Policy & Markets

Does Australia actually need nuclear power?

19 November 2026 · by Marcus Wren
6 min read·1352 words·Updated 19 Nov 2026

Zero. That’s how many nuclear power reactors are under construction in Australia right now, and it’s how many will be under construction in 2030, 2032 and most likely 2035 as well, regardless of who wins the next two federal elections. The Commonwealth’s own moratorium under the Environment Protection and Biodiversity Conservation Act 1999 is still law. So when someone asks me whether Australia needs nuclear, the first honest answer is a different question: needs it for what, and by when?

Because the case keeps shifting shape. Sometimes it’s about emissions. Sometimes it’s about reliability after coal leaves. Sometimes it’s just a general discomfort with how much steel and concrete the renewables build is chewing through. Those are three different arguments, and they don’t all point to the same answer.

What the Coalition plan actually proposes #

The Coalition’s policy, as it stands, nominates seven sites for either large reactors or small modular designs, mostly on or near retiring coal plant footprints, with the first unit notionally online sometime in the late 2030s. I went through the cost and timeline assumptions in detail in an earlier piece on the Coalition’s nuclear plan, and the short version is that even generous versions of the schedule have the first reactor arriving a decade or more after the coal unit it’s meant to replace has already shut.

That’s not a political dig. It’s a construction reality. The CSIRO’s GenCost report, produced annually with AEMO, has put large-scale nuclear at the top of the cost table for new-build generation in Australia for several years running, driven as much by the absence of a domestic supply chain and skilled workforce as by the reactor technology itself. You can disagree with GenCost’s assumptions. Plenty of nuclear advocates do. But you can’t disagree with the fact that Australia has never built a commercial reactor, has no enrichment industry, and would be starting a regulatory framework from close to scratch.

The grid-fit problem nobody argues with #

Large reactors want to run flat out, near their rated capacity, for years at a stretch. That’s how the economics work, because the capital cost is so front-loaded that every hour off-line is expensive.

Australia’s grid increasingly doesn’t want that.

On a sunny spring day the NEM now regularly sees wholesale prices go negative around midday, a pattern I covered in this piece on what negative prices actually mean. Rooftop solar alone is pushing minimum operational demand down across every mainland region, and AEMO’s own Electricity Statement of Opportunities has flagged minimum demand, not peak demand, as the harder forecasting problem in several of its recent editions. A baseload reactor sitting in that market either gets paid very little for large chunks of the day, or it needs the market to pay it to simply exist, via a contract-for-difference or a regulated asset base model, regardless of whether the grid wants its output at that hour.

That’s not a reason nuclear can’t work technically. France runs reactors alongside a renewables build and manages it through flexible output and interconnection. But France isn’t starting from Australia’s position, with one of the fastest rooftop solar uptake rates in the world and a grid built for synchronous coal, not for load-following nuclear bolted onto a solar duck curve that gets deeper every year.

Small modular reactors: the honest state of play #

Every few months someone points to small modular reactors as the escape hatch from the baseload-versus-solar problem, on the theory that smaller units can load-follow and site more flexibly.

I went deep on this in a piece on what SMR technology can actually do right now, and the state of play hasn’t moved much since. NuScale’s flagship US project, the one most often cited as proof of concept, was cancelled in 2023 over cost blowouts before a single module was poured. There is no operating commercial SMR fleet anywhere in the world that Australia could simply buy into. There are design concepts, demonstration projects in Canada and the US at early stages, and a great deal of investor material. None of that is the same as a technology ready to underwrite a 2035 commitment.

Follow the money on this one and it tells you plenty. The companies spending real capital in the Australian market right now, AGL, Origin, Tilt Renewables, AusNet, are not quietly banking land options near coal sites for SMRs. They’re building batteries, firming gas peakers and transmission. AGL and Origin’s diverging gentailer strategies both assume the 2030s grid is wind, solar, storage and firmed gas, not reactors.

What the emissions case leaves out #

The strongest version of the pro-nuclear argument is an emissions one: Australia needs firm, zero-carbon generation at scale, and wind and solar alone can’t deliver it on the geography and timeline required, so nuclear deserves a seat at the table even if it’s expensive and slow.

It’s a coherent argument. I just don’t think the timeline supports it as policy for this decade or the next.

Coal is leaving the NEM on a schedule that’s largely locked in by plant age and maintenance economics, not by policy choice. Eraring’s extension, which I covered in the deal that bought NSW two more winters, bought time, not a reprieve. Yallourn in Victoria is scheduled to close in 2028, a deadline I wrote about in relation to the Latrobe Valley workforce and battery build racing to meet it. None of those closure dates move for a reactor that’s still a decade from a concrete pour. Whatever replaces that capacity on the relevant timeline has to be buildable now, and nuclear simply isn’t on that list.

The counterfactual the advocates don’t like #

Here’s my mildly contrarian bit, and I’ll own it. I think the renewables-plus-storage-plus-gas pathway is also running behind its own schedule, and nuclear advocates are right to point that out even if their preferred fix doesn’t follow from the diagnosis.

VNI West is still arguing with a map. NSW’s renewable energy zone rollout is lagging its own coal exit. Marinus Link has been staged and restaged. If your argument is “the current plan is too slow and too expensive,” you’re not wrong. Where I part company with the nuclear camp is the conclusion. The answer to a transmission and REZ build that’s running late isn’t a generation technology with a longer lead time and a worse track record on cost overruns than the thing it’s meant to fix. It’s building the wires and the storage faster, with better planning approval processes and probably more public capital, the way Queensland’s pumped hydro and public ownership bet is attempting, however imperfectly.

I’ll admit the batteries-and-wires pathway isn’t moving fast enough either. That’s a fair criticism from the nuclear side. It just doesn’t make their alternative faster.

What would actually change my mind #

A few things would shift my view, and I say this as someone who isn’t ideologically opposed to the technology.

First, a working SMR fleet operating commercially somewhere comparable, at a cost that beats GenCost’s current estimates by a wide margin rather than a modest one. Second, a federal moratorium repeal paired with an actual state-level approvals pathway, since several of the proposed sites sit in states that have their own separate bans. Third, evidence that the capital markets, not just government commitments, are willing to underwrite first-of-a-kind nuclear risk in Australia on commercial terms. None of those three conditions currently holds.

Until they do, the honest position isn’t “nuclear is impossible” or “nuclear is inevitable.” It’s that nuclear is a 2040s conversation being conducted as though it’s a 2030s decision, and the grid doesn’t have the luxury of waiting to find out which one it turns out to be. A bit like declaring your Test batting line-up settled on day one of a five-day match: the headline looks confident, but the conditions haven’t finished talking yet.

Whether that changes depends less on reactor technology and more on whether the current build-out of transmission, storage and renewables actually hits its own targets over the next five years. If it doesn’t, don’t be surprised if this argument gets a lot louder, and a lot less patient with technical detail.

– Marcus Wren, Editor

Photo by Grianghraf on Unsplash