The short version is this: nobody serious disputes that a firmed renewables grid is cheaper to build over the next fifteen years, and nobody serious disputes that a paid-off nuclear fleet would be cheap to run over the fifteen years after that. The argument, every time, is about which fifteen years you’re standing in when you do the sum. I’ve sat through this debate three times now in various forms since the mid-2000s, and it always resolves to the same unresolved question: what do you do about the years in between.
So let’s agree on the units first, because half the shouting on this topic is really just two people using different denominators and accusing each other of dishonesty.
Agree on the units first #
Cost comparisons in this debate come in roughly three flavours: levelised cost of energy (LCOE, dollars per megawatt-hour averaged over plant life), total system cost (what it takes to keep the lights on across a whole grid, including transmission and firming), and total capital outlay (the sticker price, unadjusted for anything). Nuclear advocates tend to quote system-level arguments about a smaller, more compact grid needing less transmission. Renewables advocates tend to quote LCOE, where wind and solar are already the cheapest form of new generation in the National Electricity Market by a wide margin. Both numbers are real. They are just not measuring the same thing, and I’d argue a lot of the public shouting past each other on this comes down to that, not to bad faith.
CSIRO’s GenCost report, produced jointly with AEMO and updated annually, remains the closest thing Australia has to an independent reference point on this. Its most recent editions have put large-scale nuclear’s levelised cost well above firmed wind and solar even after accounting for integration costs, storage and additional transmission, largely because of long build times and the capital cost of first-of-a-kind reactors in a country with no existing supply chain, regulator experience or workforce. That’s the CSIRO’s position, not mine, and it’s worth reading the actual report rather than either side’s press release version of it.
What the Coalition’s own numbers say #
The Coalition’s nuclear policy, first detailed publicly in 2024 and refined since, proposes seven sites on the footprints of retiring coal plants, using a mix of large reactors and small modular designs. We’ve covered the cost and timeline questions around that plan in detail elsewhere, and the honest position hasn’t shifted much: the policy’s own modelling put the first reactor online in the mid-2030s at the earliest, assuming Australia repeals its long-standing moratorium on nuclear power under the Australian Radiation Protection and Nuclear Safety Act and equivalent state legislation, builds a regulator from scratch, and avoids the kind of overruns that hit comparable projects overseas.
On that last point, the reference case everyone reaches for is Georgia’s Vogtle plant in the United States, which came in at roughly double its original budget and about seven years late, for two reactors on a site with an existing nuclear workforce. Hinkley Point C in the UK has followed a similar trajectory. Neither of those is proof Australia would repeat the pattern. But it’s the pattern you’d want to actively argue you’ll avoid, not one you get to assume away, and I haven’t yet seen a version of the Coalition’s costing that grapples with why an Australian build, starting from zero domestic capability, would do meaningfully better than Georgia or Somerset did.
What firmed renewables actually cost, and where the number hides #
The renewables side has its own sleight of hand, and it’s worth naming plainly: LCOE for wind and solar looks brilliant in isolation because it doesn’t include the cost of making that power available at 7pm on a still winter evening. That’s what “firmed” is doing in the phrase firmed renewables, and it’s expensive. Big batteries, pumped hydro, new transmission to connect renewable energy zones, and enough gas peaking capacity to cover the multi-day gaps that batteries can’t reach, all add real dollars that don’t show up in a headline LCOE chart.
We’ve written before about the difference between pumped hydro and big batteries as firming technologies, and it matters here because they solve different problems at different costs. Batteries are getting cheaper fast and are excellent for daily peaks; four-hour and even eight-hour systems are increasingly standard for new projects. But they don’t cover a week of low wind and cloud cover across the eastern seaboard, which is where pumped hydro schemes, still-scarce long-duration storage, and gas step in. Projects like those backed by Akaysha Energy, now under BlackRock ownership, are being built at genuinely large scale to plug exactly this gap, and the pace of that build is one of the more underappreciated stories in the transition. The Capacity Investment Scheme, the federal mechanism underwriting a lot of this new firming capacity, has effectively become the government’s chosen instrument for de-risking exactly this problem, though whether it’s doing that even-handedly or quietly picking winners is a fair question and one we’ve asked directly.
Time is the variable both camps underplay #
Here’s the bit that gets lost in most of these arguments: cost and time aren’t separate questions, they’re the same question asked twice. AEMO’s Integrated System Plan, the document that actually schedules the transition, models coal retirements continuing through the 2030s regardless of what replaces them. If nuclear reactors aren’t generating until the mid-2030s at the earliest, something has to firm the grid in the meantime, and that something is renewables plus storage plus gas, whether or not Australia ever builds a reactor. Which means the true comparison isn’t nuclear versus renewables. It’s renewables-now versus renewables-now-plus-nuclear-later, and the second option doesn’t remove any of the near-term firming bill, it just adds a second bill on top, decades out.
I spent a clear night out at the Perth Observatory in Bickley a fortnight ago, the kind of sky Perth gets maybe twenty nights a year, and the guide running the session made the point that every forecast, astronomical or otherwise, is only as good as the assumptions you’re willing to write down. Energy cost modelling has the same problem. GenCost’s numbers assume Australia doesn’t repeat Vogtle. The Coalition’s numbers assume the same. Renewables firming numbers assume batteries keep getting cheaper at something like their current rate and that new transmission gets built on schedule, which, if you’ve followed the fights over transmission corridors in regional Victoria and New South Wales, is its own act of faith.
The bit both sides don’t like to say out loud #
Here’s my mildly unpopular view, and I’ll own it: I don’t think the argument over the maths is actually the argument that matters most. Both a heavily firmed renewables grid and a nuclear-plus-renewables grid can be made to work on paper, and both can be made to look ruinous depending on whose discount rate and whose overrun assumption you accept. What actually decides this is delivery capability, and on that front Australia has a demonstrated, current, unglamorous track record in renewables and storage that it simply does not have in nuclear. We’ve built ACEN’s New England Solar, Squadron Energy’s wind portfolio, and a genuinely enormous rooftop solar fleet that the ABS and the Clean Energy Regulator both track as world-leading on a per-capita basis. We have never built a civil nuclear reactor, never regulated one, and don’t currently have the trained workforce to run construction on seven sites at once even if the politics resolved tomorrow.
That’s not a technology argument. Nuclear plants overseas produce cheap, reliable, low-carbon power once built, and I’ve no interest in pretending otherwise, nor in pretending the safety case against modern reactor designs is what it was in the 1980s. It’s a capability and sequencing argument, and it’s the one I think decides this before the cost spreadsheets do.
Where I land #
On cost, over the horizon that actually matters for keeping the grid running through the 2030s, firmed renewables win, not because nuclear power is inherently uneconomic everywhere, but because the bill for firming has to be paid either way and nuclear adds a second, later, larger bill on top of it rather than instead of it. If Australia had started building reactors in the 2000s, this would be a different, more interesting argument. It didn’t, and the arithmetic of a fifteen-year construction runway against an electricity system that needs answers well before then isn’t really an argument any more, it’s a scheduling problem with a cost attached.
Whether that changes if small modular reactor costs come down the way batteries did, I genuinely don’t know, and anyone who tells you they do is guessing with more confidence than the engineering currently allows. We covered the broader case for and against nuclear in Australia at length a while back, and I’d still stand by most of what’s in there. For now, the grid we’re actually building is the firmed renewables one, whatever gets decided in Canberra about the other one.
— Tom Fitzgerald, Baseload & Fuels Correspondent
Photo by Gez Xavier Mansfield on Unsplash