more cheaply than a traditional large reactor because most of the manufacturing happens in a controlled factory setting rather than on a construction site. The idea is genuinely attractive on paper. The global track record so far is thin: only a small number of small modular or demonstration-scale reactors are actually operating anywhere in the world, mostly in Russia and China (approx.; IAEA PRIS), and the most advanced Western project, NuScale Power's Carbon Free Power Project in Utah — the first small modular design to receive United States regulatory certification — was cancelled in November 2023 after its projected costs rose well beyond what its prospective customers were willing to pay. That cancellation is a real, recent, and directly relevant data point: it happened in a country with an established nuclear regulator, an existing reactor fleet and a federal government actively subsidising the technology, and the economics still did not close.

On timing, the range of credible assessments — from CSIRO's GenCost work to independent engineering and industry estimates — puts the earliest plausible date for the first unit of Australian nuclear power, if the ban were repealed immediately, somewhere in the mid-2030s to the 2040s, once site selection, design licensing, workforce training and construction are all accounted for (approx.). That is later than the 2030 target this site examines in its analysis of the 43% target and the net-zero arithmetic. The Coalition's own commissioned analysis, produced by Frontier Economics and released in December 2024 (approx.), argued its nuclear-inclusive plan would cost less overall than the government's renewables-based Integrated System Plan — a claim independent economists and CSIRO both disputed, largely over differing assumptions about how much electricity Australia will need and how quickly demand will grow. The dispute was never fully resolved before the 2025 election settled the immediate policy question.

One design choice in the 2024 Coalition policy is worth examining on its own terms, because it changes the cost comparison in a way headline LCOE figures do not always capture. The policy proposed that the reactors be built and owned by the federal government, rather than by a private developer borrowing at commercial rates. Government ownership can lower the financing cost of a large, risky, long-lived asset, because government can typically borrow more cheaply than a private company can, and because taxpayers rather than private investors absorb the risk of delay and cost overrun. Some countries new to nuclear power, such as Turkey with its Akkuyu plant, built and financed by Russia's state nuclear company under a build-own-operate arrangement, have used similar logic to manage the same risk (approx.). That does not close the cost gap GenCost identifies — some of GenCost's own scenarios already assume a lower, government-linked cost of capital — but it does mean the fairest comparison is not simply "nuclear versus renewables" as if both faced identical financing terms. It is a comparison between a technology-neutral market, in which developers of solar, wind, batteries and gas compete for finance on commercial terms and bear their own risk of failure, and a proposal to make one specific technology, nuclear, the exception by having government absorb its risk directly. Whether that trade-off — public capital and public risk in exchange for a shot at zero-emissions baseload, meaning continuous power supply that does not depend on the weather — is worth making is a legitimate question. Under the current law, it is not a question Australia is even permitted to test.

Ruling something out by statute is not the same as ruling it out on the merits

The steel-manned case above is real, and nothing in this section disputes the underlying facts about cost, time and Australia's missing industrial base. The critique is about the mechanism, not the conclusion. Every other generation technology available to Australia — solar, wind, batteries, pumped hydro, gas, even coal — has to clear ordinary planning, environmental and market-approval processes, project by project, and can be approved or rejected on its individual merits: its cost, its site, its community impact, its place in the grid. Nuclear power alone is disqualified before any of that assessment can happen, by two blanket Commonwealth prohibitions that apply regardless of design, size, location or cost. A small modular reactor proposed for a retiring coal site with existing transmission, water and workforce, backed by a private developer willing to wear the financial risk, cannot even apply for a safety licence or an environmental approval to find out whether it would clear the bar other technologies have to clear. The GenCost findings that make up the strongest part of the government's case are not what is keeping nuclear off the grid. The law is. GenCost describes a cost problem regulators could, in principle, still test project by project. The Acts foreclose the test itself.

That is a genuine inconsistency worth naming plainly. Australia mines and exports uranium to France, South Korea, the United States, Japan and other countries that use it to generate a large share of their own zero-emissions electricity — supplying, in effect, the fuel for exactly the technology it will not let its own energy market consider. No other zero-emissions or low-emissions generation technology faces an equivalent blanket ban. Pumped hydro schemes, big batteries and offshore wind all face real planning and community objections in specific locations — the Snowy 2.0 project's cost and schedule blowouts are a documented example — but none of those objections is written into Commonwealth law as a prohibition on the technology itself, anywhere, under any circumstances. Gas-fired power, which produces meaningful greenhouse emissions and is examined in this site's analysis of gas, energy security and the interventions, remains fully legal to build. Nuclear is the single generation technology in Australia's energy mix that the law rules out before the economics are even weighed.

There is also a gap in the evidence itself that the ban creates and then hides. Every other major generation technology proposed in Australia eventually meets a real test: a specific project, at a specific site, goes through environmental assessment, community consultation and market approval, and the result — approved, rejected, delayed, redesigned — becomes part of the public record other projects can be judged against. No nuclear power project has ever been through that process in Australia, because none has ever been legally permitted to start it. That means the entire debate about nuclear power in this country remains an abstract, national argument about technology in general, fought with country-level comparisons and modelled cost estimates, rather than a specific argument about a specific site's actual costs, actual environmental findings and actual community reception. Advocates and opponents alike are arguing from projection, because the law has never allowed a single Australian project to generate the site-specific facts that would settle the argument either way.

The timing case in the steelman — that renewables can be built faster, so nuclear's slowness makes it the wrong priority for this decade — is the strongest part of the argument, and it holds for the 2020s. It says less about the 2030s and 2040s than it first appears to. AEMO's own Integrated System Plan sets out the pace of transmission and renewable-capacity build required to meet the legislated 2030 target, a pace this site's analysis of the 43% target finds has repeatedly run behind schedule on transmission approvals and grid connections (approx.). If that gap widens rather than closes, the fallback available under current law is not a reassessed, technology-neutral look at every zero-emissions option, nuclear included. It is more gas generation, or ageing coal plants such as Eraring kept running past their planned retirement, because those are the only dispatchable options — meaning generation that can be switched on when needed, rather than only when the wind blows or the sun shines — that the law still permits. A blanket prohibition does not just rule nuclear out of this decade, when the timing case against it is genuinely strong. It rules it out of the 2030s and 2040s too, when a slower-building, weather-independent zero-emissions technology might matter more, and when today's decision not to even begin the decade-long process of establishing a licensing pathway will already have been made.

The Eraring deferral from this article's opening is not a hypothetical illustration of that risk. It is the risk, already realised. The New South Wales government agreed to keep Australia's largest coal power station running years beyond its original closure date specifically because the renewables, storage and transmission meant to replace it were not yet ready to carry the load reliably (approx.; NSW Government–Origin Energy agreement, 2024–25). That is exactly the kind of schedule slippage the timing case against nuclear assumes will not happen, or will not matter, because renewables can supposedly be built fast enough to avoid it. In this instance, in the largest state economy, it did happen, and the fallback was not a faster nuclear build — that was never legally available — but an ageing coal plant staying open longer than planned. None of this proves nuclear power would have been ready in time to replace Eraring instead; on the timeline evidence marshalled earlier in this article, it almost certainly would not have been. It does show, concretely, that the assumption underneath the steelman case — that the renewables path will keep pace, so a slower alternative is not worth preserving — is not guaranteed. It has already needed a coal-fired backstop once.

This is where the Greens' role in this topic differs from an ordinary policy disagreement. Most contested energy positions in Australian politics move, at least a little, when the evidence shifts: parties adjust gas policy as reliability warnings change, adjust renewable targets as costs fall, adjust rebate schemes as budgets tighten. The Greens' opposition to nuclear power has not moved with the evidence in this way across three decades, because it was never framed as evidence-conditional in the first place — it is a founding identity position, tracing to the party's origins in the anti-uranium and anti-nuclear-testing movements of the 1970s and 1980s, not a conclusion reached by weighing GenCost's numbers against the alternative. That does not make the Greens wrong about the current cost and timing case — this article has already found real force in it. It does mean that if the cost and timing case changed — if a future GenCost edition found the gap had narrowed, or a viable small modular design emerged with a stronger global track record — the Senate arithmetic described earlier suggests the ban would very likely stay in place regardless, because the party supplying part of the blocking numbers has committed not to revisit the question on the facts.

A dispatchable, zero-emissions technology several peer economies rely on to decarbonise their grids is, in Australia, not weighed and found wanting. It is simply not allowed to be weighed at all.

The caution is not baseless

A fair audit has to credit what is genuinely sound in the case for keeping the ban, not just record it and move on. The construction record cited in the steelman is real and well documented, not selectively chosen: Flamanville, Vogtle and Hinkley Point C are the flagship nuclear projects of three advanced economies with existing nuclear industries, and all three still ran years late and well over budget (approx.). If countries with sixty years of reactor-building experience cannot reliably hit their own targets, a country starting from zero regulatory and industrial capacity has every reason to expect an even steeper first-of-a-kind premium, not a smaller one. CSIRO's GenCost report deserves to be taken seriously as the current best public estimate for the same reason it is cited throughout this article: it is produced by an independent national science agency through an open, annually consulted methodology, not by a government press office defending a policy it has already chosen. Reasonable critics have queried individual GenCost assumptions — some arguing its integration-cost estimates for renewables are too low, others arguing its nuclear estimates are too pessimistic — but no credible alternative estimate published in Australia has found nuclear power cheaper than firmed renewables over the 2030s.

The small modular reactor case specifically deserves the scepticism this article has given it. NuScale's cancellation was not a hypothetical risk raised by opponents; it was the most advanced Western small modular project, backed by a licensed design and government support, failing on cost grounds in real time, in 2023. Treating small modular reactors as a near-term solution to Australia's 2030s energy gap, rather than a longer-run technology still proving itself commercially, would have been optimistic beyond what the evidence supports. And the bipartisan history matters too: this was not a Labor-and-Greens-only position invented to block a Coalition idea. The Coalition itself held government for the bulk of the twenty-five years the ban was in force — under Howard, Abbott, Turnbull and Morrison — and never moved to repeal it, only reversing course once, in 2024, under a different leader facing a different set of energy-reliability pressures. A caution that four separate governments across both major parties maintained for a quarter of a century, largely without controversy, is not obviously an ideological artefact. Some of it looks like a genuinely durable, cross-party assessment of the numbers.

It is also fair to note what the government has done instead of nuclear, rather than judge the policy only by the technology it excludes. Commonwealth and state governments have committed tens of billions of dollars to transmission and storage investment through vehicles such as Rewiring the Nation, examined elsewhere on this site, and AEMO's Integrated System Plan sets out a detailed, costed pathway to a renewables-and-storage-dominated grid that does not depend on nuclear power at any point. Whatever the merits of ruling nuclear out by statute rather than by case-by-case assessment, the government has not simply banned an alternative and left a gap; it has backed a specific, alternative build-out with real money and a public plan, and it is reasonable to judge that plan on its own delivery record rather than assume its existence was only a justification for the ban.