Ninety seconds. That’s roughly how long it takes for someone at a public meeting on this topic to say the word “baseload” without defining it, and for someone else to say “firmed” without defining that either. I’ve sat through enough of these sessions, from a community hall in the Hunter to a webinar out of Adelaide, to know the argument rarely gets past that point. So let’s agree on the units first, because that’s the whole ballgame here.
Agree on the units before you agree on anything else #
Nuclear people quote levelised cost of electricity, LCOE, for a single large reactor running flat out for sixty years. Renewables people quote LCOE for a wind or solar project running for twenty-five, then add a separate number for storage and transmission. Neither number, on its own, tells you what it costs to keep the lights on in the NEM at five o’clock on a still winter evening in Traralgon. That’s a system cost question, not a plant cost question, and it’s the reason CSIRO’s GenCost report, done jointly with AEMO each year, tries to build a “levelised cost including integration” figure rather than a bare plant number. It’s not perfect. But it’s the most honest attempt going, and both sides of this argument tend to quote the bits of it that suit them and skip the rest.
The 2025-26 edition of GenCost put nuclear, on Australian-specific assumptions, at several times the cost per megawatt-hour of firmed wind and solar, even after loading the renewables number with the batteries, gas peaking capacity and extra transmission needed to make them dispatchable around the clock. That’s not a new finding. CSIRO’s had nuclear at the expensive end of the ledger for several years running now, and the gap has narrowed a little each edition as first-of-a-kind cost estimates get revised, but it hasn’t closed.
What the small modular reactor pitch actually promises #
The industry’s counter-argument, for a while now, has been that large reactors are the wrong comparison – small modular reactors are supposed to be cheaper because they’re factory-built and repeatable. We’ve covered the state of that technology in detail before, and the short of it is that SMRs remain mostly a paper product outside a handful of demonstration units overseas, none of them yet delivering the promised cost curve at scale. If you want the fuller argument, our piece on small modular reactors and the actual state of the technology goes through the vendor claims plant by plant. The honest position is that nobody has proven the cost case yet, in any country, at commercial scale. Betting Australia’s grid on a technology still proving itself elsewhere is a genuine risk, and it cuts against the reactor case more than its supporters like to admit.
The Coalition’s numbers and the ten-year gap #
The federal Coalition’s own costed plan, released before the last election cycle and revised since, put seven reactor sites on old coal ground with a headline capital figure that critics, including modelling commissioned by the outgoing government, put well beyond that, largely on construction time. We went through the plan’s cost and timeline assumptions in detail in our earlier look at the Coalition’s nuclear plan, and the core problem hasn’t changed: even optimistic builders elsewhere in the world, working from a standing domestic supply chain, are taking twelve to fifteen years from decision to first power. Australia has no domestic reactor supply chain, no enrichment industry to speak of, and a federal moratorium under the Australian Radiation Protection and Nuclear Safety Act 1998 and its state-level equivalents that would need repealing first. None of that makes nuclear impossible. It makes it slow, and slow is expensive in a way that doesn’t show up cleanly in an LCOE table, because every year of coal running past its planned retirement date to cover the gap adds its own cost, on top of whatever the reactor eventually costs.
Where the firmed renewables number actually gets soft #
I don’t think the renewables side gets a free pass here either, and this is the bit I’d push back on if I were sitting across from a clean energy lobbyist rather than a reactor one. The system-cost figures in GenCost lean on AEMO’s Integrated System Plan for transmission cost assumptions, and those REZ transmission builds have been running late and over budget almost everywhere they’ve been tried – we’ve tracked that slippage in New South Wales specifically, where the coal exit is comfortably ahead of the transmission that’s meant to replace it. See our piece on the NSW energy transition and its REZ rollout running late for the detail. If you price firmed renewables using the transmission cost assumed in the ISP three years ago, you’re understating the real number. If you price nuclear using the vendor’s brochure figure from the same era, you’re doing exactly the same thing in the other direction. Fair’s fair.
Firming is doing more work in this argument than either side admits #
We’ve written before about what “firming” actually means in NEM terms, and it’s worth restating here because the word gets used as if it’s a solved problem on the renewables side and an unsolved one on the nuclear side, when really both technologies need it, just differently. A reactor firms itself, mostly, running near-flat output around the clock, though even nuclear fleets overseas need peaking gas or storage to cover planned outages and demand peaks, so it’s not the zero-firming product it’s sometimes sold as. Wind and solar need batteries, pumped hydro or gas to cover the gaps, and the debate over which of those firms best is its own long argument – read our comparison of pumped hydro versus batteries for firming if you want that one in full. What both technologies share is that the firming cost is real, it’s large, and it rarely appears in the number quoted at a press conference.
The Capacity Investment Scheme has already placed its bet #
Here’s the part I keep returning to when this argument resurfaces every few months, usually timed to a state election or a coal closure announcement. The federal government isn’t waiting for the nuclear-versus-renewables debate to resolve itself. The Capacity Investment Scheme, run through AEMO and underwritten by Commonwealth revenue guarantees, is already contracting several gigawatts of firmed renewable capacity (batteries, pumped hydro, wind and solar bundled with storage) into the grid on rolling tender rounds. We’ve set out the mechanics in how the Capacity Investment Scheme actually works, and the practical point is this: whatever you think the fairest cost comparison looks like on paper, the market operator and the Commonwealth are deploying real capital against one side of the ledger, at scale, right now, while the other side remains a policy position with no contracted construction start. You can argue the CIS is picking winners rather than letting the market decide, and we’ve made that argument ourselves, but it’s still the closest thing Australia has to a revealed cost preference at the moment.
My own read #
I’ll declare an old habit here: I like a clear night sky, and out past Northam or up in the Wheatbelt you get one most nights of the year, no light pollution, just a slow twenty-four-hour rotation that doesn’t care what anyone in Canberra decides this decade. There’s something almost appropriate about that when you’re thinking about baseload, because the honest physics of a reactor (steady output, decade-scale build, forty- to sixty-year asset life) is a genuinely different bet to a grid built from thousands of distributed, weather-dependent assets stitched together with batteries and wires. Neither bet is wrong on its face. But on cost, on the numbers Australia actually has in front of it today, firmed renewables win, and they win by enough that the gap isn’t close to closing on any construction timeline anyone credible is proposing. Nuclear might make more sense in a country starting from a different industrial base, or with twenty spare years and a settled bipartisan position. Australia doesn’t have either. If someone wants to have the argument again in five years, once an SMR has actually been built somewhere on budget and on time, I’ll happily run the numbers again. Until then, agree on the units, look at what’s actually being contracted through the CIS and the state REZs, and the cost verdict isn’t really close.
For a broader look at whether the technology question even needs answering yet, our earlier feature does Australia actually need nuclear power? is worth revisiting – the politics have moved since it was published, the underlying maths largely hasn’t.
Source data referenced above comes from CSIRO’s GenCost project and AEMO’s Integrated System Plan, both worth reading in full rather than taking anyone’s summary of them, mine included.
– Tom Fitzgerald, Baseload & Fuels Correspondent
Photo by Energie-portal.sk on Unsplash