CSIRO plays GenCost critics off a break

Hey Reader, welcome to The Energy's weekly data newsletter. This week we explore how CSIRO's GenCost report has evolved.

All in

Like the Australian Energy Market Operator’s Integrated System Plan, the CSIRO’s annual GenCost report, which estimates the all-in cost of different generation technologies, has been the target of a sustained campaign by clean energy opponents.

And just like AEMO, CSIRO has listened to its critics, tweaked its models and come to the same broad conclusion as in past GenCost reports: The lowest cost path to net zero in the electricity system is based almost entirely on wind, solar and storage.

That won’t placate critics, but it’s worth examining how we got here and, to use a cricket analogy, how CSIRO has played its critics off a break. (It was helped by big increases in the capital costs of “baseload” technologies such as nuclear power).

GenCost has traditionally provided Levelised Cost of Electricity or LCOE estimates for the eastern states' National Electricity Market. LCOE is the all-in cost of generating electricity, taking into account capital costs and operating costs including any fuel costs.

GenCost has for some years now consistently found solar and wind generation to have the lowest LCOEs and nuclear (large or small modular reactor) to have the highest, challenged only by adding carbon capture and storage (CCS) to coal or gas which pushes their costs skywards too.

The estimates included some allowance for transmission and factored in renewables with storage, but critics said this was not enough to take into account the system costs of running an electricity grid with ever-increasing shares of variable wind and solar.

This critique was not without substance, but it ignored the costs of climate change. It also ignored that bare LCOEs would still help developers and large energy users — industrial plants, smelters, data centres etc — planning to co-locate these large loads with wind, solar and batteries "behind the meter", for whom system costs were hardly relevant.

Neatly delineates

LCOE (Levelised Cost of Electricity) range by technology and System LCOE for 2030
*Source: CSIRO's Draft GenCost 2025-26*

Still, CSIRO took it on board. It added a System Levelised Cost of Energy estimate to GenCost, which doesn’t help many 'system cost' critics who were also advocates of retaining coal and gas power, or adding new coal, gas or nuclear generation to the mix.

That’s because it more neatly delineates the generation technologies that fall below or in line with the SLCOE — onshore wind and solar — and those that fall above (black coal, gas, offshore wind) and far above (coal or gas with CCS, large or SMR nuclear, and solar thermal).

The first chart shows the GenCost estimate of LCOEs in 2030 compared to the SLCOE, which comes in at $91/MWh including transmission, or $81/MWh excluding transmission. This compares with a volume weighted average cost of $129/MWh in 2024-2025.

The SLCOE increases to between $135/MWh and $148/MWh including transmission costs in 2050, or between $115/MWh and $124/MWh not including transmission. This is because increasing shares of wind and solar require more transmission, storage and gas peaking, adding to costs.

In the 2050 estimates, black coal and gas without CCS straddle the SLCOEs, as does offshore wind. But black coal without CCS emits too much carbon to play a role in any serious net zero effort.

CSIRO also points out that to hit the bottom of its estimated LCOE ranges, coal, gas (without CCS) and nuclear plants would need to achieve 89% utilisation, which is virtually impossible in a renewables-rich grid.

Futility

The futility of wishing coal generation back into the grid mix is starkly illustrated by this chart, which shows the capital cost component of generation over the eight years since CSIRO began compiling GenCost.

Nuclear is not included in this chart because its capital costs are so high it would be off the chart (see below).

Black coal plant costs have increased at a steady clip to about $7000 per kilowatt capacity by 2025; add CCS and the capital cost goes nuclear ($13,000/KW). Gas (combined cycle or open cycle) are cheap to build (around $2000/KW) but costly to fuel, and very costly with CCS (around $7000/KW).

Utility scale solar costs have fallen steadily to be as cheap as open cycle gas to build (below $2000/KW) and of course there is no fuel cost. Onshore wind generation has increased to be almost twice the cost of solar to build, but eased this year. Solar thermal —as yet commercially unproven — is nearly $8000/KW, more costly than everything except black coal/CCS (and nuclear).

Eye-watering

Nuclear SMR includes a large “first of a kind premium” (92%) and starts at an eye watering $30,000/KW, falling to somewhere between $16,000/Kw and $18,000/KW if enough get built — which seems highly unlikely — under serious net zero policies.

Large-scale nuclear — FOAK of 120% — starts off above $10,000/KW, falls just below in the 2030s and rises above in the 2040s. In a market as small and fragmented as the NEM, too few nuclear plants are likely to get built for costs to fall significantly.

One reason costs for thermal technologies from coal to gas and nuclear are rising so sharply — open cycle gas plants have almost doubled in four years — is demand for gas plants from 'hyperscaler' data centre users. This is pushing up the cost of not just gas plant but anything that uses a thermally powered turbine.

At the same time battery and solar costs continue to fall. Battery costs are down 15-20% two years in a row. Large scale solar costs increased 9% — a timing issue — after dropping by similar amounts in each of the previous two years.

It should be noted that rooftop solar and home batteries — the target of big government subsidies — are following utility solar and batteries down China’s precipitous cost curve.

Large-scale solar is expected to drop below $1000/KW well before 2030, with rooftop solar costs following suit in the early 2040s (these projections are almost always beaten). Utility battery costs are expected to fall from around $500/KWh now — with small batteries before subsidies just over twice that — to $200/KWh by 2040.

Efficient

Not all emissions abatement is created equal. CSIRO compares economy-wide abatement costs to average and marginal abatement costs to find the most economically efficient target for electricity emissions.

It isn’t zero emissions, because the marginal cost of abatement to achieve this is nearly $800 per tonne of CO2, or more than 50% above the highest estimate for the economy wide cost of abatement ($500/TCO2).

CSIRO estimates a “strong net zero” target of between 20 kg/MWh and 50 kg/MWh — or between 4% and 10% of current emissions intensity — is optimal. Anything weaker leaves cheaper abatement than is available in other sectors going begging.

In that scenario, CCS and nuclear only get a look-in if mandated, gas (and hydrogen) play a very small role and the grid is dominated by “mature renewables”, or wind and solar, backed by storage.