CSIRO and AEMO published the latest GenCost draft a few months back, and the number that mattered most sat quietly in a footnote: the “first-of-a-kind” premium on large nuclear in Australia, if construction started today, would stretch a plant’s levelised cost well beyond anything firmed renewables are currently contracting at. Nobody at the launch dwelled on it. Everybody arguing nuclear on talkback radio the following week ignored it completely. That’s roughly where this debate has sat since the Coalition put its seven-site nuclear plan on the table, and it hasn’t moved much since.
I’ve written about this before, and so has pretty much every energy desk in the country, but the arguments keep recycling because the underlying maths hasn’t actually changed. So let’s do it properly, fuel by fuel, without the flag-waving on either side.
Agreeing on what we’re comparing before we start #
The nuclear-versus-renewables fight usually collapses because the two sides are pricing different things. Nuclear proponents quote the overnight capital cost of the reactor alone, sometimes borrowed from a Korean or Chinese build with none of Australia’s approvals overhead attached. Renewables proponents quote the levelised cost of a stand-alone wind or solar farm, no firming included, which is a bit like pricing a car without an engine.
The only honest comparison is full-system cost: nuclear’s total build and financing cost per megawatt-hour delivered, against wind or solar plus however many hours of battery, pumped hydro or gas peaking it takes to make that generation dispatchable on demand. CSIRO’s GenCost report does exactly this and is the best public reference going, precisely because it’s forced to make these assumptions explicit rather than leaving them for talkback hosts to invent. I’d send any reader arguing this point at a barbecue straight to the CSIRO GenCost page before they go further.
What nuclear actually costs, and why the range is so wide #
Large nuclear in a country with no existing fleet, no domestic supply chain and no regulator currently licensed to approve reactors carries what the industry calls a first-of-a-kind premium, and it is not small. GenCost’s modelling puts a hypothetical Australian large-scale plant at a levelised cost multiple times higher than onshore wind or large-scale solar, even before firming is added to the renewables side of the ledger. Small modular reactors look more attractive on paper, but as I wrote in the piece on where SMR technology actually sits, there is still no commercially operating SMR fleet anywhere in the world delivering power at the cost curve the brochures promise. Every cost figure for SMRs remains a projection, not a delivered result.
The Coalition’s own nuclear plan, released with seven proposed sites on retiring coal land, puts first power sometime in the 2030s at the earliest, assuming legislative change, a new regulator, workforce training and a construction programme starting from scratch. I went through the detail in the cost and timeline breakdown and the honest conclusion there still holds: even generous assumptions put commissioning well past the retirement dates of the coal plants nuclear is meant to replace.
What firmed renewables cost, and where the hidden bill sits #
Wind and solar are now comfortably the cheapest form of new generation in the NEM on a standalone basis, and nobody serious disputes that anymore. The argument moves, correctly, to firming: batteries, pumped hydro, gas peakers and transmission, all of which cost real money and all of which get left out of the simpler renewables pitches.
Four hours of battery storage attached to a solar or wind farm adds meaningfully to the levelised cost, but even with that addition, GenCost’s firmed renewables figures sit well below nuclear’s range across every scenario modelled. The gap narrows for longer-duration firming, which is exactly why pumped hydro projects like Snowy 2.0 and Queensland’s Borumba matter so much to the overall equation; I covered the Snowy cost blowouts in the piece asking whether Snowy 2.0 will ever earn its keep and the number there, past $12 billion and still climbing, is a sober reminder that firming costs are real and have already run over budget once. The gas peaker alternative has its own maths, covered in the firming comparison between gas peakers and big batteries, and the short version there is that gas still wins on flexibility even as batteries eat into its territory on price.
None of that changes the headline comparison. Even an honest, fully firmed renewables build, mistakes and all, comes in under nuclear’s range by a wide margin on every public modelling exercise going.
The transmission and timeline problem nobody prices properly #
Here’s where I think both camps are a bit dishonest. Renewables advocates routinely understate the transmission cost of the Renewable Energy Zone build-out, the cost of connecting wind and solar scattered across regional NSW, Queensland and Victoria to where the demand actually sits. I’ve covered the blowouts and delays on this in enough detail, from VNI West stuck in a paddock dispute to who actually pays for the transmission build-out, that I don’t need to relitigate it here. It is expensive and it is running late, and that cost belongs in the renewables ledger whether or not it shows up in a standalone levelised cost figure.
But nuclear’s advocates have the opposite problem and a bigger one: a construction timeline that doesn’t fit the physical reality of when Australia’s coal fleet is actually leaving. Eraring is now pencilled in for 2027 closure after the deal I wrote up in the Eraring extension piece, Yallourn isn’t far behind, and EnergyAustralia’s own closure countdown makes clear those retirement dates aren’t moving much further. A nuclear plant that enters service sometime in the mid-2030s, on the most optimistic reading of the Coalition’s own numbers, is simply not available to fill that gap. You can build the best reactor in the world and it still won’t turn up to the right decade.
A small aside from the back shed, if you’ll indulge me #
I keep a cheap telescope in the back shed in Perth’s southern suburbs, and most clear nights lately I’ve been looking at Jupiter rather than reading another nuclear cost estimate, which tells you something about my patience for this argument by October. But the comparison with astronomy actually holds up: you can calculate an orbit precisely with good enough instruments, and the two camps in this energy debate are both, in their own way, refusing to use the instruments on offer. GenCost isn’t perfect. It gets revised every year and the nuclear figures in particular carry wide error bars because there’s no Australian build to benchmark against. But it is a published, peer-reviewed, consistently updated reference, and the alternative on offer from most advocates on both sides is a number picked because it suits the conclusion they’d already reached.
Where I land on it #
My honestly mildly contrarian view, and it won’t make either camp happy: nuclear proponents are right that firmed renewables carry more hidden cost and delay than the glossier pitches admit, particularly on transmission and long-duration storage. But they are wrong that this closes the gap to nuclear, because it doesn’t come close. The maths, even generously read, has nuclear landing at several multiples of firmed renewables’ delivered cost, and the timeline mismatch with coal retirements is arguably the bigger problem of the two. A plant that might generate its first megawatt-hour in the mid-2030s doesn’t help anyone managing the grid through the back half of this decade, which is when the actual gap opens up.
Firmed renewables have real problems. The state renewable target scorecard I put together a while back shows several jurisdictions well behind their own build schedules, and the Capacity Investment Scheme tender process, which I went through in the CIS tender clock piece, favours large incumbent developers in ways that aren’t always good for competition. These are legitimate criticisms. They are not, however, arguments for nuclear on cost, because nuclear’s own numbers don’t get better just because renewables’ numbers have some rough edges.
AEMO’s Integrated System Plan, which I broke down properly in the piece on what the ISP actually says, doesn’t include large-scale nuclear in its optimal development path for a reason, and it isn’t ideology. It’s because the plan has to work within a timeframe nuclear can’t meet and a cost envelope nuclear doesn’t fit. Reasonable people can disagree about whether Australia should have built a nuclear industry forty years ago, when the sums looked rather different. But arguing we should bet the 2030s grid on it now, at this cost, on this timeline, is a different claim entirely, and on the evidence currently published, it doesn’t stack up.
– Tom Fitzgerald, Baseload & Fuels Correspondent
Photo by Evangelos Mpikakis on Unsplash