Seven caveats behind CSIRO’s GenCost ‘NoProgressToNetZero’ finding

CSIRO recently published its GenCost 2025-26 report, estimating the cost of various electricity technologies. This year’s report echoed the findings of the last few years, with solar PV and onshore wind shown to be the lowest cost new build technologies.

This year, as a new, secondary step – distinct from GenCost’s long-running technology cost estimates – CSIRO modelled overall system costs across five scenarios, defined by 2050 electricity sector emissions intensity. Only one, NoProgressToNetZero, is anchored to the emissions intensity consistent with Australia’s 2030 82% renewables target; the other four are scaled to progressively greater consistency with net zero by 2050. NoProgressToNetZero does not represent “no action”: it still reaches 70% renewables, 19% gas and 11% coal by 2050, a substantial transition from today. Across all scenarios, deploying solar PV, wind, gas and storage for the majority of generation was found to be the least-cost 2050 option.

One result of the CSIRO modelling was that electricity system costs could be 5% lower in NoProgressToNetZero than in the next-least-ambitious scenario. Some stakeholders have cited this finding to argue net zero should be abandoned. However, this scenario comes with seven caveats – a number of them explicitly put forward by CSIRO – meaning it can’t be relied on as the lowest-cost pathway for Australia’s economy overall. These caveats relate to this new 2050 scenario modelling specifically, not to GenCost’s separately established technology cost findings.

The seven caveats are: 

1. NoProgressToNetZero is not lowest-cost overall – abatement costs across the wider economy would be higher 

CSIRO finds this scenario is not the cheapest option overall: choosing it means higher emissions reduction costs across the rest of the economy. CSIRO says: “Achieving weak or no progress in reducing electricity sector emissions is not efficient for achieving net zero because electricity sector emissions reduction is a third to a half the cost of emissions reduction elsewhere in the economy.” 

CSIRO found the efficient range for the electricity sector’s emissions intensity level sits between the ModerateNetZero and StrongNetZero scenarios, at 0.05–0.02 tonnes of carbon dioxide equivalent per megawatt-hour (tCO2e/MWh), where the electricity sector’s marginal cost of abatement remains below the cost of abatement elsewhere in the economy.

2. CSIRO’s model is a 2050 snapshot, not a full build pathway like the ISP

CSIRO’s modelling costs the system in 2050 only, generated almost from a clean slate starting point. The 2050 scenarios assume nearly all existing generation has retired, except for the long-lived assets of transmission, hydro and pumped hydro. CSIRO says: “By 2050, we assume all current generation is retired and only currently existing or committed hydro, pumped hydro and transmission remains.” 

By contrast, the Integrated System Plan (ISP) builds the system year by year from today’s fleet, accounting for the project pipeline. CSIRO describes this 2050 scenario tool as “a simplified electricity model” that is “substantially less sophisticated than the multi-model state-of-the-art framework deployed in the ISP”. It recommends: “Where results differ, GenCost advises that the ISP should be given greater weight.”

3. Fossil fuel price doesn’t vary with how much fuel each scenario burns; correcting this largely eliminates the electricity cost advantage

NoProgressToNetZero has 19% gas generation in 2050, burning roughly six times more gas than StrongNetZero, and is the only scenario to use meaningful quantities of coal — yet all five scenarios use the same gas and coal prices drawn from a blend of AEMO scenarios.

CSIRO identifies fossil fuel price assumptions as a live problem, stating: “There is an inconsistency in fossil assumptions for the NoProgressToNetZero emission intensity scenario. … The NoProgressToNetZero emission intensity scenario might be more consistent with increasing demand for fossil fuels which could mean increasing fossil fuel prices. With this context, the competitiveness of fossil fuels in this scenario should be interpreted with some caution. Projections of fossil fuel prices for a non-decarbonising world are not readily available.” 

We recalculated fossil fuel prices using the Australian Energy Market Operator (AEMO)’s Slower Growth scenario, the highest-priced of AEMO’s three scenarios, and thematically closest to NoProgressToNetZero’s sustained fossil-fuel reliance, weighting state-based prices by where CSIRO’s model actually builds gas and coal capacity. This gives a 2050 gas price of $16.2 per gigajoule (GJ) (vs. CSIRO’s original $13.4/GJ) and a coal price of $4.1/GJ (vs. $3.7/GJ). We re-ran CSIRO’s published model with these gas and coal prices, finding an electricity system cost of $128.5/MWh for NoProgressToNetZero – just $2.3/MWh below WeakNetZero ($130.8/MWh), down from a prior $6.6/MWh gap.

Using CSIRO’s published high-band prices instead of the average ($17.4/GJ for gas, $4.4/GJ for coal, both drawn from the highs in the Slower Growth scenario), the gap narrows further, to just $0.6/MWh – making the difference between NoProgressToNetZero and WeakNetZero negligible, using nothing beyond CSIRO’s own figures.

AEMO’s Slower Growth scenario appears as the closest match to CSIRO’s NoProgressToNetZero due to the sustained global fossil fuel reliance. However, coal and gas prices could plausibly be even higher still, though this is uncertain. No scenario published by AEMO or by advisory firm ACIL Allen actually models Australia’s electricity sector plateauing in terms of emissions reductions after 2030, or coal prices for new coal generators. Gas prices are also shaped by capacity, utilisation and broader market dynamics that could cut either way.

Either way, a cost advantage this small, and this sensitive to uncertain gas and coal price assumptions that do not vary in line with different scenario settings, should not be read as a settled result.

4. Coal’s financing cost is assumed equal to every other technology; testing a higher, more realistic financing cost removes it from the solution

GenCost applies a single 7% discount rate to every technology, with no technology-specific risk premium anywhere. In practice, financiers routinely apply a higher cost of capital to new fossil fuel assets (in particular higher-emissions ones like coal) given policy, stranded asset, and long-term demand risks. 

AEMO’s commissioned research puts coal’s weighted average cost of capital (WACC) at 12.0% (its central estimate, with a 10.0-15.0% range), the highest of any technology assessed. Oxford Economics cautioned that the estimate should be treated with low confidence, as “market participants do not see these assets as viable... many could not foresee a project that they would pursue regardless of the WACC”. At 7% WACC, black coal’s 2050 levelised cost of electricity (LCOE) is $107–$182/MWh; at 12%, $154–$268/MWh.

We re-ran CSIRO’s published model across the lower Oxford Economics WACC range (testing both 10% and 12%), while keeping fuel prices at CSIRO’s original figures, and coal was removed from NoProgressToNetZero’s solution at both WACC levels tested. This raised NoProgressToNetZero’s system cost slightly, to $125.5/MWh, still $5.3/MWh cheaper than WeakNetZero ($130.8/MWh). A higher coal WACC alone, without also correcting fuel prices, didn’t overturn NoProgressToNetZero’s overall electricity cost advantage, but it did show that the scenario’s specific 11% coal share is not itself a robust part of the NoProgressToNetZero solution once a realistic financing cost is applied.

5. Fossil fuel supply is assumed unlimited, while renewables are modelled using the worst weather year available

The CSIRO modelling assumes there is no limit to the physical quantity of gas or coal available to generators. It does not model constraints on gas pipeline infrastructure, fuel reserves or extraction rates. Griffith University researchers have found that Australian electricity market modellers “implicitly, and universally, assume the east Australian gas market is endlessly flexible and capable of meeting any level of demand from the GT [gas turbine] fleet”, and that heavy reliance on gas turbines “may routinely breach the outer operational boundaries of Australia’s existing gas network”. 

Further, major gas reserves in southern Australia are close to depletion, triggering warnings of gas supply shortfalls for some states in recent years. The CSIRO modelling does not explore this dynamic – it has no constraint linking the volume of gas it selects back to reserve or infrastructure limits. It also does not consider 2050 coal availability, which is uncertain given most thermal coal mines seeking extensions or expansions in Australia are export oriented rather than domestic focused. 

Renewables face the opposite treatment: CSIRO anchors its renewables profiles to a single historical weather year, 2011, “the single most costly weather year” among the 15 years of AEMO data available.

This embeds asymmetry into the model: wind and solar resources are assumed to be available as per the worst case weather year, while gas and coal resources are assumed to be available however the model requires. 

6. Fossil fuel price volatility is smoothed out entirely

CSIRO’s modelling did not consider disruptive events and price shocks, using smooth coal and gas price trajectories developed by ACIL Allen for AEMO. This means events like the 2022 Ukraine gas supply shock – cited elsewhere in GenCost as a driver of historical electricity price volatility – are not reflected as a risk in the 2050 costings. A fossil fuel-heavy scenario like NoProgressToNetZero carries far more exposure to this kind of volatility than the renewables heavy scenarios, a risk differential the modelling does not capture. CSIRO notes this protection renewables provides saying “renewable energy supported by storage is helping to protect Australia against global energy shocks and continues to provide the lowest cost pathway for Australia’s electricity system to achieve net zero emissions”.

7. Wholesale market dynamics are excluded 

CSIRO explored system cost, but not wholesale electricity market dynamics and prices. Gas typically sets the price at high levels when it is the marginal price-setter, so more gas in the system could mean higher realised wholesale prices than a simple cost comparison implies. 

As CSIRO notes: “GenCost calculates the breakeven cost of electricity needed for investors to recover their capital, fuel and operating costs, including a reasonable return on investment. This is an indicator of the electricity price needed to encourage new investment, but it does not control the electricity price. Electricity prices are controlled by the balance of supply and demand. If supply is tight relative to demand, then prices go up. If supply is significantly more than demand, then prices go down. Changes in fossil fuel prices are another source of volatility. Price increases in recent years are a combination of lack of supply and fuel price volatility.”

In conclusion, GenCost’s results cannot be read as showing a lower electricity decarbonisation target is the lower cost pathway for Australia’s economy overall. NoProgressToNetZero’s modelled 5% electricity cost advantage rests on a number of simplifications and assumptions that may not eventuate, and according to CSIRO’s analysis, following this pathway would lead to higher emission reduction costs in the broader economy.

Data sources and calculation notes

• Gas price $16.2/GJ: ACIL Allen’s latest regional gas powered generation price forecasts for AEMO, Slower Growth scenario, published November 2025. Regional prices are weighted by the share of NoProgressToNetZero’s modelled gas generation built in each state sourced from CSIRO’s GenCost 2025-26 Final Appendix Tables.
• Coal price $4.1: AEMO’s Draft 2026 ISP Input, Assumptions and Scenarios Workbook based on ACIL Allen November 2024 forecasts for AEMO – these are the only available forecast for coal prices based on existing generators, as AEMO does not model new-build coal separately. Regional prices from the Slower Growth scenario have been weighted by the share of NoProgressToNetZero’s modelled coal generation built in each state (sourced from CSIRO’s GenCost 2025-26 Final Appendix Tables).
• Upper-bound prices ($17.4/GJ gas, $4.4/GJ coal): CSIRO’s published high-band figures (sourced from Apx Table B.9 in Gencost 2025-26 Final Appendix Tables and Parameter Data 2050), used as inputs to CSIRO’s 2050 SLCOE modelling in place of the average band prices.

IEEFA thanks Dr Dylan McConnell from UNSW for rerunning CSIRO’s published model using updated assumptions drawing on IEEFA analysis. 

This article first appeared in RenewEconomy.