In November and December 2024, Frontier Economics released two reports on the transition required in the NEM to achieve net zero emissions by 2050, drawing on AEMO’s 2024 Integrated System Plan (ISP).
This was followed by various summaries and takeaways of the Frontier modelling by, amongst others, Steven Hamilton, Matt Kean, Kane Thornton, and David Leitch.
In contrast to some of the public reaction to date, I think Frontier’s work adds to the debate about the potential for nuclear power generation in Australia. In particular, Frontier’s work raises a philosophical question about the pace of decarbonisation we want, and our willingness to pay for it.
We might all subscribe to “Net Zero by 2050”, but the pathway to get there, as AEMO’s and Frontier’s modelling show, can clearly differ. I will return to this issue, below.
This article differentiates from what’s been publicly said and written about Frontier’s work in two important ways:
1. It focuses on the assumptions in and implications of Frontier’s modelling, whereas the bulk of the existing public discussion mixes aspects of Frontier’s analysis with political parties’ and politicians’ statements on nuclear power and their selective use of Frontier’s analysis to support their statements.
2. It discusses both of the Frontier reports, whereas the bulk of existing discussion focuses on selected aspects of Frontier’s second report.
A quick summary of the two Frontier reports
To heavily summarise:
1. Report #1 shows how Frontier is, broadly speaking, able to recreate the generation outcomes modelled under ISP’s Step Change and Progressive Change scenarios. As part of this, Frontier estimates the generation and transmission costs required under Step Change and also argues emissions costs should be removed from the cost-benefit analysis as a CO2 price does not exist in the NEM. These modelling assumptions and justifications set the scene for the analysis in Frontier’s second report.
2. Report #2 shows the impact on generation and transmission costs (but, as just noted, doesn’t consider emissions costs impacts) from adding ‘baseload’ large scale nuclear plants into the NEM under each of Step Change and Progressive Change, starting with 1GW of nuclear during FYE2036 and up to 13.3GW by FYE2049. It shows cumulative emissions in the NEM are much higher over the modelled horizon (FYE2025 – FYE2051) and generation and transmission costs lower, when nuclear is included in the NEM vs. being excluded.
Both reports have common ‘narratives’ and associated challenges, as outlined below.
Five key narratives from the two reports – and the challenges of each
Narrative #1: there are no adverse cost or reliability implications from extending the life of ageing coal plants
Both Frontier reports note publicly-announced coal plant exit dates are up to a decade after dates required to meet the emissions budget under Step Change. As such, the starting point for Frontier’s nuclear cost-benefit analysis is a pushback to coal plant closures to their publicly-announced dates, after which they are replaced by nuclear.
In Frontier’s analysis, the cost associated with “sweating these coal plants” is the additional fuel and operations and maintenance costs from their extended operation. The key premise here seems to be that coal plants do not become less reliable as they age – a premise at odds with evidence showing increased outages.
One example of this was in NSW during late Nov 2024, when hot weather resulted in unplanned outages at Eraring just when it was most needed to supply the increased demand. Such risks would increasingly arise under Frontier’s modelling that delays closure of all existing coal plants by up to a decade without any investment in plant refurbishment.
However, lack of plant refurbishment is likely to result in extra costs for consumers in the following two ways:
1. Higher and more volatile wholesale prices are likely to result, as higher-priced gas generation needs to run and run more often to make up for increased coal plant outages, and;
2. The cost from the extra risk of load shedding (‘unserved energy’) as coal is unexpectedly and at short notice withdrawn from the market at times when it is most needed (as the recent NSW experience demonstrated).
Conversely, including coal plant refurbishment costs will reduce the costs and risks associated with 1. and 2. above, but then such refurbishment costs need to be included in the cost-benefit analysis.
Hence by omitting both plant refurbishment costs and the costs associated with no refurbishment, it is unsurprising Frontier’s cost-benefit analysis concludes a net consumer benefit from coal plant “sweating”.
Eagle-eyed readers will note that the above discussion does not consider the emissions costs associated with delayed coal plant closure. These costs are discussed below.
Challenge #1: Delaying coal closures will pose significant risks and drive up costs for consumers from increased plant unreliability
Narrative #2: renewables have to make way for nuclear: pause developments, increase renewables curtailment
Frontier’s modelling excludes behind-the-meter supply and storage options, namely rooftop PV, residential batteries, and electric vehicles. Frontier notes it assumes “that this is likely to be roughly constant across the scenarios” (p15, Report #2).
This assumption is inconsistent with the lived experience: households’ uptake of rooftop PV reflects their price responsiveness and so it is likely that households’ investment in rooftop PV will be adversely impacted if large-scale nuclear was to be installed.
Tesla was recently noted to have warned that Australian households face severe curtailment of their rooftop PV systems if nuclear enters the NEM. This was based on just 2GW of nuclear being added, not 13.3GW modelled by Frontier Economics. Others have also expressed concern that inflexible, “always on”, plant like nuclear and coal do not complement solar.
Given this lack of complementarity, it is striking that Frontier’s report does not contain the volume of renewables curtailed under its modelling. Some claim rooftop PV curtailment could be as high as 70% if large-scale nuclear were installed, which – if true – would significantly undermine the returns to households from investing in renewables. We would expect households to respond to this signal.
Given more than one in three Australian households currently have rooftop PV – a ratio that will rise between now and the mid-2030s if current rooftop PV uptake trends continue – introducing nuclear may not be well-received by these households.
Challenge #2: households are unlikely to accept losing money on their solar PV investments from nuclear
Narrative #3: not much transmission is required under nuclear
Frontier’s report #1 lists 28 transmission projects needed under AEMO’s Step Change that, Frontier argues, will not be needed if nuclear is built. However, as the CEC noted, some of these 28 projects are already in construction and can’t and/or shouldn’t be easily reversed – such as Project EnergyConnect and Waratah Super Battery.
Furthermore, even if nuclear is located where existing coal plants reside, the network in these areas will need replacement as they reach their end of technical life – which will likely be from the late-2030s onward as these network assets will be c.60 years old by then.
Given this, Frontier’s estimates of transmission costs under its two with-nuclear scenarios appear an underestimate. How significant an underestimate we don’t yet know. Hopefully this is addressed in a with-nuclear scenario in AEMO’s ISP 2026, to provide additional information on transmission with large-scale nuclear in place.
This creates Challenge #3: More transmission is likely needed to connect new large-scale nuclear plants than is estimated by Frontier Economics
Narrative #4: nuclear will cost less in the NEM than what it costs in other countries with experience with nuclear power generation
Frontier’s choice of nuclear technology is large-scale nuclear, with a capital cost for the first plant of $10,000/kW. While Frontier notes this cost is higher than CSIRO GenCost’s – CSIRO GenCost feeds into AEMO’s ISP – it fails to note these costs are well below actual costs of large-scale nuclear plants in countries that geopolitically are like Australia, but have significantly more experience with large-scale nuclear than Australia does. For example, UK’s Hinkley C nuclear plant is likely to cost at least $27,500/kW – almost three times more than Frontier’s estimate.
Therefore, it would be useful to assess how sensitive Frontier’s with-nuclear cost savings are to different large-scale nuclear cost assumptions; for example, large-scale nuclear costs of $25,000/kW. Frontier discusses at length the ‘blowout’ in observed transmission costs in Australia.
This has occurred due to the increases in steel costs and labour, as well as increased cost of biodiversity remediation actions and longer development and construction times. It is unclear why some of these cost drivers, especially from supply chain constraints, would not also impact large-scale nuclear plants. Moreover, the cost of complying with safety-related regulatory requirements will be much greater than for wind or solar PV.
All this said, in truth we don’t know how much large-scale nuclear will cost in Australia for the simple reason that we haven’t built them here (yet). Hence the need for any cost benefit analysis of nuclear to consider alternative scenarios for nuclear costs.
This leads to Challenge #4, aptly put by the Grattan Institute’s Tony Wood: “We know how difficult it’s been to build [nuclear] in the past 25 years, so how do we assume it’s going to be so easy in the next 25 years?”
Narrative #5: it is ok for the NEM to significantly slow its pace of decarbonisation
For readers that have made it this far, the most important implication (in my view) of Frontier’s analysis has been saved for last.
It is worth recalling AEMO’s ISP scenarios are surprisingly well crafted: the scenarios emanate from economywide, whole-of-system modelling of different Net Zero-by-2050 scenarios differentiated by the extent of global warming: commitments consistent with collective action to limit global warming to 1.5°C (Green Energy Exports), 1.8°C (Step Change), and 2.4°C (Progressive Change).*
(*These correspond to Representative Concentration Pathways (RCPs) of 1.9, 2.6, and 4.5, respectively. A fourth ISP scenario, Slow Change, a 4.5°C-aligned scenario with an RCP of 7.0 or higher, was discontinued from ISP 2024 onward due to the high unlikelihood of such weak global climate action.)
By being a signatory to the Paris Accord, Australia has effectively committed to a Step Change pace of decarbonisation for its economy and its power sector. In addition, the power sector is widely seen as providing the cheapest forms of emissions abatement, and so achieving economywide Net Zero by 2050 requires Net Zero in the power sector before 2050.
Moving from Step Change to Frontier’s Progressive Change incl. nuclear would see Australia backslide significantly on its Paris commitments; Progressive Change incl. nuclear’s cumulative NEM-wide CO2-e emissions are around 1 billion tonnes higher than Step Change. The economy-wide implications of this include:
– Can Australia still achieve economy-wide Net Zero by 2050 under Frontier’s with nuclear scenarios?
– If so, from which sector(s) will these emissions reductions occur and what is the cost to Australians from abating these emissions? And are these higher costs greater than any savings within the power sector?
– If not, then what are the impacts for Australia from geopolitical and trade perspectives from not fulfilling our commitments? Impacts include lower export revenues from exports made less competitive due to the impact of carbon border adjustment mechanisms.
Frontier’s report is silent on these implications, which leads to Challenge #5: delaying the NEM’s decarbonisation may significantly threaten Australia’s ability to achieve its economy-wide emissions reduction commitments, and may pose significant sovereign risks for Australia – potentially offsetting any savings by delaying the power sector’s transition
Of all the narratives, this one may be the most damaging geopolitically, given the leadership role Australia plays on global climate action, especially climate action in the Asia-Pacific.
In conclusion
To reiterate what I started with, I think Frontier’s work adds to the debate about the potential for nuclear power generation in Australia – their modelling poses some important philosophical questions for us.
However, the above-noted challenges means Frontier’s modelling falls short of a definitive answer to whether nuclear is appropriate, at any scale and at any future time horizon, for Australia’s power sector. As Frontier appropriately noted, their modelling is not “the last word on this matter.”
